INTRODUCTION
The technology that made coronary artery bypass grafting (CABG) surgery with cardiopulmonary bypass (CPB) possible was developed in the 1950s, and by the early 1960s, CABG was introduced as a clinical procedure. Since that time, advances in surgical and perfusion techniques have led to significant reductions in both the mortality and morbidity associated with this complex surgical procedure. Nonetheless, adverse postoperative cognitive and neurological outcomes continue to be a major concern. In a study from 1987, the 6-month incidence of postoperative cognitive decline in patients undergoing CABG surgery was found to be nearly 60% (Shaw et al., Reference Shaw, Bates, Cartlidge, French, Heaviside and Julian1987). Although the degree of this decline in cognitive performance was described as mild, the findings from this study nonetheless motivated several investigators to examine the incidence, nature, and etiology of cognitive decline after CABG. While there is little disagreement that even in the context of contemporary surgical practices, some patients may suffer cognitive decline after CABG, the longitudinal course of the recovery of these changes and the etiologic role of cardiopulmonary bypass remain controversial.
METHODOLOGICAL ISSUES
A large number of studies have examined the incidence of postoperative cognitive decline in patients undergoing CABG surgery (van Dijk et al., Reference van Dijk, Keizer, Diephuis, Durand, Vos and Hijman2000), but methodological limitations have resulted in widely discrepant findings (Symes, Maruff, Ajani, & Currie, Reference Symes, Maruff, Ajani and Currie2000). First, there is no agreement regarding the optimal time point after surgery at which follow-up cognitive testing should be performed. Neuropsychological testing at or around the time of hospital discharge after CABG may be subject to confounding effects of pain, fatigue, medication side effects, or other clinical issues. Therefore, many investigators have opted to defer follow-up testing until weeks or months after surgery. Although this may yield follow-up data that are less contaminated by nonspecific surgical factors, extending the follow-up time runs the risk of missing short-term postoperative cognitive changes. Second, most studies have not included appropriate comparison groups, such as nonsurgical patients with coronary artery disease who are managed medically, or patients undergoing major noncardiac surgery with general anesthesia. This is of particular relevance for studies that have evaluated patients within a week or two after surgery, since it is now increasingly clear that noncardiac surgery with general anesthesia can also be associated with some short-term cognitive changes (Newman, Stygall, Hirani, Shaefi, & Maze, Reference Newman, Stygall, Hirani, Shaefi and Maze2007). Third, in the absence of control subjects, most studies have been forced to rely on the use of largely arbitrary statistical criteria, such as 20% decline on 20% of tests, for estimating the incidence of postoperative decline. In a study that illustrates the importance of the choice of such a criterion, the incidence of cognitive decline at the time of discharge was found to range from 15% to 66% in the same study population (Mahanna et al., Reference Mahanna, Blumenthal, White, Croughwell, Clancy and Smith1996). Fourth, study inclusion criteria have varied widely, with some studies excluding patients with comorbid illnesses, such as peripheral vascular disease and diabetes.
The growing interest in evaluation of neuroprotective procedures with cardiac surgery and other medical interventions will require careful pre- and postoperative cognitive assessments. The most critical, as well as most challenging, part of the design of future studies may be the choice of an appropriate control group. In addition, if nonsurgical controls are used, the interval before the postoperative follow-up cognitive testing must be long enough to avoid nonspecific confounding effects of major surgery.
For investigators who prefer to go beyond traditional null hypothesis significance testing, recent methods to define incidence of cognitive decline based on reliable change indices have proven more valid than the previously used arbitrary “standard deviation methods” mentioned earlier. The reliable change methods take into account the variability of within-subject change scores in control subjects, and can thus adjust for sources of errors, such as random test-retest variability, regression to the mean, and practice effects when estimating the incidence of true change in postoperative test performance (Kneebone, Andrew, Baker, & Knight, Reference Kneebone, Andrew, Baker and Knight1998).
