A predictive model for screening cerebrovascular disease in patient undergoing coronary artery bypass grafting
http://www.100md.com
《血管的通路杂志》
Department of Thoracic & Cardiovascular Surgery and Vascular Medicine, Dupuytren University Hospital, Limoges, France
Presented at the 53rd International Congress of the European Society for Cardiovascular Surgery, Ljubljana, Slovenia, June 2–5, 2004.
Abstract
Objective: The occurrence of stroke during coronary artery bypass grafting (CABG) is multifactorial but the coexistence of carotid disease is considered as one of the avoidable sources. Beyond the perioperative management, the detection of carotid disease in a coronary patient could be of prognostic significance. A systematic screening for all candidates for CABG could, however, be a non-efficient strategy. We aimed to optimize the Duplex screening of candidates for CABG by studying risk factors of significant concomitant carotid lesions. Methods: We prospectively studied 1043 consecutive candidates for CABG by Duplex scanning. A first subgroup of 825 patients permitted to establish the predictive model of >50% stenosis. A multivariate analysis provided independent predictive factors. The ability of the model to predict >50% and >70% stenosis of neck arteries has been prospectively assessed on the 218 consecutive patients. Results: In the first group, 108 (13.1%) and 58 (7%) had respectively at least a >50% and >70% stenosis. The independent risk factors were: past history of stroke or transient ischemic attack, neck bruit, clinically apparent peripheral arterial disease (PAD) or subclinical PAD (ABI <0.85 or >1.5), and age >70 years (P<0.05). Among the subsequent 218 patients, the presence of at least one of these factors was able to detect 24 out of 26 patients (92.3%) with a >50% stenosis, and 100% of those with >70% stenosis, and could rule out 41% from a systematic Duplex screening. Conclusions: The excellent sensitivity of this risk assessment approach, makes an efficient screening of cerebrovascular disease possible in CABG patients.
Key Words: Carotid disease; Coronary surgery; Screening; Ultrasound
1. Introduction
Stroke and other ischemic neurological events are one of the most important side effects of cardiac surgery. Besides substantial improvements in cardiac surgery and perioperative care, the benefit is unfortunately partially offset by an increasing rate of very old patients addressed to cardiac surgery, presenting a higher risk of perioperative stroke [1–3]. In a recent general review, the overall estimation of stroke in a pooled population of 106 211 patients undergoing CABG in 33 studies is estimated at 2% [3]. This rate increases dramatically with the age of the patients, up to 9% in octogenarians [3]. The mortality of stroke after CABG is very high, from 21% to 27% [4,5], and its cost is estimated between 2 to 4 billion dollars yearly worldwide.
Several mechanisms contribute to the occurrence of post-CABG stroke, among which embolic, hemodynamic and hemostatic etiologies can be cited. In many patients, especially in the older ones, coronary disease is included in a multifocal atherosclerotic disease, with a significant rate of concomitant atherosclerotic cerebrovascular disease [6], which contributes to the risk of post-operative stroke after CABG. Consequently, the screening of cerebrovascular disease in these patients could be of prognostic interest. Carotid Duplex is currently the most accurate non-invasive method for screening.
In a prospective study, we aimed to emphasize on the risk of concomitant cerebrovascular disease as one of the embolic sources of post-CABG stroke, and attempted to determine the predictive factors of coexistent cerebrovascular disease in patients undergoing CABG, in order to establish a predictive model of selection of those requiring carotid Duplex.
2. Materials and methods
2.1. Data collection
We prospectively studied 1043 consecutive candidates for CABG. All patients accepted their enrollment in the study, as the cerebrovascular Duplex screening is systematically performed in our unit prior to any coronary surgery. All patients benefited from a standardized protocol including data collection on cardiovascular risk factors, past history and those issued from cardiac catheterization. In all patients, the presence of neck bruits was systematically searched by auscultation prior to Duplex screening. Clinically apparent peripheral arterial disease (PAD) was considered present if there was a history of peripheral arterial surgery for the limbs and/or the presence of intermittent claudication. In order to detect subclinical PAD, the ankle-brachial index (ABI) has been measured by investigators prior to Duplex screening by dividing the mean of posterior tibial (PT) and dorsalis pedis (DP) arteries systolic pressures by the mean of systolic pressures of both humeral arteries. For each patient, the lowest ABI between both ankles was kept for further analysis. This mode of ABI determination is known to present the lowest inter-observer variability [7]. Patients with no history of clinically apparent PAD but with an abnormal ABI (<0.85 or >1.5) were considered as presenting subclinical PAD.
Duplex of neck arteries have been systematically performed in all patients by using a 10 MHz linear probe (ATL, HDI 3000, Philipps Ultrasound, Bothell, Wa., USA). The presence of >50% and >70% carotid stenosis was based on systolic and diastolic flow velocities, derived from Duplex criteria published by Fillis et al. [8]. Briefly, >50% stenosis was defined by the presence of following criteria: systolic peak velocity >1.5 m/s, end-diastolic velocity >0.5 m/s and peak systolic internal carotid/common carotid ratio >2.2. For the >70% stenosis, velocity criteria were as follows: systolic peak velocity >2.5 m/s, end-diastolic velocity >1 m/s and peak systolic internal carotid/common carotid ratio >3. According to that paper and compared to arteriography, the sensitivity of those criteria is around 80–90% with a specificity of 95–99% [8]. In an internal evaluation in our department of Duplex stenosis quantification on a series of 50 patients with carotid endarterectomy based on the presence of a stenosis >70%, none of the surgical pieces neither arteriograms invalidated the Duplex diagnosis of >70% stenosis.
2.2. Elaboration and validation of the predictive model
After the enrollment of all of the 1043 patients, analysis was performed in two parts. First, for the elaboration of the model, the data of the first 825 patients (elaboration group) were used. The model elaborated should highlight predictive factors of >50% carotid stenosis. The age factor was both studied as a continuous variable, as well as a discrete variable, using different thresholds, and selecting the one being the most discriminative at each analysis. After statistical analysis, the predictive model has been applied on the 218 following patients (validation group) in order to determine the ability of the model to predict which patients should benefit from carotid Duplex for cerebrovascular disease screening (both >50% and >70% carotid stenosis). Finally, the overall diagnostic performances of the model were calculated in the whole population studied.
