| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the *Department of Anesthesiology and General Intensive Care, Division of Cardiothoracic and Vascular Anesthesia and Intensive Care; Department of Surgery, Division of Cardiothoracic Surgery, University Hospital Vienna, Austria.
Address correspondence and reprint requests to Martin Dworschak, MD, Division of Cardiothoracic and Vascular Anesthesia and Intensive Care, University Hospital Vienna, Waehringer Guertel 18-20, A-1090 Vienna, Austria. Address e-mail to martin.dworschak{at}meduniwien.ac.at.
 |
Abstract |
|---|
 Top Abstract IntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
0.05). The difference between preoperative and postoperative Digit Span backward scores correlated with the increase in serum NSE levels (r2 = 0.3, P 0.05). Moreover, P300 latencies increased in 13 of 17 ICD patients, but decreased in 7 of 10 PM patients (P 0.05). PM patients even improved in the Trailmaking test (P 0.05). Neuronal injury from even brief periods of global brain ischemia seems to be associated with deteriorating neurocognitive function. |
Introduction |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Not surprisingly, cognitive dysfunction and minor neurologic deficits have been reported after ICD implantation under general anesthesia (4). At our institution, ICDs, like cardiac pacemakers (PM), are routinely implanted with monitored anesthesia care under local anesthesia, with ICD patients receiving small doses of etomidate immediately before induction of ventricular fibrillation. Thereby, a potential influence of general anesthesia on postoperative cognitive function can be minimized. In this setting, we could observe an increase in biochemical serum markers of neuronal injury (i.e., serum neuron-specific enolase [NSE] and S100) postoperatively without major neurological dysfunction (5). However, we did not test for postprocedural cognitive deficits that may be less obvious at first but similarly disabling.
Certain psychometric tests—such as the Mini Mental State Examination, the Digit Span, and the Trailmaking A test, as well as P300 auditory evoked potentials—have been used to evaluate postoperative cognitive brain function (6,7). Thereby, attention, short-term memory, psychomotor function, and the ability to discriminate can be assessed and dementia can be excluded in the early postoperative period. In previous studies, for example, atrial fibrillation after coronary artery bypass surgery as well as mitral valve replacement had a negative effect on P300 peak latencies and on Trailmaking A test results (6,7). In these patients, P300 peak latencies and the time required to complete the Trailmaking A test were both prolonged.
The goal of this study was to elucidate whether cognitive impairment after ICD insertion can be detected in our setting of ICD implantation and whether it is associated with the increase in serum NSE, i.e., the degree of neuronal injury that we have previously reported after ICD implantation (5).
|
METHODS |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Preoperatively, patient weight, medical history, left ventricular ejection fraction (LVEF), and serum creatinine values were determined and recorded. Exclusion criteria were cardiopulmonary resuscitation (CPR) during insertion and preexisting neurologic impairment. Women of childbearing potential were required to produce a negative pregnancy test for inclusion.
Anesthetic service was provided in the form of monitored anesthesia care. Patients received 0.02 mg/kg midazolam IV as premedication immediately before surgery and a bolus of 3 mg of piritramide (Dipidolor®), a synthetic opioid, when they felt pain despite adequate infiltration of local anesthetic for analgesia. ICD patients additionally received 0.1 mg/kg etomidate IV just before induction of ventricular fibrillation. Monitoring consisted of electrocardiogram, pulse oximetry, capnography via an oxygen facemask, and arterial blood pressure (invasive in ICD patients).
Before surgery, at the end of surgery, and 2, 6, and 24 h after surgery, blood for blood gases, electrolytes, glucose, free hemoglobin, and NSE were drawn either from an IV line or from the arterial catheter when in place. NSE was measured using an NSE enzyme immunoassay test kit (Cobas Core NSE EIA II; Hoffmann-La Roche, Diagnostic Division, Basel, Switzerland), which has a functional sensitivity of 0.25 g/L. The 95th percentile of normal serum concentrations is reported by the manufacturer to be 15.2 g/L. Values of 12.5 g/L, and in some investigations, 9.2 g/L, are considered to be increased. Visibly hemolytic samples had to be discarded and increased NSE values in conjunction with plasma free hemoglobin of 
18 mg/dL were not to be used for statistical analysis.