Preoperative Baseline Cognitive Performance
By definition, patients who are candidates for CABG have coronary artery disease. In addition, these patients frequently have associated vascular disease of the brain. MRI or other imaging studies are only rarely obtained before surgery in candidates for CABG, but in a study of 421 patients from Japan, 30% of patients were found to have small infarctions, and 20% multiple or large infarctions before surgery. The presence of one or more lacunar infarctions was associated with worse neurologic and cognitive outcomes (Goto et al., Reference Goto, Baba, Honma, Shibata, Arai and Uozumi2001). It is not surprising, therefore, that some candidates for cardiac surgery have been found to have abnormal cognitive performance even before surgery. The number of patients who might meet criteria for Mild Cognitive Impairment (MCI) preoperatively has not been reported, but one study found that 45% of patients had preoperative cognitive test performance significantly below that of control subjects without coronary artery disease (Hogue, Jr. et al., Reference Hogue, Hershey, Dixon, Fucetola, Nassief and Freedland2006). Other studies have reported that a surprising 20% of surgery candidates had Mini-Mental test scores below 24 when evaluated before surgery (Jensen, Rasmussen, & Steinbruchel, Reference Jensen, Rasmussen and Steinbruchel2008). The cognitive domains most commonly reported to be impaired preoperatively include new learning, executive functions, and motor speed, thus resembling the pattern of cognitive findings observed in patients with vascular cognitive impairment. These observations suggest that patients with coronary artery disease have cognitive deficits even before surgery, and emphasize the need for careful preoperative baseline cognitive assessments in order to interpret specific effects of the cardiac surgery itself on cognitive outcomes (Selnes & Zeger, Reference Selnes and Zeger2007). Additionally, these findings also draw attention to the importance of preoperative patient-related variables, such as age and degree of pre-existing cerebrovascular disease, as predictors of cognitive outcomes after CABG (Ho et al., Reference Ho, Arciniegas, Grigsby, McCarthy, McDonald and Moritz2004).
Short-term Cognitive Changes
A large number of studies have examined the incidence of cognitive decline at various intervals after bypass surgery. More recent studies that have recognized the importance of including control patients have generally reported a lower incidence of postoperative cognitive decline than older, uncontrolled studies. Depending on the specific criteria used, at 7 days after surgery, one study found that 4% to 33% of patients could be classified as having cognitive decline (Raymond, Hinton-Bayre, Radel, Ray, & Marsh, Reference Raymond, Hinton-Bayre, Radel, Ray and Marsh2006). Other studies have reported similar outcomes (Andrew, Baker, Bennetts, Kneebone, & Knight, Reference Andrew, Baker, Bennetts, Kneebone and Knight2001; Fearn et al., Reference Fearn, Pole, Wesnes, Faragher, Hooper and McCollum2001). Although it has been assumed that the early cognitive decline is directly related to cerebral complications associated with the use of cardiopulmonary bypass, nonspecific effects of major surgery undoubtedly also play a role (Newman et al., Reference Newman, Stygall, Hirani, Shaefi and Maze2007). One study comparing outcomes after major noncardiac surgery with that of cardiac surgery patients found no differences in the incidence of short-term cognitive decline, suggesting that factors like postoperative fatigue, pain, medication side effects, and sleep deprivation, may still be present at 7 days in some patients (Sugiyama et al., Reference Sugiyama, Kawaguchi, Yoshitani, Sugiyama, Inoue and Sakamoto2002). The incidence of postoperative cognitive decline is time-dependent, with evidence of gradual improvement in cognitive performance over time. In a recent small study of acute postoperative cognitive outcomes, a group of 67 CABG patients underwent cognitive testing before surgery and at days 3, 6, and 9 after surgery (Mullges, Berg, Schmidtke, Weinacker, & Toyka, Reference Mullges, Berg, Schmidtke, Weinacker and Toyka2000). A group of inpatients with peripheral neuropathy served as controls. Although there was a significant decline in neuropsychological test performance in the CABG group at days 3 and 6, return to baseline levels or above was observed as early as postoperative day 9. Another prospective study found that, by about 3 months after surgery, there was no longer any difference in the cognitive performance of patients who had had surgery compared with medically and demographically similar nonsurgical patients (Selnes et al., Reference Selnes, Grega, Borowicz, Royall, McKhann and Baumgartner2003). These findings suggest that for those patients who do experience postoperative cognitive decline, these cognitive changes are relatively mild and generally improve to baseline levels by 3 months after surgery.