2.3. Statistical analysis
Chi-square test was used to compare categorical variables and Student t-test was used to compare continuous variables. Logistic multiple regression analysis was used to identify the predictive value of various factors for the presence of >50% and >70% carotid arteries. For each test, a P-value <0.05 was considered as significant.
3. Results
Among the first 825 subjects, 108 (13.1%) presented at least a >50% stenosis lesion. Fifty-eight (7%) had >70% stenosis lesion. In 105 cases (97.2%) the lesions affected carotid (internal or common) arteries. In three cases, vertebral arteries were involved (unilateral vertebral artery occlusion in two cases and bilateral occlusion in one case). Table 1 presents general data of this first group as well as the comparison of those presenting significant and severe cerebrovascular disease compared to the remaining patients. Excepted the age as a continuous variable, the statistically significant factors differing between patients with >50% vs. <50% stenosis are the same other than those differing between patients with >70% vs. <70% stenosis. Logistic multiple regression analysis provided independent predictive variables for the presence of significant CVD in these patients (Table 2). The presence of an abnormal ABI was predictive of the presence of significant CVD, and that was independent of the presence of a clinical PAD. Fig. 1illustrates the relationship between ABI and the risk of presence of significant CVD in the overall study group.
As the presence of a past history of CVD or the presence of a neck bruit are evident clinical factors suspecting the presence of significant CVD in patients undergoing CABG, we aimed to determine which factors are predictive of significant CVD in asymptomatic patients without neck bruit. A second logistic multiple regression has then been performed in those without any past history of CVD and with normal neck auscultation. This second analysis concerned 603 patients (73.1% of the first group). Two independent variables predicted the presence of significant CVD in patients without any history of CVD and without neck bruit: an age >70 years and the presence of PAD (clinically defined by a past history of PAD surgery and/or presence of intermittent claudication or diagnosed by an abnormal ankle-arm index). These results are displayed in Table 3.
Finally, according to these analysis, 4 independent factors are predictive of significant CVD in patients undergoing CABG: an age >70 years, a past history of CVD, the presence of a neck bruit, and the presence of PAD, defined clinically or by the presence of an abnormal ABI. In order to have the best sensitivity for the identification of patients at risk of concomitant CVD, we considered that patients with at least one of these factors would be those with a high level of risk of >50% stenosis cerebrovascular disease, requiring carotid Duplex for additional investigation.
In the second part of the study, we prospectively compared the strategy of performing neck arteries Duplex only in patients with at least one of the four risk factors, to the one performing a systematic Duplex in all patients. For that purpose we compared both strategies in the following 218 patients in order to validate our predictive model. The characteristics in this second group are displayed in Table 4. These patients were comparable to the first ones excepted a significantly higher proportion of low ejection fraction (<40%) and presence of significant left main coronary artery stenosis (>50%).
In the test group, >50% carotid stenosis was found in 26 patients. Among them, 13 patients (6.0%) presented >70% carotid stenosis. Compared to the strategy of Duplex performed systematically, the strategy of requiring Duplex only in case of presence of at least one of the four risk factors presented the following results for the detection of carotid stenosis >50%: sensitivity 92.3%, specificity 42.2%, positive predictive value 17.8% and negative predictive value 97.6%. For the detection of >70% carotid stenosis the results of the model were as follows: sensitivity 100%, specificity 39.5%, positive predictive value 9.5% and negative predictive value 100%. In case of performing Duplex only when one of the predictive variables is at least present, 24 out of the 26 patients (92%) with a carotid stenosis >50% and all the 13 patients (100%) with stenosis >70% would have been detected. Age was the major cause of requiring Duplex, as 72.2% of patients directed to Duplex ultrasound were older than 70 years. In 40.7% of cases the age factor was the sole criterion for a Duplex scanning. The prevalence of the other predictive factors in patients oriented to Duplex ultrasound was: PAD (clinical and subclinical) 45.8%, past history of cerebrovascular disease 17.4% and neck bruit 16%.
Overall, the prevalence of significant carotid stenosis (>50%) in the 1043 patients undergoing CABG was 12.8%, and the pre-selection of the patients by the predictive factors for the requirement of Duplex presents the following performances, respectively for >50% and >70% stenosis: sensitivity 90.3% and 90.1%, specificity 37.6% and 35.8%, positive predictive value 17.6 and 9.3% and negative predictive value 96.3% and 98%. By the use of this strategy, Duplex could be avoided in 34% of cases (355 sonographies).
4. Discussion
Among post-CABG stroke risk factors, carotid lesions are of particular interest for several reasons. First, carotid stenosis is one of the major sources of peroperative cerebral emboli. In a general review compiling 59 studies [3], the odds ratio of occurrence of stroke after cardiac surgery is at 4.3 in case of >50% stenosis compared to <50% stenosis. Second, in patients with a high-grade coronary atherosclerosis, the coexistence of carotid disease is not rare. Table 5 presents the prevalence of >50% carotid stenosis in different studies on CABG candidates. Thirdly, this stroke risk factor can be easily screened by a reliable non-invasive method, namely carotid Duplex. Last, among all risk factors, carotid stenosis is one of the rare risk factors potentially correctable during the short lap between the indication and the realization of CABG. According to a study on a series of 19 224 subjects [15], the independent risk factors of post-CABG stroke were aortic calcifications, renal failure, history of stroke, smoking, carotid stenosis, age, clinical PAD, diabetes and on-pump cardiac surgery. Besides the choice of an off-pump cardiac surgery, the treatment of a carotid stenosis appears as a unique factor which can potentially be treated prior (during) surgery. The carotid stenosis management in CABG candidates is largely debated and overpasses the issue of our study. The aim of our study was to enhance our ability to screen patients. As we did not randomize patients for different types of management, we cannot recommend any specific management after the screening.
