Atherosclerosis of the ascending aorta is an independent predictor of long-term neurologic events and mortality

Patient population

The study population consisted of 1,969 consecutive patients ≥50 years old (mean 68 ± 9 years, range 50 to 91) undergoing cardiac surgery and intraoperative evaluation of the ascending aorta with epiaortic ultrasound at Barnes-Jewish Hospital during the period from January 1990 to May 1994. Long-term follow-up was complete in 1,957 patients. The cardiac surgical procedures performed in these patients included: coronary artery bypass graft (CABG) surgery in 1,472 (75.2%), valve repair or replacement in 224 (11.5%) and CABG and valve surgery in 261 (13.3%). Carotid endarterectomy or graft replacement of the ascending aorta was performed in combination with one of the above procedures in 65 (3.3%) and 47 (2.4%) patients, respectively. The study protocol was approved by the Human Studies Committee at Washington University Medical Center.

Variables analyzed

At the time of cardiac surgery, the following characteristics were prospectively recorded: 1) age; 2) gender; 3) body mass index (kg/m²); a history of any of the following (requiring treatment): 4) atrial fibrillation, 5) hypertension, and 6) diabetes mellitus; 7) history of smoking, 8) use of anticoagulants (warfarin and/or aspirin); 9) presence of mitral stenosis (mitral valve area <2 cm²); 10) left ventricular (LV) function; 11) extent of coronary artery disease (CAD); 12) extent of carotid artery disease; and 13) serum cholesterol level. The LV function was assessed by angiography or by echocardiography according to the Coronary Artery Surgery Study, based on left ventricular ejection fraction (LVEF), as follows: LVEF >55% (score of 5) = normal; 40% to 54% (score 6–10) = mild dysfunction; 25% to 39% (score of 11–15) = moderate dysfunction, and <24% (score >16) = severe dysfunction (37).

Ultrasonic methods

Ultrasonic epiaortic imaging of the ascending aorta was performed during cardiac surgery in each patient by use of a 7-MHz linear ultrasound transducer (Acuson, Mountain View, California) inserted in a sterile sheath and placed directly over this vessel, as previously described (21,24–26,38). In brief, transverse and longitudinal images were obtained from the level of the aortic root to the level of the proximal aortic arch, just distal to the innominate artery. The ultrasonic images were evaluated for the presence of atherosclerosis by two observers blinded to the clinical data, as follows: 1) normal aorta (no intimal thickening), 2) mild atherosclerosis (<3 mm intimal thickening without intimal irregularities), 3) moderate aortic atherosclerosis (intimal thickening ≥3 mm, with diffuse irregularities and/or calcification) and 4) severe atherosclerosis (>5 mm intimal thickening and one or more of the following: large protruding or mobile atheromatous debris, ulcerated plaques and/or thrombi).

Carotid artery Doppler and duplex ultrasound was performed in all patients ≥65 years old and in younger patients with carotid bruits and/or symptoms or a history of neurologic events, including transient ischemic attack (TIA) and stroke. Carotid artery stenosis was graded based on results of the carotid artery ultrasound performed in the longitudinal and cross-sectional axes, as previously described (39,40): 1) insignificant or no disease (luminal narrowing ≤50%), 2) moderate disease (luminal narrowing >50% but <80%), 3) severe disease (luminal narrowing ≥80% but ≤99%) and 4) complete occlusion. For the purpose of analysis, severe disease and complete occlusion were combined.

Long-term follow-up

Each patient completed a questionnaire to identify those who had suffered a neurologic event such as stroke or TIA during the follow-up period. A stroke was defined as any neurologic impairment of motor, sensory or cognitive function that persisted >24 h, or was associated with death within 24 h, and that could not be explained by other neurologic (ie, dementia, head trauma), and/or medical (ie, metabolic abnormalities, drugs) etiologies; and a TIA as neurologic impairment that resolved within 24 h, and was unexplained by other neurologic and/or medical etiologies. All neurologic events were confirmed by the attending neurologist who cared for the patient at the time of the event, or were documented in the medical records, including results of computed tomography or magnetic resonance imaging of the brain. Death was investigated through a relative, the primary care physician or the hospital records. In all instances, death was documented by the death certificate. Three mailings of the questionnaire were sent within a six-month period; 1,676 (85%) of the patients responded, and for those who did not, a search of the National Death Index was performed, which yielded information on 281 (14%) of the patients. Long-term follow-up was unavailable on 12 patients (0.6% of the population); data from these patients were excluded from the analysis. The duration of follow-up was recorded as that from the day of cardiac surgery to the date of patient contact, the date of the neurologic event or the date of death. The mean follow-up time was 3.0 ± 1.9 years (range 0.1 to 7.6 years).

