Neuropsychological Change and S-100 Protein Release in 130 Unselected Patients Undergoing Cardiac Surgery
Background and Purpose—S-100 protein promises to be a valuable surrogate end point for cerebral injury. This is of particular interest within the context of cardiac surgery. We sought to explore the relationship between change in neurospychological performance attributable to cardiopulmonary bypass and the release of brain-specific S-100 protein.
Methods—In an observational comparative study in a University Hospital Cardiac Surgical Unit, S-100 protein release during and 5 hours after the onset of cardiopulmonary bypass was compared with change (from preoperative to 6 to 8 weeks postoperative) in neuropsychological tests in 130 patients undergoing the full range of cardiac surgical procedures.
Results—Neuropsychological performance usually improved, being significantly so in 10 of 25 parameters. S-100 area under the curve (AUC) protein release correlated with age (r=0.24, P<0.008) and bypass time (r=0.17, P<0.02). S-100 Cmax correlated with bypass times (r=0.29, P<0.0001). Bypass times correlated with memory performance (Rey R5; r=−0.21, P<0.03). Less S-100 protein release was associated with better neuropsychological performance, as indexed by significant correlations with the Rey Auditory Verbal Learning memory test, descending Critical Flicker Fusion thresholds, and the Hospital Anxiety and Depression rating scales, typically around r=0.2. Multiple regression models showed that neuropsychological tests accounted for 23% of the variance associated with S-100 AUC release, after partialing out the effects of age and bypass time.
Conclusions—The correlation between S-100 protein release and neuropsychological function supports the belief that it is a measure of brain injury, which may be useful in future studies of mechanisms and prevention.
Cerebral complications of cardiac surgery are now well characterized,1 and a great deal more is known about them than 10 years ago, when we2 3 and others4 5 first made systematic attempts to study this problem. For routine cardiopulmonary bypass, the risk of both diffuse brain injury and stroke has been reduced; however, we operate on ever-older patients, with worse atherosclerotic vascular disease, so it remains a significant issue.
There are 2 general patterns of brain injury seen after cardiopulmonary bypass.1 Stroke is the more obvious, associated with 1 or more discrete cerebral infarcts, resulting in a range of neurological manifestations, which depend on site and size. Stroke is relatively uncommon after cardiopulmonary bypass, and its occurrence is sporadic. The second pattern is common and was demonstrated, in our earlier studies,2 3 as a deterioration in the ability to perform neuropsychological tests for 6 to 8 weeks after surgery in approximately 25% of routine coronary patients. It is a diffuse injury, not characterized by discrete infarction or focal neurological signs. In clinical practice, there may be an interplay between the 2 patterns. For example, atheroembolism6 may contribute to both patterns, whereas microembolism of air bubbles causes a more subtle evanescent and apparently transient form of brain injury.7 8
A major difficulty in studying this problem has been the establishment of sensitive and specific measures of outcome that can be used reliably and efficiently. Neuropsychological tests have been the mainstay of this research9 but have several major drawbacks. They are time consuming, taking as much as an hour to perform, and there must be a preoperative baseline so that change can be determined. Their use is therefore limited to elective, stable patients, who can cooperate with the testing process before and after operation. Particular groups, including infants, the critically ill, and those with a different language, are automatically excluded from such studies. Furthermore, the results can be obscured by learning effects on test-retest, mood state, and medication. Nevertheless, neuropsychological tests have been used in controlled randomized trials to demonstrate the benefit of the use of an arterial filter in the bypass circuit,7 the hazards of glucose in the prime,10 and the advantage of alpha-stat over pH-stat perfusion11 ; however, we are seeking subtle changes, and very large studies will be required to show further improvement.