Beyond the perioperative risk, patients with a significant CVD are more prone to present further cerebrovascular events. The screening and knowledge of significant lesions in these patients is therefore of prognostic significance. That explains why we not only aimed to detect severe CVD (>70% stenosis) whose surgical correction is plausible, but also 50–70% stenosis, which will have to be subject to a long-term close surveillance. In the overall study group, in addition to the 71 patients (6.8%) with a >70% carotid stenosis, a group of 63 patients (6%) presented a 50–70% stenosis. Besides, the risk factors of presenting >50% stenosis and >70% stenosis are similar (Table 1), and the selection of the former provided a more powerful statistical analysis, with a larger number of patients with CVD. A majority of studies does present results of >50% stenosis (Table 5), and we can consequently compare our results to them. The overall prevalence of 12.8% of >50% stenosis is concordant with the literature. Differences can be explained by population's age, the exclusion of symptomatic patients, the mode of recruitment (consecutive or not) as well as the criteria for stenosis quantification.
We attempted to determine predictive factors issued from all available preoperative factors. The independent predictive factors after the first multivariate analysis were: age >70, smoking, presence of neck bruit, past history of CVD, clinical PAD, subclinical PAD. Two remarks should be made:
First, this study confirms the weakness of neck auscultation alone for the detection of patients with significant stenosis, as 78% of cases with stenosis were ‘silent’ and conversely 70% of those with a neck bruit did not have significant lesions. This is confirmed by prior studies reporting the presence of significant lesions only in 25 to 50% of cases with neck bruit [16–19].
The presence of PAD converts those with coronary artery disease in multifocal disease patients. In an epidemiological study in a population >62 years, two-third of those with PAD had a multifocal disease, and the coexistence of CAD and PAD increased dramatically the presence of CVD [20]. In a series of patients undergoing CABG [21], 63.9% of those with clinical PAD presented severe CVD (stenosis >80%) vs. only 16.9% in those without clinical PAD (P=0.00001). This high prevalence of coexistent carotid lesions in patients with PAD probably explains why PAD is an independent prognostic factor of occurrence of stroke after CABG [5,22,23].
One originality of our study is the detection of subclinical PAD. In the ARIC study in general population, the risk of carotid atherosclerosis was significantly higher in those with an ABI <0.9, with a multivariate-adjusted odds ratio around 2 [24]. Conversely to the neck arteries Duplex, the ABI measurement is easy. It does not require long-term training and requires only a low cost hand-held device. It is feasible in the patient's room or during the preoperative consultation of the surgeon or anesthesiologist, or prior to hospital discharge after coronary angiography. In our study, it permitted to detect two additional subclinical PAD patients for every clinical PAD patient. The multivariate analysis highlighted the additional and independent contribution of subclinical PAD detection in the prediction of CVD. The measurement of ABI increased of 12% the number of patients who should undergo to a Duplex assessment, which conducted to detect 1.3% additional patients with significant lesions, corresponding to a sensitivity increase of 8.2%.
We secondarily restricted our analysis in patients without neck bruit neither past history of CVD, as the presence of one of these two latter factors would accordingly indicate a Duplex investigation [25], even though the insufficiencies of neck auscultation have been underlined. That explains the second logistic regression analysis, in strictly asymptomatic and ‘silent’ patients, corresponding to 73% of all patients.
We finally determined 4 predictive factors of presence of concomitant CVD in CABG candidates: an age >70 years, a history of CVD, the presence of a neck bruit, and the presence of a clinically- or ABI-defined PAD. One might question the absence of determination of any risk score calculated from the odds-ratio, with the determination of a threshold by a ROC analysis. These approaches that have already been used for the assessment of the risk of stroke after cardiac surgery [23] favor the accuracy of the screening method by looking for the best balance between sensitivity and specificity. Here, we aimed to make the CVD screening more feasible in terms of time, human and material cost. As we consider carotid severe stenosis as a correctable risk factor, we attempted to focus on the sensitivity, in order to minimize the number of false negative patients. That explains that we considered Duplex mandatory in any case presenting one of the four predictive factors. Among those without any of these 4 risk factors, the risk of presence of a >50% carotid stenosis is at 3.7% compared to the 17.6% risk in the subgroup with at least one of these risk factors.
To our knowledge, only one prior study presented a probabilistic approach of carotid lesions as in our study, focusing on CABG candidates >65 years [10], with a prevalence of >50% carotid stenosis at 17%. Berens et al. [10] found the following predictive factors: a history of cerebrovascular events, a history of PAD, left main coronary stenosis, female gender, smoking. If Duplex was only proposed in case of at least one of these factors, this investigation would only be avoided in 20% of cases, with a 91% sensitivity and 20% specificity. Our predictive model is issued from an overall population without selection. Moreover, our study is the only one with a prospective validation of the risk model. Data issued from the univariate statistical analysis are very close to those issued from the Berens et al. study [10], but with a higher rate of investigations avoided (41% in the group 2, 34% in the overall study group) despite equivalent diagnostic performances. This could partly be due to the use of ABI measurement, as a multifocal atherosclerosis marker. Despite neck auscultation insufficiencies as a screening method alone, it seems, in the setting of clinical practice, regrettable that it was not studied in Berens et al. study.
The cost of Duplex of neck arteries is around 75 in our country, up to $500 in the United States. Taking into account the high number of CABG performed yearly worldwide, the reduction of the number of neck sonographies can be considerable. We did not perform an economic study of our probabilistic approach, as a comprehensive assessment should include the management of severe detected lesions which is not well standardized.
5. Conclusion
In this study, we prospectively assessed all the available preoperative data in patients undergoing CABG in order to determine those useful to select patients with a high probability of concomitant significant carotid disease. The satisfying sensitivity of the predictive model, enhanced by the detection of asymptomatic PAD by the measurement of ABI, allows us to propose this model for a probabilistic approach of carotid Duplex screening. To our knowledge this is the first study underlining the importance of asymptomatic PAD as a marker of concomitant carotid disease in CABG patients. The cheapness and easiness of ABI measurement makes the applicability of the model realistic. Beyond the detection of significant cerebrovascular disease, further analysis will provide data on the usefulness of this screening in this setting, not only for the occurrence of cerebral damage in the in-hospital post-operative period, but also during long-term follow-up.