Nature of Early Postoperative Cognitive Changes
Because the majority of studies in the literature have focused on defining the postoperative incidence of cognitive decline, less information is available regarding the nature of the cognitive decline. Although self-reported cognitive symptoms from both patients and spouses after cardiac surgery procedures typically have focused on memory changes (Bergh, Backstrom, Jonsson, Havinder, & Johnsson, Reference Bergh, Backstrom, Jonsson, Havinder and Johnsson2002), the objective neuropsychological test findings have not necessarily found memory to be the most commonly impaired domain. For example, Raymond and colleagues found that the highest incidence of decline was in the domains of attention and speed of information processing, with relatively few patients showing decline in memory (Raymond et al., Reference Raymond, Hinton-Bayre, Radel, Ray and Marsh2006). Other studies of short-term (7 days) cognitive outcomes have reported a high incidence of decline in speed of processing (Trail Making Test B; Digit Symbol), but relatively low incidence of decline on measures of new verbal learning and delayed recall (Andrew et al., Reference Andrew, Baker, Bennetts, Kneebone and Knight2001). The only study that has examined specific patterns of postoperative memory deficits after CABG also found that nearly one half of the patients had normal memory performance at 6 months after surgery. Of those with impaired memory, 35% showed a predominantly subcortical pattern of impaired delayed recall, but preserved recognition memory, and 17% had a cortical pattern of impairments in both delayed recall and recognition memory (Kneebone, Luszcz, Baker, & Knight, Reference Kneebone, Luszcz, Baker and Knight2005). Contrary to their expectations, however, the two memory subgroups did not differ on other measures typically associated with either cortical or subcortical cognitive impairment. Data currently available regarding the cognitive domains most likely to be affected by CABG have not shed much light on the possible etiology of these changes. For example, the findings of early deficits in attention and speed of processing could be consistent with both effects of factors specific to cardiac surgery but also with possible early effects of anesthesia and other nonspecific effects of major surgery.
Etiology of Postoperative Cognitive Decline
The neurological and cognitive adverse events that occur in some patients after cardiac surgery have been hypothesized to be secondary to three main procedure-related causes: atherosclerotic micro- and macroemboli released during the surgery, hypoperfusion, and the systemic inflammatory response. The vast majority of studies have focused on the role of microemboli, but surprisingly, very few studies have found a robust correlation between the number of emboli released during surgery and degree of decline in postoperative cognitive test performance. There are several possible reasons why a direct relationship between absolute number of emboli and neurocognitive outcomes has not been observed. First, it is possible that the makeup and size of the emboli are more predictive of neurological injury than the absolute number of emboli. Second, the functional consequences of showers of emboli released into the cerebral circulation might depend on the degree of pre-existing vascular disease of the brain. Atherosclerosis of the heart tends to be closely correlated with microvascular disease of the brain. Thus, older patients with CAD of sufficient severity to require surgery are likely to have vascular disease of the cerebral circulation, with accompanying changes in perfusion pressure and other hemodynamic changes that may affect their ability to clear or wash out any embolic particles released during the surgery (Caplan & Hennerici, Reference Caplan and Hennerici1998). These patients may thus be more susceptible to injury from microemboli than younger patients with little or no cerebrovascular disease.
There are no evidence-based guidelines for the optimal level of maintaining mean arterial pressure during coronary artery bypass surgery. Because many of the elderly patients who undergo CABG have a longstanding history of hypertension before surgery, the customary levels of mean arterial pressure (50–60 mm HG) maintained during the surgery may result in certain areas of the brain, most notable watershed regions, being hypoperfused. While severe intraoperative hypotension is known to be associated with watershed strokes, the neurocognitive effects of more subtle blood pressure changes have not been well established (Gottesman et al., Reference Gottesman, Sherman, Grega, Yousem, Borowicz and Selnes2006).