S-100 protein release is a plausible candidate for a sensitive and specific marker for brain injury.12 13 We have shown that S-100 protein release increases commensurately with the anticipated degree of cerebral hazard,14 and a consistent relationship to routine bypass has also been shown.15 We had found that the use of an arterial filter improved performance on neuropsychological tests in a trial of 100 patients.7 This has been reevaluated with S-100 protein as a surrogate marker, and in a study of only 40 patients, there was a significant difference.16
Although it is known individually that cardiac surgery is associated with S-100 protein release and neuropsychological deficit, no one has directly graduated S-100 with neuropsychological deficit. It is the relationship between these 2 outcomes, cognitive change and S-100 protein release, which we set out to study. If S-100 release is predictive of change in performance on neuropsychological tests after bypass, it could serve as a surrogate in prospective studies of a range of strategies aimed to optimize cardiopulmonary bypass, including choice of temperature, pulsatility, pH management, and putative neuroprotective drugs.
Subjects and Methods
All patients scheduled for cardiac surgery involving cardiopulmonary bypass who were able to complete a set of preoperative neuropsychological tests were approached and asked to participate in the study. These included urgent cases awaiting surgery in hospital, but unstable and emergency cases were excluded. Patients with recent stroke were excluded. Ethical approval was obtained. The study began in November 1996 and concluded on October 1, 1997. Before surgery, patients had the purpose of the study explained to them by the research nurse and gave written informed consent. Patients were free to withdraw at any time without necessarily giving an explanation.
Patients were tested in a private interview room by a psychologist (S.K.). The tests included those used in over 500 patients in previous studies: Rey Auditory Verbal Learning Test (RAVLT), Digit Symbol Substitution Test (DSST), Trail-Making Test (TMT A+B), Spot the Word Test (as an index of pre-morbid IQ), Speed and Capacity of Language Processing (SCOLP), and Critical Flicker Fusion (CFF). Patients also completed the Cognitive Failures Questionnaire (CFQ), the General Health Questionnaire 28 (GHQ), and the Hospital Anxiety and Depression Rating Scale (HADS). The tests are well described and validated,2 9 and most are used routinely in clinical practice in the United Kingdom. They were administered before surgery and 8 weeks after surgery.
Blood was taken for S-100 protein analysis at specific time points: before bypass and at 0.5, 1, 1.5, 2, 3, and 5 hours after going onto bypass, as guided by pilot study data.14 From these we derived 2 measures: the area under the curve (AUC), which is an estimate of total protein release, and the maximum plasma concentration (Cmax). S-100 protein was measured with the commercial Sangtec S-100(R) kit (Sangtec Medical Sweden). This is a monoclonal, 2-site immunoradiometric assay. The Sangtec assay measures the beta subunit of the protein S-100, which is specific to Schwann and Glial cell damage. The sample is incubated with 125I-labeled monoclonal antibody to S-100 protein. The concentration of S-100 in the sample is determined by measuring the radioactive count rate against that of known calibration standards.
Plots of S-100 Cmax and AUC data versus age, bypass time, and all neuropsychological change scores were studied for relationships. Correlations were examined with use of the Pearson product moment for age and bypass time and AUC release. Other more detailed analyses of interactions between neuropsychological tests, risk factors, and S-100 protein (stepwise multiple regression, analysis based on z scores and ranked poorest 15% of change scores on neuropsychological testing) were also carried out.
Of the 130 patients aged 63.5±8.8 (mean ±SD) years who entered the study (Table 1⇓), 84% had coronary artery bypass grafts, with 4 of 109 being second operations; 16 (12%) had single-valve operations; and the others had various combinations of valve and coronary artery surgery, with a single case of atrial septal defect. Bypass time varied from 0 to 290 minutes, with a mean of 84.5 (SD 37.6) minutes and a median of 81 minutes. In 2 patients coronary artery grafts were performed without the need for cardiopulmonary bypass, but these were excluded from the study analysis. Only measurement of their S-100 protein was taken to act as control for nonextracorporeal circulation. S-100 protein was found in 1 patient. Of those who went on a bypass machine, 11 of 128 showed no protein release.