Acknowledgements
The authors thank Mrs J. Deguillon, S. Penichon and J. Vigneron for their technical assistance during this study.
References
Jones EL, Weintraub WS, Craver JM, Guyton RA, Cohen CL. Coronary bypass surgery: is the operation different today J Thorac Cardiovasc Surg 1991;101:108–115.
Peterson ED, Cowper PA, Jollis JG, Bebchuk JD, DeLong ER, Muhlbaier LH. Outcomes of corocary artery bypass graft surgery in 24,461 patients aged 80 years or older. Circulation 1995;92:II-85–II-91.
Naylor AR, Mehta Z, Rothwell PM, Bell PRF. Carotid artery disease and stroke during coronary artery bypass: a critical review of the literature. Eur J Vasc Endovasc Surg 2002;23:283–294.
Bilfinger TV, Reda H, Giron F, Seifert FC, Ricotta J. Coronary and carotid operations under prospective standardized conditions: incidence and outcome. Ann Thorac Surg 2000;69:1792–1798.
Roach GW, Kanchuger M, Mangano CM, Newman M, Nussmeier N, Wolman R, Aggarwal A, Marschall K, Graham SH, Ley C. Adverse cerebral outcomes after coronary bypass surgery. N Engl J Med 1996;335:1857–1863.
Kuller LH, Shemanski L, Psaty BM, Borhani NO, Gardin J, Haan MN, O'Leary DH, Savage PJ, Tell GS, Tracy R. Subclinical disease as an independent risk factor for cardiovascular disease. Circulation 1995;92:720–726.
Aboyans V, Lacroix P, Lebourdon A, Preux PM, Ferrières J, Laskar M. The intra- and inter-observer variability of ankle-arm blood pressure index according to its mode of calculation. J Clin Epidemiol 2003;56:215–220.
Filis KA, Arko FR, Johnson BL, Pipinos II, Harris EJ, Olcott C IV, Zarins CK. Duplex ultrasound criteria for defining the severity of carotid stenosis. Ann Vasc Surg 2002;16:413–421.
Breslau PJ, Fell G, Ivey TD. Carotid arterial disease in patients undergoing coronary artery bypass operations. J Thorac Cardiovasc Surg 1999;17:765–767.
Berens ES, Kouchoukos NT, Murphy SF, Wareing TH. Preoperative carotid artery screening in elderly patients undergoing cardiac surgery. J Vasc Surg 1992;15:313–321.
Schwartz LB, Bridgman AH, Kieffer RW. Asymptomatic carotid artery stenosis and stroke in patients undergoing cardiopulmonary bypass. J Vasc Surg 1995;21:146–153.
Brener BJ, Brief DK, Alpert J, Goldenkrantz RJ, Parsonnet V. The risk of stroke in patients with asymptomatic carotid stenosis undergoing cardiac surgery: A follow-up study. J Vasc Surg 1987;5:269–279.
Riccotta JJ, Faggiolli GL, Castilone A, Hasset JM. Risk Factors for stroke after cardiac surgery: Buffalo Cardiac-Cerebral Study Group. J Vasc Surg 1995;21:359–364.
Fukuda I, Gomi S, Watanabe K, Seita J. Carotid and aortic screening for coronary artery bypass grafting. Ann Thorac Surg 2000;70:2034–2039.
John R, Choudri AF, Weinberg AD, Ting W, Rose EA, Smith CR, Oz MC. Multicenter review of preoperative risk factors for stroke after coronary artery bypass grafting. Ann Thorac Surg 2000;69:30–36.
Barnes RW, Rittgers SE, Putney WW. Real-time Doppler spectrum analysis. Arch Surg 1982;117:52–57.
Brewster DC, Schaen HH, Raines JK, Abbott WH, Darling RC. Rational management of asymptomatic carotid bruit. Arch Surg 1978;113:927–930.
Bihari-Varga M, Szekely J, Gruber E. Plasma high density lipoproteins in coronary, cerebral, and peripheral vascular diseases. The influence of various risk factors. Atherosclerosis 1981;40:337–344.
Keagy BA, Battaglini JW, Lucas CL, Thomas DD, Wilcox BR. Identification of internal carotid artery stenosis in coronary artery bypass candidates. South Med J 1983;76:996–999.
Aronow WS and Ahn C. Prevalence of coexistence of coronary artery disease, peripheral arterial disease, and atherothrombotic brain infarction in men and women > or = 62 years of age. Am J Cardiol 1994;74:64–65.
Salasidis GC, Latter DA, Steinmetz OK, Blair JF, Graham AM. Carotid artery Duplex scanning in preoperative assessment for coronary artery revascularisation: the association between peripheral vascular disease, carotid artery stenosis, and stroke. J Vasc Surg 1995;21:154–160.
Goto T, Baba T, Yoshitake A, Shibata Y, Ura M, Sakata R. Craniocervical and aortic atherosclerosis as neurological risk factors in coronary surgery. Ann Thorac Surg 2000;69:834–840.
Newman MF, Wolman R, Kanchuger M, Marschall K, Mora-Mangano C, Roach G, Smith R, Aggarwal A, Nussmeier N, Hrskowitz A, Mangano DT. participants in the Multicenter Study of Perioperative Ischemia Research Group. Multicenter preoperative stroke risk index for patients undergoing coronary artery bypass graft surgery. Circulation 1996;94:II-74–II-80.
Zheng ZJ, Sharrett AR, Chambless LE, Rosamond WD, Nieto FJ, Sheps DS, Dobs A, Evans GW, Heiss G. Associations of ankle-brachial index with clinical coronary heart disease, stroke and preclinical carotid and popliteal atherosclerosis: the Atherosclerosis Risk in Communities (ARIC) Study. Atherosclerosis 1997;131:115–125.