Forward and backward Digit Span tests (8), the Mini Mental State Examination (9), and the Trailmaking A test (10) have been used to evaluate neurocognitive function in patients affected by brain ischemia and also in people with an intact cerebral circulation.
In a subset of patients that included patients from a preliminary study who were managed with the same protocol, cognitive evoked P300 potentials were assessed. For this purpose we used the Nicolet 2000 machine (Nicolet, Madison, WI) and determined the respective latencies. Ag/AgCl electrodes were placed at the vertex (Cz) and at the frontal cortex (Fz) and referenced to an earlobe A1/2 electrode. Patients had to discriminate rare tones (20%) at 2000 Hz from frequent tones (80%) at 1000 Hz both applied at 75 dB. Two sets of measurements were performed to ensure reproducibility, and a low band-pass filter (0.01–30 Hz) was used (7).
Neurologic assessment, consisting of the psychometric test battery and, in a subgroup of patients, P300 evoked potentials, was always performed by the same experienced investigators (MW, RF) 1 day before and 2 days after surgery. None of these patients needed analgesics or sedatives during the evaluation, which was always scheduled at the same time of day.
A power calculation conducted to detect a change of 1.5 sd in NSE at a significance level of 5% with a power of 80% revealed that a sample size of 21 patients per group would be required. Furthermore, data from Grimm et al. (7) suggest that statistically significant group differences in the Trailmaking A test and in P300 latencies should be detectable in a cohort of 40 patients. The 
2 test was used to compare differences in patient characteristics, and linear regression was used to determine significant associations between the increases in serum NSE and psychometric test scores. Student's t-tests were used to evaluate differences from baseline within groups (paired Student's t-test) and differences between groups (unpaired Student's t-test). In the case of an uneven data distribution or uneven group size, we used the Wilcoxon's signed rank test and the Mann-Whitney U-test, respectively.
In a previous investigation we found that NSE serum levels increase until 24 h after ICD insertion (5). This was also the time point when the largest differences in NSE between ICD and PM patients were detected. We thus assumed that this also reflects the time the greatest neuronal injury becomes apparent. Therefore, we chose that time point to relate the difference in serum NSE levels from their corresponding baseline values to the preoperative and postoperative difference in the Digit Span backward scores obtained in each patient. We used StatView 5.0 (SAS Institute Inc., NC) statistical software. A P value of 0.05 was considered significant. Values are reported as means ± sem.
|
RESULTS |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
0.05). More male patients were in the ICD group (P 0.05). Patients of this group also showed a reduced left ventricular pump function (LVEF < 50%; 17 versus 8 patients; P 0.05). Only one patient in the PM group exhibited an increased serum creatinine level preoperatively (3.48 mg/dL). Preoperative neurologic incidents (i.e., sleep apnea in one patient of each group and prior stroke without persisting neurological symptoms in three PM patients and one ICD patient) were reported more frequently in the PM group.
|
Induced ventricular fibrillation caused an immediate decrease of mean arterial blood pressure, i.e., perfusion pressure, with loss of pulsatility, which we used to determine the beginning of CA. The end of CA was defined as reappearance of pulse pressure on the pressure curve. There was, however, some variability in the amount of time needed to return to pre-arrest arterial blood pressure. On average, 3 arrests were induced, which resulted in a mean duration of CA during ventricular fibrillation of 10 s and in an average cumulative duration of CA of 25 s. The mean duration of the reperfusion period, i.e., the time between 2 successive shocks, was 305 s.
