Why Are Cerebral Microemboli Not Associated with Cognitive Decline?

In 1996, Roach et al.1 published a prospective cohort study that evaluated neurologic outcomes in 2108 coronary artery bypass grafting (CABG) surgery patients from 24 United States institutions. Neurologic outcomes were classified into two categories. Type I outcomes were defined as focal neurological injury, stupor, or coma at hospital discharge. Type II outcomes were defined as deterioration in intellectual function, disorientation, memory deficit, or seizures, without evidence of focal injury. Adverse cerebral outcomes occurred in 6% of the patients. Three percent had Type I neurologic outcomes and 3% had Type II outcomes. Notably, in-hospital mortality for patients with Type I outcomes was 21% and for patients with Type II outcomes 10%. Moreover, half of the patients with Type I outcomes and one-third of the patients with Type II outcomes were discharged to skilled nursing facilities or rehabilitation centers.1 Type II outcomes are comparable with postoperative cognitive decline (POCD), which has been evaluated in a large number of studies in cardiac and noncardiac surgical patients.2,3 Although Roach et al.1 based Type II outcomes on clinical data and hospital discharge summaries evaluated by an expert panel, POCD is more often diagnosed with repeated administration of a battery of psychometric tests.2 Compared with an expert panel, psychometric tests are believed to be much more sensitive for detecting subtle cognitive changes.4 It is, therefore, not surprising that studies evaluating POCD with psychometric tests have consistently reported a higher incidence of POCD than the 3% Type II outcomes found by Roach et al.1 Depending on the timing of neuropsychological assessments and the definition of POCD, the reported incidence of POCD after cardiac surgery varies between 20% and 70%.2 The high incidence of POCD after cardiac surgery has been attributed to the use of cardiopulmonary bypass (CPB).2,5 This seems reasonable because several studies have demonstrated potentially harmful side effects of CPB on the brain. Brain swelling can be observed in patients who undergo magnetic resonance imaging immediately after CABG surgery using CPB.6 Serum concentrations of S-100 protein, a biochemical marker of cerebral injury, are higher during “on-pump” than “off-pump” CABG surgery.7 Transcranial Doppler (TCD) monitoring has been widely used during CPB to detect cerebral microemboli,5 and postmortem examination of the brains of patients who had undergone cardiac surgery reveals evidence of cerebral emboli.8 Despite these disturbing side effects of CPB, however, randomized studies have failed to show a difference in the frequency of POCD between patients undergoing surgery with and without CPB.9,10 In this issue of Anesthesia & Analgesia, Liu et al.11 report a study that compared the effects of “off-pump” versus “on-pump” CABG surgery on cerebral emboli count and the risk of POCD. Intraoperative cerebral emboli were detected with bilateral TCD monitoring of the middle cerebral arteries, and POCD was measured with a psychometric testing battery administered preoperatively, and 1 wk and 3 mo after surgery. Avoiding CPB resulted in a 200-fold reduction of the number of intraoperative cerebral emboli. Despite the marked reduction in TCD-detected microemboli, however, there was no association between the use of CPB and the frequency of POCD. These results, and a body of other data, illustrate a major dilemma of POCD research.11 We know that CPB is associated with cerebral embolism and with POCD. Why then, does not avoiding CPB and reducing the number of cerebral microemboli result in improved cognitive outcomes? One explanation might be that cerebral microemboli do not cause persistent cognitive impairment. It is difficult to distinguish gaseous microemboli from solid microemboli with TCD, and it has been suggested that properties such as size and composition instead of the number of microemboli determine the risk of adverse cerebral outcomes.2,5 Gaseous microemboli during CPB may dissolve within seconds and could have minimal effects on brain function. Likewise, cerebral edema after CPB may have a benign course. None of the patients with brain swelling immediately after CABG surgery had edema during repeated magnetic resonance imaging 6 days later.6 It is also conceivable that advances in CPB technology have resulted in improved cognitive outcomes. It is, therefore, possible that in contemporary practices, the risk of POCD is largely determined by factors other than CPB. This may include the inflammatory response to major surgery, certain anesthetic drugs, the presence of generalized atherosclerotic disease, and advanced age.2,11 For the development of long-term cognitive decline, even the causal role of CABG surgery in itself is now uncertain, regardless of the use of CPB. Selnes et al.12 have recently compared cognitive outcomes at 6-yr follow-up between 152 CABG surgery patients and 92 nonsurgical cardiac comparison patients. Compared with baseline performance, mild late cognitive decline was observed for both groups, but there were no differences between the surgical and nonsurgical patients. The Octopus Study Group13 performed a largely similar study and compared 281 CABG surgery patients with 112 healthy subjects without coronary artery disease. At 5-yr follow-up, a statistically significant difference in cognitive decline between the CABG surgery patients and the healthy subjects could not be demonstrated. Both studies suggest that natural aging contributes much more to long-term POCD than undergoing CABG surgery.12,13 A final possible explanation for the findings that avoiding CPB and reducing cerebral microembolism do not result in a lower frequency of POCD is that the psychometric tests used have limited precision.14,15 These tests are administered to patients at baseline, usually the day before their surgery, and a second time after their surgery. When a patient’s postoperative performance is lower than his or her preoperative performance by some degree (e.g., −1sd or −20%), he or she is classified as having POCD. However, most psychometric tests were never designed to detect within-subject cognitive decline over time, and it has become apparent that there is considerable natural variation in test performance when subjects undergo the same test twice.14,15 It appears to be very difficult to discriminate true cognitive decline from natural variations in psychometric test performance.14,15 Studies in healthy subjects have shown, for example, that most classical definitions of POCD using a fixed cutoff value (such as a 1sd or 20% lower performance than at baseline) generate a high rate of false positive results and thus overestimate the incidence of POCD.14,15 The assumption that POCD based on repeated psychometric testing is a precursor or surrogate for clinically relevant cognitive decline is therefore questionable. Indeed, the correlation between POCD according to psychometric tests and subjective, patient-reported POCD is poor.16 Considering the possibility of a declining true incidence of POCD, measurement errors with psychometric testing, and the multifactorial etiology of the condition, it is easy to understand why even prospectively randomized studies have failed to establish a relationship between CPB and POCD.9,10 The studies lacked statistical power, and instead of a few hundred patients, thousands of patients would be needed to prove or exclude the association between microemboli from CPB and cognitive outcome. Repeated psychometric testing of thousands of patients, however, is laborious and expensive. Perhaps we should return to the approach used by Roach et al.1 and use clinical criteria to identify patients with Type II cerebral outcomes. An adjudication committee evaluating the patient’s hospital course, the need for admission to advanced care facilities, and a patient/family questionnaire might be more appropriate than psychometric tests to detect the cognitive outcomes that are relevant to the patients themselves.