Walker WA, Harvey WR, Gaschen JR, Appling NA, Pate JW, Weiman DS. Is routine carotid screening for coronary surgery needed Am Surgeon 1996;62:308–310.(Victor Aboyans, Philippe )
Presented at the 53rd International Congress of the European Society for Cardiovascular Surgery, Ljubljana, Slovenia, June 2–5, 2004.
Abstract
Objective: The occurrence of stroke during coronary artery bypass grafting (CABG) is multifactorial but the coexistence of carotid disease is considered as one of the avoidable sources. Beyond the perioperative management, the detection of carotid disease in a coronary patient could be of prognostic significance. A systematic screening for all candidates for CABG could, however, be a non-efficient strategy. We aimed to optimize the Duplex screening of candidates for CABG by studying risk factors of significant concomitant carotid lesions. Methods: We prospectively studied 1043 consecutive candidates for CABG by Duplex scanning. A first subgroup of 825 patients permitted to establish the predictive model of >50% stenosis. A multivariate analysis provided independent predictive factors. The ability of the model to predict >50% and >70% stenosis of neck arteries has been prospectively assessed on the 218 consecutive patients. Results: In the first group, 108 (13.1%) and 58 (7%) had respectively at least a >50% and >70% stenosis. The independent risk factors were: past history of stroke or transient ischemic attack, neck bruit, clinically apparent peripheral arterial disease (PAD) or subclinical PAD (ABI <0.85 or >1.5), and age >70 years (P<0.05). Among the subsequent 218 patients, the presence of at least one of these factors was able to detect 24 out of 26 patients (92.3%) with a >50% stenosis, and 100% of those with >70% stenosis, and could rule out 41% from a systematic Duplex screening. Conclusions: The excellent sensitivity of this risk assessment approach, makes an efficient screening of cerebrovascular disease possible in CABG patients.
Key Words: Carotid disease; Coronary surgery; Screening; Ultrasound
1. Introduction
Stroke and other ischemic neurological events are one of the most important side effects of cardiac surgery. Besides substantial improvements in cardiac surgery and perioperative care, the benefit is unfortunately partially offset by an increasing rate of very old patients addressed to cardiac surgery, presenting a higher risk of perioperative stroke [1–3]. In a recent general review, the overall estimation of stroke in a pooled population of 106 211 patients undergoing CABG in 33 studies is estimated at 2% [3]. This rate increases dramatically with the age of the patients, up to 9% in octogenarians [3]. The mortality of stroke after CABG is very high, from 21% to 27% [4,5], and its cost is estimated between 2 to 4 billion dollars yearly worldwide.
Several mechanisms contribute to the occurrence of post-CABG stroke, among which embolic, hemodynamic and hemostatic etiologies can be cited. In many patients, especially in the older ones, coronary disease is included in a multifocal atherosclerotic disease, with a significant rate of concomitant atherosclerotic cerebrovascular disease [6], which contributes to the risk of post-operative stroke after CABG. Consequently, the screening of cerebrovascular disease in these patients could be of prognostic interest. Carotid Duplex is currently the most accurate non-invasive method for screening.
In a prospective study, we aimed to emphasize on the risk of concomitant cerebrovascular disease as one of the embolic sources of post-CABG stroke, and attempted to determine the predictive factors of coexistent cerebrovascular disease in patients undergoing CABG, in order to establish a predictive model of selection of those requiring carotid Duplex.
2. Materials and methods
2.1. Data collection
We prospectively studied 1043 consecutive candidates for CABG. All patients accepted their enrollment in the study, as the cerebrovascular Duplex screening is systematically performed in our unit prior to any coronary surgery. All patients benefited from a standardized protocol including data collection on cardiovascular risk factors, past history and those issued from cardiac catheterization. In all patients, the presence of neck bruits was systematically searched by auscultation prior to Duplex screening. Clinically apparent peripheral arterial disease (PAD) was considered present if there was a history of peripheral arterial surgery for the limbs and/or the presence of intermittent claudication. In order to detect subclinical PAD, the ankle-brachial index (ABI) has been measured by investigators prior to Duplex screening by dividing the mean of posterior tibial (PT) and dorsalis pedis (DP) arteries systolic pressures by the mean of systolic pressures of both humeral arteries. For each patient, the lowest ABI between both ankles was kept for further analysis. This mode of ABI determination is known to present the lowest inter-observer variability [7]. Patients with no history of clinically apparent PAD but with an abnormal ABI (<0.85 or >1.5) were considered as presenting subclinical PAD.
Duplex of neck arteries have been systematically performed in all patients by using a 10 MHz linear probe (ATL, HDI 3000, Philipps Ultrasound, Bothell, Wa., USA). The presence of >50% and >70% carotid stenosis was based on systolic and diastolic flow velocities, derived from Duplex criteria published by Fillis et al. [8]. Briefly, >50% stenosis was defined by the presence of following criteria: systolic peak velocity >1.5 m/s, end-diastolic velocity >0.5 m/s and peak systolic internal carotid/common carotid ratio >2.2. For the >70% stenosis, velocity criteria were as follows: systolic peak velocity >2.5 m/s, end-diastolic velocity >1 m/s and peak systolic internal carotid/common carotid ratio >3. According to that paper and compared to arteriography, the sensitivity of those criteria is around 80–90% with a specificity of 95–99% [8]. In an internal evaluation in our department of Duplex stenosis quantification on a series of 50 patients with carotid endarterectomy based on the presence of a stenosis >70%, none of the surgical pieces neither arteriograms invalidated the Duplex diagnosis of >70% stenosis.
2.2. Elaboration and validation of the predictive model
After the enrollment of all of the 1043 patients, analysis was performed in two parts. First, for the elaboration of the model, the data of the first 825 patients (elaboration group) were used. The model elaborated should highlight predictive factors of >50% carotid stenosis. The age factor was both studied as a continuous variable, as well as a discrete variable, using different thresholds, and selecting the one being the most discriminative at each analysis. After statistical analysis, the predictive model has been applied on the 218 following patients (validation group) in order to determine the ability of the model to predict which patients should benefit from carotid Duplex for cerebrovascular disease screening (both >50% and >70% carotid stenosis). Finally, the overall diagnostic performances of the model were calculated in the whole population studied.