A systemic inflammatory response is known to occur during CABG, due to the fact that circulating blood comes into contact with the nonendothelial surface of the plastic tubing of the bypass circuitry. It has been associated with transient cerebral edema, but the cognitive effects are less well understood. In one of the few studies examining associations between markers of inflammation and postoperative cognitive performance, the authors concluded that there was no relationship between levels of inflammatory markers and cognitive outcomes at 5 days or 3 months (Westaby et al., Reference Westaby, Saatvedt, White, Katsumata, van Oeveren and Halligan2001).
Long-term Cognitive Changes
Several observational studies have reported a late cognitive decline, several years after CABG surgery. Newman and colleagues found that a surprising 42% of their patients had lower cognitive performance at 5 years follow-up than before surgery (Newman et al., Reference Newman, Kirchner, Phillips-Bute, Gaver, Grocott and Jones2001). One of the predictors of late decline was having had cognitive decline during the immediate postoperative period, leading the authors to hypothesize that there might be a causal link between the use of CPB 5 years earlier and late cognitive decline. Even though this was an uncontrolled study, the findings generated new interest in developing strategies to protect the brain during cardiac surgery. On the largely unproven assumption that adverse cognitive and neurologic outcomes were directly related to the use of extracorporeal circulation, this inspired the development of so-called off-pump surgery. In this procedure, the patient undergoes anesthesia and median sternotomy, but no cardiopulmonary bypass is used. Instead, the surgery is performed on the beating heart. Although early studies reported a reduction in adverse neurologic outcomes with off-pump surgery, this was most likely related to biases in patient selection. Subsequent randomized trials that compared patients with conventional on-pump surgery with those having off-pump surgery, were unable to demonstrate lower rates of short- or long-term cognitive decline in the patients having off-pump surgery (van Dijk et al., Reference van Dijk, Spoor, Hijman, Nathoe, Borst and Jansen2007; Takagi, Tanabashi, Kawai, & Umemoto, Reference Takagi, Tanabashi, Kawai and Umemoto2007).
These findings are also consistent with results from recent studies comparing long-term cognitive outcomes in patients after CABG with those in demographically similar patients with coronary artery disease but no surgery. One such study confirmed that while some “late” cognitive decline does occur after CABG surgery, there was no significant difference in the degree of decline between CABG patients and the nonsurgical controls with CAD (Selnes et al., Reference Selnes, Grega, Borowicz, Barry, Zeger and Baumgartner2005). These findings emphasize that patient-related factors, in particular degree of preexisting vascular disease of the brain, may play a more important role for cognitive outcomes after CABG than was previously thought.
Predictors of Postoperative Cognitive Decline
There are several studies that have developed predictive models, based on information available before surgery, for the risk of postoperative stroke in patients who undergo CABG (McKhann, Grega, Borowicz, Jr., Baumgartner, & Selnes, Reference McKhann, Grega, Borowicz, Baumgartner and Selnes2006; Anyanwu, Filsoufi, Salzberg, Bronster, & Adams, Reference Anyanwu, Filsoufi, Salzberg, Bronster and Adams2007). Similar models have not yet been developed for the risk of postoperative cognitive decline. In part, this relates to the fact that unlike stroke, cognitive decline is not a dichotomous event but occurs along a continuum. Among demographic predictors, older age and fewer years of education have been associated with increased risk of postoperative cognitive decline (Newman et al., Reference Newman, Croughwell, Blumenthal, Lowry, White and Spillane1995). Certain comorbid medical conditions, such as history of cerebrovascular disease, peripheral vascular disease, or diabetes also appear to be strong predictors of adverse neurological and cognitive outcomes (Ho et al., Reference Ho, Arciniegas, Grigsby, McCarthy, McDonald and Moritz2004; Collison, Smith, & Engel, Reference Collison, Smith and Engel2006). However, the presence of these conditions by themselves does not reliably predict outcomes, in part because information about severity or duration of the comorbid conditions may not be readily available. Genetic risk factors for postoperative cognitive decline have also been examined, but to date, there is no evidence that the APOE e4 genotype is associated with increased risk of postoperative cognitive decline (Silbert, Evered, Scott, & Cowie, Reference Silbert, Evered, Scott and Cowie2008; Hogue et al., Reference Hogue, Fucetola, Hershey, Freedland, vila-Roman and Goate2008).