Statistical analysis

Data were analyzed with SAS version 6.11 (SAS Institute, Carey, North Carolina) and expressed as the mean ± SD. The association between the four categorical measurements of severity of atherosclerosis of the ascending aorta and the continuous variables was analyzed by analysis of variance (ANOVA). Wilcoxon’s test was used to assess the relationship between ascending aortic atherosclerosis and all dichotomous variables. Kaplan-Meier survival analysis was used to describe the time to the first event (neurologic events and all-cause mortality). Comparisons of survival curves were performed by use of a log-rank test. In all event-specific survival analyses, data were censored at the time of death if the patient had not previously suffered a neurologic event. A Cox proportional-hazards multivariate regression analysis was used to assess the independent significance of six predefined potential predictors of the composite measure of neurologic events, all-cause mortality and all events. In these regression analyses, findings with respect to severity of atherosclerosis of the ascending aorta were grouped into three categories: normal-mild, moderate and severe. The normal and mild categories were grouped together for two reasons: it led to a better regression fit, and there were no differences between these two groups. A p value <0.05 was indicative of a statistically significant difference.

Importance of hypertension, diabetes mellitus and atherosclerosis of the ascending aorta as predictors of stroke

The role of hypertension as the most powerful risk factor for stroke, and the reduction of this risk with adequate therapy, has been documented in numerous studies (41–45). Similarly, but to a lesser extent, diabetes mellitus has been associated with increased risk of stroke and a poorer prognosis after stroke (46–48). Thus, it is not surprising that these two risk factors, in addition to atherosclerosis of the ascending aorta, were the most significant predictors of neurologic events in the present study. Numerous case reports have identified an atherosclerotic thoracic aorta as a likely source of arterial embolism to the brain and peripheral vessels (14–19,31,49–50), and have been documented to occur spontaneously (20), after initiation of anticoagulation (51,52), in association with cardiac surgery (21–25,53), and after invasive procedures (54). Atherosclerosis of the ascending aorta is a risk factor that has an independent association with previous cerebrovascular events (26). A recent autopsy study showed that the presence of ulcerated plaques in the ascending aorta and/or the aortic arch was associated with a significantly greater prevalence of cerebral embolic events than that in the absence of such plaques (28). In addition, three recent prospective studies have shown an association between an atherosclerotic thoracic aorta (detected by transesophageal echocardiography [TEE]), and cerebral and/or peripheral vascular ischemic events (33–35).

Our findings support the results of these three studies, but differ in several respects. First, our patient population was almost six times that in the French Aortic Plaques in Stroke (FAPS) study, and more than 10 times larger than those in the studies by Tunick and by Mitusch. Second, the mean follow-up period in the present study was 3.2 years, compared with 2.4, 1.4, and 1.2 years in the FAPS, the Mitusch, and in the Tunick studies, respectively. Third, the techniques used for detection of atherosclerotic disease and the locations at which disease was sought differed. The three previous studies evaluated, by use of TEE, atherosclerotic disease of the aortic arch only. In the present study, only the ascending aorta was evaluated, and we used epiaortic ultrasound. Although the utility of TEE for the detection of atherosclerosis of the aortic arch and the descending thoracic aorta has been demonstrated (29–36), our group and Konstadt et al. have shown that intraoperative visualization of the ascending aorta is suboptimal with TEE as compared with epiaortic ultrasound (38,55). Fourth, the patient populations differ: we followed 1,957 consecutive patients undergoing cardiac surgery, and of these, only 87 (4.4%) had a history of a previous neurologic event. In the FAPS study, all 331 patients were hospitalized for recent brain infarctions, in the study of Tunick et al, over half of the 42 patients had suffered a previous vascular embolic event, and in the study of Mitusch et al., over one third of the patients had a history of thromboembolic disease. Since a history of previous cerebrovascular event is known to be a predictor of future events, it could be argued that there was selection bias in these three studies. Fifth, the incidence of recurrent neurologic events according to severity of aortic arch atherosclerosis in the FAPS study was higher than in our study (Table 5). This is likely due to differences in study populations and in duration of follow-up. Thus, our study confirms the findings of the FAPS study, that the incidence of neurologic events increases as the severity of aortic atherosclerosis increases. However, our study is unique because it documents, for the first time, that increasing severity of atherosclerotic disease of the ascending aorta is a significant predictor of long-term mortality.