2.3. Statistical analysis
Chi-square test was used to compare categorical variables and Student t-test was used to compare continuous variables. Logistic multiple regression analysis was used to identify the predictive value of various factors for the presence of >50% and >70% carotid arteries. For each test, a P-value <0.05 was considered as significant.
3. Results
Among the first 825 subjects, 108 (13.1%) presented at least a >50% stenosis lesion. Fifty-eight (7%) had >70% stenosis lesion. In 105 cases (97.2%) the lesions affected carotid (internal or common) arteries. In three cases, vertebral arteries were involved (unilateral vertebral artery occlusion in two cases and bilateral occlusion in one case). Table 1 presents general data of this first group as well as the comparison of those presenting significant and severe cerebrovascular disease compared to the remaining patients. Excepted the age as a continuous variable, the statistically significant factors differing between patients with >50% vs. <50% stenosis are the same other than those differing between patients with >70% vs. <70% stenosis. Logistic multiple regression analysis provided independent predictive variables for the presence of significant CVD in these patients (Table 2). The presence of an abnormal ABI was predictive of the presence of significant CVD, and that was independent of the presence of a clinical PAD. Fig. 1illustrates the relationship between ABI and the risk of presence of significant CVD in the overall study group.
As the presence of a past history of CVD or the presence of a neck bruit are evident clinical factors suspecting the presence of significant CVD in patients undergoing CABG, we aimed to determine which factors are predictive of significant CVD in asymptomatic patients without neck bruit. A second logistic multiple regression has then been performed in those without any past history of CVD and with normal neck auscultation. This second analysis concerned 603 patients (73.1% of the first group). Two independent variables predicted the presence of significant CVD in patients without any history of CVD and without neck bruit: an age >70 years and the presence of PAD (clinically defined by a past history of PAD surgery and/or presence of intermittent claudication or diagnosed by an abnormal ankle-arm index). These results are displayed in Table 3.
Finally, according to these analysis, 4 independent factors are predictive of significant CVD in patients undergoing CABG: an age >70 years, a past history of CVD, the presence of a neck bruit, and the presence of PAD, defined clinically or by the presence of an abnormal ABI. In order to have the best sensitivity for the identification of patients at risk of concomitant CVD, we considered that patients with at least one of these factors would be those with a high level of risk of >50% stenosis cerebrovascular disease, requiring carotid Duplex for additional investigation.
In the second part of the study, we prospectively compared the strategy of performing neck arteries Duplex only in patients with at least one of the four risk factors, to the one performing a systematic Duplex in all patients. For that purpose we compared both strategies in the following 218 patients in order to validate our predictive model. The characteristics in this second group are displayed in Table 4. These patients were comparable to the first ones excepted a significantly higher proportion of low ejection fraction (<40%) and presence of significant left main coronary artery stenosis (>50%).
In the test group, >50% carotid stenosis was found in 26 patients. Among them, 13 patients (6.0%) presented >70% carotid stenosis. Compared to the strategy of Duplex performed systematically, the strategy of requiring Duplex only in case of presence of at least one of the four risk factors presented the following results for the detection of carotid stenosis >50%: sensitivity 92.3%, specificity 42.2%, positive predictive value 17.8% and negative predictive value 97.6%. For the detection of >70% carotid stenosis the results of the model were as follows: sensitivity 100%, specificity 39.5%, positive predictive value 9.5% and negative predictive value 100%. In case of performing Duplex only when one of the predictive variables is at least present, 24 out of the 26 patients (92%) with a carotid stenosis >50% and all the 13 patients (100%) with stenosis >70% would have been detected. Age was the major cause of requiring Duplex, as 72.2% of patients directed to Duplex ultrasound were older than 70 years. In 40.7% of cases the age factor was the sole criterion for a Duplex scanning. The prevalence of the other predictive factors in patients oriented to Duplex ultrasound was: PAD (clinical and subclinical) 45.8%, past history of cerebrovascular disease 17.4% and neck bruit 16%.
Overall, the prevalence of significant carotid stenosis (>50%) in the 1043 patients undergoing CABG was 12.8%, and the pre-selection of the patients by the predictive factors for the requirement of Duplex presents the following performances, respectively for >50% and >70% stenosis: sensitivity 90.3% and 90.1%, specificity 37.6% and 35.8%, positive predictive value 17.6 and 9.3% and negative predictive value 96.3% and 98%. By the use of this strategy, Duplex could be avoided in 34% of cases (355 sonographies).
4. Discussion
Among post-CABG stroke risk factors, carotid lesions are of particular interest for several reasons. First, carotid stenosis is one of the major sources of peroperative cerebral emboli. In a general review compiling 59 studies [3], the odds ratio of occurrence of stroke after cardiac surgery is at 4.3 in case of >50% stenosis compared to <50% stenosis. Second, in patients with a high-grade coronary atherosclerosis, the coexistence of carotid disease is not rare. Table 5 presents the prevalence of >50% carotid stenosis in different studies on CABG candidates. Thirdly, this stroke risk factor can be easily screened by a reliable non-invasive method, namely carotid Duplex. Last, among all risk factors, carotid stenosis is one of the rare risk factors potentially correctable during the short lap between the indication and the realization of CABG. According to a study on a series of 19 224 subjects [15], the independent risk factors of post-CABG stroke were aortic calcifications, renal failure, history of stroke, smoking, carotid stenosis, age, clinical PAD, diabetes and on-pump cardiac surgery. Besides the choice of an off-pump cardiac surgery, the treatment of a carotid stenosis appears as a unique factor which can potentially be treated prior (during) surgery. The carotid stenosis management in CABG candidates is largely debated and overpasses the issue of our study. The aim of our study was to enhance our ability to screen patients. As we did not randomize patients for different types of management, we cannot recommend any specific management after the screening.