S-100 protein release was measured in all of these cases and is displayed as maximum concentration (Cmax), AUC, and its logarithm (log AUC). Table 3⇓ shows the various correlations between protein, age, and neuropsychology tests. Scattergrams showed these to be all linear relationships.
Age correlated with AUC (r=0.24, P<0.008), Cmax (r=0.28, P=0.002), and log AUC (r=0.33, P<0.0001). Bypass time correlated with AUC (r=0.20, P<0.02), Cmax (r=0.34, P<0.0001), and log AUC (r=0.33, P<0.0001). Older patients had significantly higher protein release (AUC and Cmax). Multiple regression models showed that age and bypass time independently accounted for significant proportions of the variance associated with protein release. The proportion of variance accounted for by both age and bypass time (with n=128) was 8% for AUC, 19% for log AUC, and 15% for Cmax.
Of the 130 patients, 114 completed preoperative and postoperative neuropsychological tests. There were 6 deaths (4.6%), and 10 patients did not perform the second set of tests because of illness or refusal.
Tests of mood state (HADS) showed highly significant improvement, with the mean anxiety score falling from 8.2 to 5.3 and depression from 4.5 to 2.5 (t test, P<0.0001). It should be noted that all of these mean scores are below the cutoffs that would suggest clinical abnormality of mood. There was also significant improvement in the GHQ, from 6.1 (mildly clinically abnormal) preoperatively to 3.6 (normal) postoperatively (P<0.0001). Most cognitive test results improved or were unaffected by the operation (Table 2⇓).
Investigation of changes in cognitive test scores and self report questionnaires, from before to after the operation, by Student t test revealed a significant (P<0.05) improvement in 10 of 25 variables and a deterioration in the 3 derived memory test parameters of proactive interference. Changes in 8 tests remained significant after Bonferroni correction for multiple testing. These 7 tests pertained to speed of information processing (DSST, SCOLP), mood and psychological distress (HADS, GHQ), subjective view of memory ability (CFQ), descending CFF (DESCFF) indexing sedation, and also the previously mentioned subscale test of proactive interference from the Rey Auditory Learning Test (B/1+2).
There are competing movements in the data that need to be accounted for. In short, while the majority of patients showed a global improvement in each cognitive test postoperatively, there is a subset of patients whose test results remained stable or were impaired. Taking average (mean) scores masks these 2 movements. Additionally, although there was a global improvement, as indexed by >1 test showing improvement, some patients’ scores may have deteriorated on some tests and improved on others. To provide a more thorough understanding of the data, analyses relating S-100 protein release to neuropsychological function were carried out at 3 levels, and a summary of the analyses is given below.
Correlations With S-100 Protein
Pearson product moment correlations between the three S-100 protein variables or bypass time with the 28 psychological variables produced 17 significant results (see Table 3⇓). The direction of the correlation was in every case consistent with higher protein release/longer bypass times and poorer postoperative neuropsychological scores. All examples are were plotted to confirm linear relationships before correlations were carried out.
This analysis included the variable for (1) all change scores on psychology tests as independent variables and (2) age, sex, body mass index (BMI), Parsonnet risk scores,17 and bypass time. The measures of protein release were separately regressed upon the independent variables and subjected to stepwise regression procedures. The dependent variable was always the continuous measures of S-100 protein and was not dichotomized so that logistic regression was not required. The model that accounted for the highest proportion of variance (24%) is given below. The equation only includes variables which made a significant contribution in predicting S-100 protein release (multiple R=48.9). Because bypass time was included in the models, the 2 nonbypass cases were dropped out of the analysis. Where the model includes both positive and negative weightings for Rey memory subtests, this is probably due to the rebalancing of the underestimate/overestimate of variables entered into the stepwise model first and collinearity of the subscales. S-100 (AUC)=0.875+0.31×bypass time−0.3×R3+0.19 ×CFF+0.42×Rey B+0.19×TMTB−0.29×Rey B1/2
Psychological factors accounted for 14.4% of the variance and bypass time for 9.5%. For Cmax, 25.6% of the variance was accounted for by the following model (multiple R=50.6). S-100 (Cmax)=−1.43+0.41×bypass time+0.19×age+0.19 ×Rey B−0.16×Rey 4
In this case psychological factors accounted for 6% of the variance, bypass time 16.3%, and age 3.5%. Hence, after other factors were accounted for, psychological variables alone accounted for 6% to 14% of the variance.