CABG as a Risk Factor for Alzheimer’s Disease
The results from the Duke 5-year outcome study also raised the issue of whether patients who undergo CABG might be at greater risk for eventually developing Alzheimer’s disease (Brown, Moody, Tytell, Ghazi-Birry, & Challa, Reference Brown, Moody, Tytell, Ghazi-Birry and Challa1997). A study using case-control methodology concluded that CABG is not a major risk factor for dementia or Alzheimer’s disease (Knopman, Petersen, Cha, Edland, & Rocca, Reference Knopman, Petersen, Cha, Edland and Rocca2005). By contrast, in a retrospective cohort analysis of Veteran Affairs patients, patients who had undergone CABG were found to be at greater risk for the emergence of AD within a 5-year follow-up window after the procedure than patients who had undergone stenting (Lee, Wolozin, Weiss, & Bednar, Reference Lee, Wolozin, Weiss and Bednar2005). The overall incidence rates of AD reported for both study groups, both with known cardiovascular disease, were significantly lower than previously published 5-year incidence rates of dementia in community cohorts (Fitzpatrick et al., Reference Fitzpatrick, Kuller, Ives, Lopez, Jagust and Breitner2004), particularly when considering that some studies have found a higher incidence of dementia in patients with risk factors for cerebrovascular disease (Newman et al., Reference Newman, Fitzpatrick, Lopez, Jackson, Lyketsos and Jagust2005; Luchsinger et al., Reference Luchsinger, Reitz, Honig, Tang, Shea and Mayeux2005). Therefore, is it perhaps worth considering the possibility that the low rates of AD in the VA study could be because cardiac revascularization procedures may have a protective effect against the progression of cerebrovascular disease? Further epidemiologic studies would be needed to explore this possibility.
Summary
Cognitive decline after conventional CABG with CPB does occur in some patients, but lack of control groups and other methodological issues have made it difficult to determine if this decline is specifically related to the use of CPB, anesthesia, or other factors. Older patients and those with comorbid conditions associated with increased risk of cerebrovascular disease of the brain are the most likely to have postoperative cognitive decline. For those patients who do suffer some cognitive decline, it is generally transient and reversible by three to six months after surgery. Cognitive outcomes after off-pump surgery do not appear to be significantly different from those after conventional on-pump surgery, suggesting that factors related to the use of cardiopulmonary bypass alone cannot account for adverse cognitive outcomes. More recent studies have therefore focused on patient-related as opposed to procedure-related variables in the search for likely etiologies of postoperative cognitive decline. Some have suggested that preoperative cognitive testing could be one strategy for identifying patients who are at greatest risk for cognitive decline after CABG. Because the window of opportunity for such testing is generally quite brief, it is unclear whether it could be implemented outside of elective surgeries. Another hurdle would be to convince cardiologists and cardiac surgeons of the actual predictive value of preoperative cognitive assessment. Long-term cognitive decline has been shown to occur five or more years after CABG with CPB. However, in studies with appropriate controls, it has been shown that a similar degree of late decline also occurs in patients undergoing off-pump surgery, and in medically treated patients. This suggests that the late cognitive decline may be associated with progression of cerebrovascular disease rather than late effects of cardiopulmonary bypass. There is no evidence to suggest that CABG with CPB increases the risk of developing dementia or Alzheimer’s disease. Studies to examine the effects of stricter postoperative management of modifiable risk factors for cerebrovascular disease, such as blood pressure, cholesterol, diet, and exercise, on long-term cognitive outcomes could potentially yield strategies to reduce the incidence of late cognitive decline after CABG surgery.
ACKNOWLEDGMENTS
This study was supported by grant 35610 from the National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD, USA; by the Charles A. Dana Foundation, New York, NY, USA and the Johns Hopkins Medical Institution GCRC grant RR 00052. We thank Guy M. McKhann, Maura Grega, Scott Zeger, Luu Pham, Maryanne Bailey, William Baumgartner, Pamela Talalay, cardiac surgeons and cardiologists at Johns Hopkins Medical Institutions who helped with our studies.