Beyond the perioperative risk, patients with a significant CVD are more prone to present further cerebrovascular events. The screening and knowledge of significant lesions in these patients is therefore of prognostic significance. That explains why we not only aimed to detect severe CVD (>70% stenosis) whose surgical correction is plausible, but also 50–70% stenosis, which will have to be subject to a long-term close surveillance. In the overall study group, in addition to the 71 patients (6.8%) with a >70% carotid stenosis, a group of 63 patients (6%) presented a 50–70% stenosis. Besides, the risk factors of presenting >50% stenosis and >70% stenosis are similar (Table 1), and the selection of the former provided a more powerful statistical analysis, with a larger number of patients with CVD. A majority of studies does present results of >50% stenosis (Table 5), and we can consequently compare our results to them. The overall prevalence of 12.8% of >50% stenosis is concordant with the literature. Differences can be explained by population's age, the exclusion of symptomatic patients, the mode of recruitment (consecutive or not) as well as the criteria for stenosis quantification.
We attempted to determine predictive factors issued from all available preoperative factors. The independent predictive factors after the first multivariate analysis were: age >70, smoking, presence of neck bruit, past history of CVD, clinical PAD, subclinical PAD. Two remarks should be made:
First, this study confirms the weakness of neck auscultation alone for the detection of patients with significant stenosis, as 78% of cases with stenosis were ‘silent’ and conversely 70% of those with a neck bruit did not have significant lesions. This is confirmed by prior studies reporting the presence of significant lesions only in 25 to 50% of cases with neck bruit [16–19].
The presence of PAD converts those with coronary artery disease in multifocal disease patients. In an epidemiological study in a population >62 years, two-third of those with PAD had a multifocal disease, and the coexistence of CAD and PAD increased dramatically the presence of CVD [20]. In a series of patients undergoing CABG [21], 63.9% of those with clinical PAD presented severe CVD (stenosis >80%) vs. only 16.9% in those without clinical PAD (P=0.00001). This high prevalence of coexistent carotid lesions in patients with PAD probably explains why PAD is an independent prognostic factor of occurrence of stroke after CABG [5,22,23].
One originality of our study is the detection of subclinical PAD. In the ARIC study in general population, the risk of carotid atherosclerosis was significantly higher in those with an ABI <0.9, with a multivariate-adjusted odds ratio around 2 [24]. Conversely to the neck arteries Duplex, the ABI measurement is easy. It does not require long-term training and requires only a low cost hand-held device. It is feasible in the patient's room or during the preoperative consultation of the surgeon or anesthesiologist, or prior to hospital discharge after coronary angiography. In our study, it permitted to detect two additional subclinical PAD patients for every clinical PAD patient. The multivariate analysis highlighted the additional and independent contribution of subclinical PAD detection in the prediction of CVD. The measurement of ABI increased of 12% the number of patients who should undergo to a Duplex assessment, which conducted to detect 1.3% additional patients with significant lesions, corresponding to a sensitivity increase of 8.2%.
We secondarily restricted our analysis in patients without neck bruit neither past history of CVD, as the presence of one of these two latter factors would accordingly indicate a Duplex investigation [25], even though the insufficiencies of neck auscultation have been underlined. That explains the second logistic regression analysis, in strictly asymptomatic and ‘silent’ patients, corresponding to 73% of all patients.
We finally determined 4 predictive factors of presence of concomitant CVD in CABG candidates: an age >70 years, a history of CVD, the presence of a neck bruit, and the presence of a clinically- or ABI-defined PAD. One might question the absence of determination of any risk score calculated from the odds-ratio, with the determination of a threshold by a ROC analysis. These approaches that have already been used for the assessment of the risk of stroke after cardiac surgery [23] favor the accuracy of the screening method by looking for the best balance between sensitivity and specificity. Here, we aimed to make the CVD screening more feasible in terms of time, human and material cost. As we consider carotid severe stenosis as a correctable risk factor, we attempted to focus on the sensitivity, in order to minimize the number of false negative patients. That explains that we considered Duplex mandatory in any case presenting one of the four predictive factors. Among those without any of these 4 risk factors, the risk of presence of a >50% carotid stenosis is at 3.7% compared to the 17.6% risk in the subgroup with at least one of these risk factors.
To our knowledge, only one prior study presented a probabilistic approach of carotid lesions as in our study, focusing on CABG candidates >65 years [10], with a prevalence of >50% carotid stenosis at 17%. Berens et al. [10] found the following predictive factors: a history of cerebrovascular events, a history of PAD, left main coronary stenosis, female gender, smoking. If Duplex was only proposed in case of at least one of these factors, this investigation would only be avoided in 20% of cases, with a 91% sensitivity and 20% specificity. Our predictive model is issued from an overall population without selection. Moreover, our study is the only one with a prospective validation of the risk model. Data issued from the univariate statistical analysis are very close to those issued from the Berens et al. study [10], but with a higher rate of investigations avoided (41% in the group 2, 34% in the overall study group) despite equivalent diagnostic performances. This could partly be due to the use of ABI measurement, as a multifocal atherosclerosis marker. Despite neck auscultation insufficiencies as a screening method alone, it seems, in the setting of clinical practice, regrettable that it was not studied in Berens et al. study.
The cost of Duplex of neck arteries is around 75 in our country, up to $500 in the United States. Taking into account the high number of CABG performed yearly worldwide, the reduction of the number of neck sonographies can be considerable. We did not perform an economic study of our probabilistic approach, as a comprehensive assessment should include the management of severe detected lesions which is not well standardized.
5. Conclusion
In this study, we prospectively assessed all the available preoperative data in patients undergoing CABG in order to determine those useful to select patients with a high probability of concomitant significant carotid disease. The satisfying sensitivity of the predictive model, enhanced by the detection of asymptomatic PAD by the measurement of ABI, allows us to propose this model for a probabilistic approach of carotid Duplex screening. To our knowledge this is the first study underlining the importance of asymptomatic PAD as a marker of concomitant carotid disease in CABG patients. The cheapness and easiness of ABI measurement makes the applicability of the model realistic. Beyond the detection of significant cerebrovascular disease, further analysis will provide data on the usefulness of this screening in this setting, not only for the occurrence of cerebral damage in the in-hospital post-operative period, but also during long-term follow-up.