Atrial fibrillation and carotid artery disease

Atrial fibrillation and carotid artery disease have been associated with an increased risk of neurologic events (6,7,11–12). In the present study, 75% of patients underwent carotid artery ultrasound, only 88 (6%) had >80% occlusion of one or both carotid arteries, and most of these underwent carotid artery endarterectomy in association with their cardiac surgery. Even though severe carotid artery stenosis was a univariate predictor of neurologic events and of all-cause mortality, this variable was not a predictor in the multivariate analysis. The small number of patients with uncorrected severe carotid artery disease may explain why this was not a predictor of subsequent neurologic events in our study. Our findings do not contradict those of previous studies in which severe carotid artery disease was found to be an independent predictor of neurologic events.

Atrial fibrillation also has been shown to be an independent predictor of neurologic events. In the present study, a history of atrial fibrillation was a univariate predictor of long-term neurologic events, but it was not a predictor in the multivariate analysis. Since only 92 patients suffered neurologic events, it is likely that statistical power is lacking to definitively determine the independent predictive value of atrial fibrillation for the development of neurologic events. Pooled data from studies comparing treatment with warfarin and placebo in patients with atrial fibrillation have shown an event rate of approximately 4.5% per year in the placebo group versus 1.4% in the warfarin group (56). Since 246 of the patients in the present study had a history of atrial fibrillation on enrollment, approximately 30 neurologic events attributable to atrial fibrillation would be expected over a three-year follow-up, but only if atrial fibrillation were left untreated. In the present study, most patients with atrial fibrillation were treated with aspirin and/or warfarin, which would mitigate against an effect of atrial fibrillation on neurologic events.

Limitations of the study

Our study population was limited to cardiac patients who underwent cardiac surgery, but because of the known association between CAD, hypertension and neurologic events, the results of this study may be generalizable to a large number of patients with these three conditions, who are at high risk for suffering neurologic events. In addition, a large number of deaths in patients with neurologic events occur in this high-risk population (57,58).

The strength of the relationship found in this study between aortic atherosclerosis and neurologic events may have been limited by a number of factors. First, we collected data on atherosclerosis of the ascending aorta only, not on the aortic arch or the descending thoracic aorta. Studies have documented that atherosclerotic disease in the aortic arch or the descending aorta is associated with an increased risk of stroke and/or peripheral vascular ischemic events (33–36). If the association between cerebrovascular events and atherosclerosis of the aortic arch is at least as strong as the association observed between the former and ascending aortic atherosclerosis, additional neurologic events could be explained by atherosclerosis of the aortic arch in the absence of disease in the ascending aorta. Thus, it is possible that the true incidence of neurologic events associated with atherosclerosis of the aorta was underestimated. Second, even though we obtained confirmation of the occurrence of neurologic events (eg, we required neurologic confirmation and/or hospital records clearly documenting the occurrence of stroke or TIA) and mortality (eg, the death certificate) in all cases, we were unable to determine the exact anatomic location of some of the neurologic events. If we had been able to confine our analysis to neurologic events that could not have been explained by significant contralateral carotid artery disease (in those who underwent carotid artery ultrasound), it may have been possible to ascertain the relative contribution of aortic atherosclerosis more definitively. Third, some patients underwent alterations in the surgical technique as a result of the epiaortic ultrasound findings, which may have reduced the observed long-term risk related to atherosclerosis of this vessel (24,25).

The therapeutic implications of the findings in this study are unknown. Further investigation will be needed to define better diagnostic strategies and to determine whether treatment with anticoagulants and/or neuroprotective agents will be effective in the prevention of atheroembolic neurologic events in these high-risk patients (59,60).

Conclusions

In this prospective, long-term follow-up study of 1,957 patients, we have shown that atherosclerosis of the ascending aorta is an independent predictor of neurologic events and of all-cause mortality. In addition to being a direct cause of atheroembolism, the atherosclerotic aorta may be a marker of generalized atherosclerosis and thus of increased morbidity and mortality.