Laboratory Results
Baseline neuron-specific enolase (NSE) serum levels were not different between groups. Serum NSE increased over time in the ICD group, whereas it remained at baseline level in PM patients. Values immediately after surgery and at 6 and 24 h after surgery were significantly higher in ICD patients as compared with baseline (P 0.05, Wilcoxon test). Group differences were noted at the end of surgery and 24 h thereafter (P 0.05, Mann-Whitney U-test) (Fig. 1A). The increase from each individual's NSE baseline level is depicted in Figure 1B. This variable better reflects the true increase over time in each patient. Nevertheless, we found a strong correlation between absolute NSE values and the difference from baseline (r2 = 0.42, P = 0.001, Spearman Rank correlation). Significant differences between groups were seen in the second, fourth, and fifth samples, i.e., immediately after as well as 6 and 24 h after surgery (P 0.05, Mann-Whitney U-test) (Fig. 1B). These values were also significantly different from baseline in the ICD group (P 0.05, Wilcoxon test).
|
No hyperglycemic, hypercapnic or hypocapnic events were recorded in either group. Furthermore, no signs of severe hemolysis, evident as increased free hemoglobin levels, could be determined. Therefore, no NSE value had to be rejected. In addition, serum sodium and potassium levels were not altered throughout the study period in any patient.
Psychometric Tests and Acoustic Evoked Potentials
Digit Span Forward Test
Baseline scores were slightly lower in PM patients (43 ± 2 versus 45 ± 2 of 60 attainable points). After surgery, scores in PM and ICD patients decreased significantly in each group (P 0.05). However, the groups' scores were similar postoperatively (38 ± 3 for the PM group versus 39 ± 3 for ICD patients; P > 0.05). There were also no statistically significant differences when the differences between preoperative and postoperative values were compared between groups.
Digit Span Backward Test
Contrary to the Digit Span forward test, this test requires that a series of numbers heard be repeated in reverse order. It therefore requires more attention on the part of the subject being evaluated. The maximum score attainable on the Digit Span backward test is 50 points. Baseline scores obtained before surgery were significantly lower in PM patients (P 0.05) (Fig. 2A). Scores decreased in both groups postoperatively, yet the decline was only significant in ICD patients (P 0.05). Moreover, a weak positive correlation was found between the preoperative and postoperative score differences (i.e., larger score difference meaning greater deterioration in attention and memory) and the increase from baseline in NSE values obtained 24 h after surgery (P 0.05, r2 = 0.3) (Fig. 3). Less pronounced correlations were also found between the postoperative increase in serum NSE values and the difference in the preoperative and postoperative scores of another two tests, namely, the Digit Span forward test and the Mini Mental State Examination. However, after Bonferroni correction for multiple testing only the correlation between the postoperative increase in serum NSE levels and the preoperative versus postoperative difference in the Digit Span backward test remained significant.
|
|
Trailmaking A Test
Before surgery, the amount of time needed to complete the Trailmaking A test was longer (P 0.05 versus ICD group) in PM patients; i.e., in relation to the ICD group they needed more time to correctly connect the numbers they were shown on a piece of paper. However, PM patients improved their skills to complete the same task after the operation; i.e., they needed less time to complete the test as compared with baseline (P 0.05) (Fig. 2B). In contrast, ICD patients were not able to connect the numbers more quickly after surgery. The preoperative time difference between groups was also significantly larger than the postoperative difference (P = 0.02).
Mini Mental State Examination
Mean preoperative and postoperative scores were in the normal range in both groups (i.e., >28 of 30 obtainable points). Thus, no signs of dementia (<27 points) were observed.