A further method of regressing the data is by dichotomizing the change score data from psychological measures. Poor scores were defined as those in the lowest 15% of ranked scores. Poor performance was dummy coded 1 for those in the bottom 15% and 0 for those above this cutoff. The dependent variable was always the continuous measure of S-100 protein. Using this method, the proportion of variance explained by psychological tests increases to 23% for S-100 (AUC), with bypass time contributing another 9.5%. For Cmax, the figures are 19% (psychological tests) and bypass time another 16%.
Open Heart Versus Closed Cardiac Surgery
Although the subclasses of cardiac operation are not a focal point in the present research and the sample sizes are imbalanced (21 open versus 109 graft), there were some statistical differences between open and closed groups that some readers may find interesting.
Patients in the closed group were significantly heavier (81.3±12.3 versus 72.7±10.5 kg, P=0.003) and hada higher BMI than those in the open group (27.5±3.8 versus 24.2±3.1, P=0.0003). They had marginally (P=0.07) shorter bypass times (82±33.2 versus 97± 54.2 minutes, P=0.09) than the open-heart patients. These differences did not bring about any significant difference in S-100 Cmax or AUC protein release. There were no significant differences in any of the change in neuropsychology scores.
The group of 130 patients is quite typical of contemporary adult surgical practice: >80% undergoing coronary artery surgery, with the majority between 60 and 70 years of age and 25% >70 years. S-100 protein release during and after cardiopulmonary bypass correlated positively with both age and time of perfusion. All studies on cerebral consequences of cardiopulmonary bypass have found that age and bypass times are determinants, so this is consistent with the hypothesis that S-100 protein release will be a useful marker as a surrogate for brain injury in future studies.
There were some statistical differences between the open-heart (valves and atrial septal defect) and closed-heart groups that some readers may find interesting. The closed group were significantly heavier and had a higher BMI than the open group. They had marginally shorter bypass times than the open-heart patients. These differences did not bring about any significant difference in S-100 Cmax or AUC protein release, nor were there any significant differences in any of the changes in neuropsychology scores.
In 2 patients the surgeon performed coronary surgery without going on to bypass. One released a small amount (AUC=0.18 and Cmax=0.35) and the other no protein. The data are presented with these patients excluded, but whether they are included (on intention to treat) or excluded did not alter any findings. Of those who went on a bypass machine, 11 of 128 showed no protein release.
The improvement in the performance in some neuropsychological tests after surgery is a change from the findings of our first study.2 It has become evident in more recent studies10 that the end point in comparative studies would be failure to show improvement rather than absolute deterioration. This improvement could in part be attributable to the better feeling of psychological well-being (improvement in GHQ scores) and in mood state (lower HADS scores) found in the postoperative period in this study. It could also be an effect due to practice, increased confidence, a reduction in β-blocker usage, or reduced fatigue. In our earlier work these effects are presumed to have been overridden by the damaging effect of bypass as practiced at that time.
The significant correlations between S-100 protein release and change in neuropsychological test performance and the fact that the relationship is stronger with the dichotomized data supports a view that protein release is related to cognitive impairment. A cautious consideration is required, however, when interpreting findings such as these: first, because at best, neuropsychological tests account for only 23% of the variance associated with S-100 (AUC) release; and second, because there is no evidence pertaining to everyday functioning; for example, deterioration or no change from before to after operation can be associated with relatively high protein release, but we do not know whether it is associated with any measurable neuropsychological disability or handicap in daily life.
Prompted to pick the most discriminating neuropsychological test post hoc, the Rey memory test would be the obvious choice. Of note, none of our patients presented a memory impairment warranting an MRI scan, and as such the cerebral consequences of our cardiac patients studied here are well protected by our standard surgical and anesthetic protocols.