Acknowledgements
The authors thank Mrs J. Deguillon, S. Penichon and J. Vigneron for their technical assistance during this study.
References
Jones EL, Weintraub WS, Craver JM, Guyton RA, Cohen CL. Coronary bypass surgery: is the operation different today J Thorac Cardiovasc Surg 1991;101:108–115.
Peterson ED, Cowper PA, Jollis JG, Bebchuk JD, DeLong ER, Muhlbaier LH. Outcomes of corocary artery bypass graft surgery in 24,461 patients aged 80 years or older. Circulation 1995;92:II-85–II-91.
Naylor AR, Mehta Z, Rothwell PM, Bell PRF. Carotid artery disease and stroke during coronary artery bypass: a critical review of the literature. Eur J Vasc Endovasc Surg 2002;23:283–294.
Bilfinger TV, Reda H, Giron F, Seifert FC, Ricotta J. Coronary and carotid operations under prospective standardized conditions: incidence and outcome. Ann Thorac Surg 2000;69:1792–1798.
Roach GW, Kanchuger M, Mangano CM, Newman M, Nussmeier N, Wolman R, Aggarwal A, Marschall K, Graham SH, Ley C. Adverse cerebral outcomes after coronary bypass surgery. N Engl J Med 1996;335:1857–1863.
Kuller LH, Shemanski L, Psaty BM, Borhani NO, Gardin J, Haan MN, O'Leary DH, Savage PJ, Tell GS, Tracy R. Subclinical disease as an independent risk factor for cardiovascular disease. Circulation 1995;92:720–726.
Aboyans V, Lacroix P, Lebourdon A, Preux PM, Ferrières J, Laskar M. The intra- and inter-observer variability of ankle-arm blood pressure index according to its mode of calculation. J Clin Epidemiol 2003;56:215–220.
Filis KA, Arko FR, Johnson BL, Pipinos II, Harris EJ, Olcott C IV, Zarins CK. Duplex ultrasound criteria for defining the severity of carotid stenosis. Ann Vasc Surg 2002;16:413–421.
Breslau PJ, Fell G, Ivey TD. Carotid arterial disease in patients undergoing coronary artery bypass operations. J Thorac Cardiovasc Surg 1999;17:765–767.
Berens ES, Kouchoukos NT, Murphy SF, Wareing TH. Preoperative carotid artery screening in elderly patients undergoing cardiac surgery. J Vasc Surg 1992;15:313–321.
Schwartz LB, Bridgman AH, Kieffer RW. Asymptomatic carotid artery stenosis and stroke in patients undergoing cardiopulmonary bypass. J Vasc Surg 1995;21:146–153.
Brener BJ, Brief DK, Alpert J, Goldenkrantz RJ, Parsonnet V. The risk of stroke in patients with asymptomatic carotid stenosis undergoing cardiac surgery: A follow-up study. J Vasc Surg 1987;5:269–279.
Riccotta JJ, Faggiolli GL, Castilone A, Hasset JM. Risk Factors for stroke after cardiac surgery: Buffalo Cardiac-Cerebral Study Group. J Vasc Surg 1995;21:359–364.
Fukuda I, Gomi S, Watanabe K, Seita J. Carotid and aortic screening for coronary artery bypass grafting. Ann Thorac Surg 2000;70:2034–2039.
John R, Choudri AF, Weinberg AD, Ting W, Rose EA, Smith CR, Oz MC. Multicenter review of preoperative risk factors for stroke after coronary artery bypass grafting. Ann Thorac Surg 2000;69:30–36.
Barnes RW, Rittgers SE, Putney WW. Real-time Doppler spectrum analysis. Arch Surg 1982;117:52–57.
Brewster DC, Schaen HH, Raines JK, Abbott WH, Darling RC. Rational management of asymptomatic carotid bruit. Arch Surg 1978;113:927–930.
Bihari-Varga M, Szekely J, Gruber E. Plasma high density lipoproteins in coronary, cerebral, and peripheral vascular diseases. The influence of various risk factors. Atherosclerosis 1981;40:337–344.
Keagy BA, Battaglini JW, Lucas CL, Thomas DD, Wilcox BR. Identification of internal carotid artery stenosis in coronary artery bypass candidates. South Med J 1983;76:996–999.
Aronow WS and Ahn C. Prevalence of coexistence of coronary artery disease, peripheral arterial disease, and atherothrombotic brain infarction in men and women > or = 62 years of age. Am J Cardiol 1994;74:64–65.
Salasidis GC, Latter DA, Steinmetz OK, Blair JF, Graham AM. Carotid artery Duplex scanning in preoperative assessment for coronary artery revascularisation: the association between peripheral vascular disease, carotid artery stenosis, and stroke. J Vasc Surg 1995;21:154–160.
Goto T, Baba T, Yoshitake A, Shibata Y, Ura M, Sakata R. Craniocervical and aortic atherosclerosis as neurological risk factors in coronary surgery. Ann Thorac Surg 2000;69:834–840.
Newman MF, Wolman R, Kanchuger M, Marschall K, Mora-Mangano C, Roach G, Smith R, Aggarwal A, Nussmeier N, Hrskowitz A, Mangano DT. participants in the Multicenter Study of Perioperative Ischemia Research Group. Multicenter preoperative stroke risk index for patients undergoing coronary artery bypass graft surgery. Circulation 1996;94:II-74–II-80.
Zheng ZJ, Sharrett AR, Chambless LE, Rosamond WD, Nieto FJ, Sheps DS, Dobs A, Evans GW, Heiss G. Associations of ankle-brachial index with clinical coronary heart disease, stroke and preclinical carotid and popliteal atherosclerosis: the Atherosclerosis Risk in Communities (ARIC) Study. Atherosclerosis 1997;131:115–125.
Walker WA, Harvey WR, Gaschen JR, Appling NA, Pate JW, Weiman DS. Is routine carotid screening for coronary surgery needed Am Surgeon 1996;62:308–310.(Victor Aboyans, Philippe )