P300 Latencies
Twenty-seven patients (17 ICD and 10 PM patients) were evaluated. PM patients in whom P300 acoustic evoked potentials were determined were older than ICD patients (74 ± 2 versus 61 ± 3 yr; P 0.05) and showed increased latencies at baseline (412 ± 20 versus 358 ± 12 ms; P 0.05). Latencies, however, decreased in the PM group after surgery, whereas they were significantly prolonged (P = 0.005, Mann-Whitney U-test) in ICD patients (371 ± 13 versus 426 ± 19 ms). The difference between postoperative and preoperative values was also significantly different between PM and ICD patients (–41 ± 21 versus 67 ± 22 ms; P = 0.003, Mann-Whitney U-test). Thirteen of 17 ICD patients as compared with 3 of 10 PM patients demonstrated prolongation of P300 latencies (P = 0.04; Fig. 4).
|
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In a previous study (5) we observed that two serum markers of neuronal injury, S100 and NSE, increase after ICD but not after PM implantation performed under similar conditions. NSE is a dimeric enzyme almost exclusively located in neurons and cells of neuroectodermal origin. Smaller amounts are detectable in red blood cells and may be released in the course of hemolysis. Some tumors, such as neuroblastomas, small cell carcinomas, and APUDomas, can also synthesize NSE, which can thus serve as a tumor marker, as it is already increased in serum under normal conditions. Unlike S100ß, no significant extracerebral source has been implicated in the increase of serum NSE apart from tumor secretion and hemolysis. NSE release has additionally been observed after severe shock. However, severe shock is generally associated with cerebral hypoperfusion. All of these causative factors could be excluded in this study. The only difference between the two groups was that ventricular fibrillation was induced in ICD patients but not in PM patients. Consequently, the increase in NSE in this group should be attributable to transient hypoperfusion with accompanying global cerebral ischemia. This is emphasized by our previous observation that NSE was extremely elevated postoperatively in a patient who had to be resuscitated in the course of the implantation because of ventricular fibrillation that could not be terminated by internal and external shocks.
As previously described, NSE reached its maximum between 6 hours after surgery and the end of the 24-hour observation period, after which evaluation was terminated (5). Whether it is continuing cell death or simply a time delay until released NSE crosses the blood-brain barrier that accounts for the continuous increase in serum levels cannot be answered here. Despite this increase, no major neurological alterations could be detected in these patients. However, in 1997 Murkin et al. (4) reported cognitive dysfunction and minor neurologic deficits after ICD insertion. In contrast to the patients we studied, these patients were managed with general anesthesia, which may have influenced postoperative neurocognitive tests. Furthermore, it has been shown that in particular, cognitive dysfunction as a result of global cerebral hypoperfusion may not be readily obvious (11).
Successfully resuscitated victims of CA suffer primarily from impaired cognitive function that seems to correlate with the delay in the start of CPR (12). In this investigation, witnessed CA, early CPR (<1 min), early defibrillation, and shockable rhythm, as well as the absence of the need for advanced cardiopulmonary life support, resulted in better cognitive, physical, and psychosocial function. Interestingly, extreme delays in CPR were not related to poorer functioning than moderate delays. In addition, the shorter arrest durations of in-hospital CAs as compared to out-of-hospital CAs did not protect against memory impairment (8). Both of these observations can be explained by neuronal damage obviously occurring soon after onset of CA.
To elucidate whether the postoperative increase in NSE we observed previously is related to cognitive brain function, we performed three psychometric tests and investigated P300 auditory evoked potentials in a subset of patients. These tests and P300 peak latencies have been used before in the postoperative setting after cardiac surgery (7,13). In the present study it could be demonstrated that cognitive performance worsened after surgery only in ICD patients. This group also exhibited an increase in NSE serum levels, which confirms our previous finding (5). Moreover, this increase showed a positive correlation with the decline in the performance in the Digit Span backward test. Multiple testing as it was conducted here, however, has the limitation that some correlations can be significant by mere chance. This restriction makes it necessary to perform post hoc corrections. After Bonferroni correction, only the change from baseline in NSE values (supposedly the best indicator of the true increase in this serum marker for each individual patient) at 24 hours after surgery, and the difference in the Digit Span backward score (the most difficult test that requires the most attention) remained significant in the Spearman Rank correlation.