We did not find any association between Parsonnet score and S-100 release. The Parsonnet is a composite score used to assess the risk of serious adverse effects or death. It includes factors such as age, gender, left ventricular function, and mitral valve involvement. Certainly, age was found to be related to S-100 release; in a multiple regression equation, this variance would already be accounted for, leaving less-unique variance to contribute to the equations. It is also the case, that our sample of patients do fair very well, with a generalized improvement in postoperative psychometric scores. In the extreme case of death, they may be more of a relationship between S-100 and Parsonnet score.
We have tried in our previous work to understand mechanisms and now know a great deal more about the role of microbubbles,8 filters to trap them,7 the importance of acid-producing substrate load,10 pH management,11 and atheroembolism.6 Further improvement in the cerebral outcomes for patients will depend on more studies that explore both mechanisms and preventive strategies. We feel sufficiently confident that S-100 protein as the outcome measure would reliably indicate which of 2 or more perfusion strategies favored preservation of the brain. Depending on the anticipated effect of the measure, our data should be usable in power calculations in the design of these studies.
This study was funded by the British Heart Foundation (96/099). We are grateful to the other cardiac surgeons of the department, Mr E.E.J. Smith and Mr Andrew Murday, for allowing us to include their patients in the study, and to all the theater, ITU, and ward staff for cooperation and help.
- Received February 18, 1999.
- Revision received June 18, 1999.
- Accepted June 18, 1999.
- Copyright © 1999 by American Heart Association
Treasure T. Cerebral protection in adults. In: Yacoub M, Pepper J, eds. Annual of Cardiac Surgery. 7th ed. London, UK; Current Science; 1994:161–169.
Treasure T, Smith PLC, Newman S, Schneidau A, Joseph PH, Ell P, Harrison MJG. Impairment of cerebral function following cardiac and other major surgery. Eur J Cardiothorac Surg. 1989;3:216–221.
Shaw PJ, Bates D, Cartlidge NEF, Heaviside D, Julian DG, Shaw DA. Early neurological complications of coronary artery bypass surgery. BMJ. 1985;291:1384–1387.
Pugsley W, Treasure T, Klinger L, Newman S, Pascalis C, Harrison M. Microemboli and cerebral impairment during cardiac surgery. Vasc Surg. 1990;24:34–43.
Pugsley W, Klinger L, Paschalis C, Treasure T, Harrison M, Newman S. The impact of microemboli during cardiopulmonary bypass on neuropsychological functioning. Stroke. 1994;25:1393–1399.
Griffin S, Klinger L, Newman S, Hothersall J, McLean P, Harrison M, Sturridge M, Treasure T. The effect of substrate load and blood glucose management on cerebral dysfunction following cardiopulmonary bypass. Vasc Surg. 1992;26:656–664.
Patel RL, Turtle MRJ, Chambers DJ, Newman S, Venn GE. Hyperperfusion and cerebral dysfunction. Effect of differing acid-base management during cardiopulmonary bypass. Eur J Cardiothorac Surg. 1993;7:457–464.
Persson L, Hardemark H-G, Gustafsson J, Rundstrom G, Mendel-Hartvig I, Esscher T, Pahlman S. S-100 protein and neuron-specific enolase in cerebrospinal fluid and serum: markers of cell damage in human central nervous system. Stroke. 1987;18:911–918.
Taggart DP, Bhattacharya K, Meston N, Standing SJ, Kay JDS, Pillai R, Johnsson P, Westaby S. Serum S-100 protein concentration after cardaic surgery: a randomised trail of arterial filtration. Eur J Cardiothorac Surg.. 1997;11:645–649.
Parsonnett V, Dean D, Bernstein AD. A method of uniform stratification of risk for evaluating the results of surgery in acquired adult heart disease. Circulation. 1989;79(suppl I):I-3–I-12).