As the same tests were repeated after two days, one would also expect a positive learning curve to cause improvement in the patients' scores now having completed the test twice (14). This was actually the case in control (PM) patients in the Trailmaking A test. Furthermore, ICD patients did not show a positive learning effect in either test, which implies that short-term memory, attention, and learning seem to be more severely impaired in this group. As the Digit Span test evaluates memory, it is not surprising, and actually underlines the fact, that it is this cognitive function in particular that seems to be affected the most after ICD placement (13). In the setting of global brain ischemia, a potential neuroprotective potency has been ascribed to etomidate, among other anesthetics (15). Therefore, etomidate can be assumed to have somewhat mitigated the neurocognitive decline we observed in ICD patients. In other words, had we used another drug without neuroprotective effect, cognitive dysfunction might have been even worse.
P300 latencies increase with age and with the occurrence of atrial fibrillation. This may account for the longer latencies in our PM patients before surgery (16). In line with the psychometric test results, P300 latencies improved in two thirds of the PM patients but increased (i.e., got worse) in two thirds of the ICD patients. Shorter postoperative peak latencies in the PM group associated with improved performance in the Trailmaking A test may be explained by a positive learning curve, as mentioned before. A further explanation could be enhanced brain perfusion after PM implantation in patients with severe bradycardic episodes with compromised cardiac output (17). In contrast, the generally lower ejection fractions we determined in our ICD patients with presumably impaired reperfusion after CA may have contributed to the neurocognitive decline we observed in this group after implantation.
In this investigation, significant postoperative increases in serum NSE could only be detected in the ICD group, which also showed impaired cognitive performance after surgery. Similar results have been obtained after cardiac surgery, where an increase in serum NSE appeared to be a good marker for postoperative cognitive dysfunction (18). Although the hippocampal area is particularly vulnerable to hypoxia and ischemia, one cannot safely assume that the NSE increase we observed is primarily caused by injured hippocampal neurons. It may rather stem from various dying neurons from different parts of the brain. Although focal hippocampal injury has been linked to cognitive dysfunction, it seems more likely that it is a global reduction of cerebral tissue that ultimately affects memory (19).
A limitation of this study is the fact that patients could only be evaluated until postoperative day 3, when they were discharged from the hospital. Long-term follow-up is required to determine whether potentially disabling cognitive deficits persist or occur only transiently after surgery (20).
A further drawback is that the control group we chose is not really a perfect control group, as the patients were not matched for age, gender, heart rhythm, or LVEF. This difference in patient characteristics is most likely the reason for the observed differences in the preoperative P300 latencies and the psychometric test results (21). Patients who receive an ICD device because of intractable ventricular fibrillation are generally younger and show different comorbidities, usually with signs of ventricular dysfunction, which is often the cause of the malignant arrhythmia. Quite frequently, an ICD device serves as a bridge to transplant in these patients. In contrast, patients who need a cardiac PM are often older women who suffer from sick sinus syndrome and atrial fibrillation. This is why these two groups are difficult to match. Nevertheless, in our opinion, it is the best possible control group, as the surgical procedure is almost identical in duration and in performance.
Although PM patients started from a lower level, improvements in the Trailmaking A test and P300 acoustic evoked potentials without a decline of the Digit Span backward scores could be detected postoperatively. This strikingly contrasts with a decrease in both Digit Span test scores and a prolongation of P300 potentials that became evident in the ICD group.
In conclusion, even minor neuronal injury after repeated brief periods of global cerebral ischemia during ICD implantation appears to be associated with postprocedural cognitive dysfunction that primarily affects attention, memory and learning. PM patients, however, demonstrated seemingly improved neurocognitive performance after insertion.
|
Footnotes |
|---|
This study was supported by institutional funding only.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
P 
