| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2003;75:162-168
© 2003 The Society of Thoracic Surgeons
a Center of Heart and Lung Disease, Institute of Laboratory Medicine, Lund University Hospital, Lund, Sweden
b Department of Psychology, Lund University, Lund, Sweden
Accepted for publication July 22, 2002.
* Address reprint requests to Dr Johnsson, Department of Coronary Artery Disease, Center of Heart and Lung Disease, Lund University Hospital, SE 221 85, Lund, Sweden
e-mail: pelle.johnsson{at}skane.se
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsResultsCommentReferences |
|---|
METHODS: Prospectively collected clinical variables, including S100B, in 767 patients who survived more than 30 days after a cardiac operation, were analyzed with actuarial survival analysis and 678 patients were analyzed with Cox multiple regression analysis.
RESULTS: Forty-nine patients (6.4%) were dead at follow-up (range, 18 to 42 months); 11.5% (88 of 767 patients) had elevated S100B 2 days after operation (range, 38 to 42 hours). The probability for death at follow-up was 0.239 if the S100B level was more than 0.3 µg/L, and 0.041 if it was less than 0.3 µg/L. The clinical variables independently associated with mortality were preoperative renal failure, preoperative low left ventricular ejection fraction, emergency operation, severe postoperative central nervous system complication, and elevated S100B values, which turned out to be the most powerful predictor.
CONCLUSIONS: Even slightly elevated S100B values in blood 2 days after cardiac operation imply a bad prognosis for outcome, and especially so in combination with any central nervous system complication.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
Increased blood concentrations of the glial protein S100B, measured after operations using cardiopulmonary bypass (CPB) operations, are associated with cerebral complications [5]. However, this protein is not only restricted to brain tissue, but can also be found in peripheral nerves, in extracts of fat, bone marrow, and organs of mesenchymal origin [6, 7]. Furthermore, S100B accumulates in the mediastinum during and after operation [8], and therefore, substantial amounts of S100B are present in blood returned to the bypass circuit from the cardiotomy suckers, as well as in shed mediastinal blood collected after the procedure. This contamination reduces the diagnostic value of S100B measurements during the first hours after operation or, if shed blood is retransfused to the patient, a few hours after the last transfusion. Attempts to correlate increased levels of S100B early after operation with neurologic complications have shown weak or inconsistent results and negligible associations [8–11]. However, the biological half-life of S100B is only 25 minutes. The rapid elimination from the bloodstream thus makes it very unlikely that any degree of contamination would affect S100B measurements 24 hours or more after operation [12]. The S100B concentrations measured after 24 hours or beyond show considerable and consistent correlations with stroke or other more robust signs of neurologic impairment [5, 13].
A number of patients have increased levels of S100B after operation with CPB, even without evident signs of neurologic impairment. These patients could have subclinical brain injury. From studies with computed tomographic scans or magnetic resonance imaging it is well known that morphologic signs of brain injury may be present in what were considered normal patients [14]. We measured, as part of the clinical routine, S100B levels in patients undergoing operation with CPB during 1996 and 1997. The present study aimed to investigate the long-term outcome in these patients and to explore any possible associations between the postoperative S100B levels and outcome.
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
Patient variables were prospectively collected and stored in two databases, of which one contained general variables and was based on a Swedish translation of the Society of Thoracic Surgeons patient record form. This database was the official quality registry for the Lund and Malmoe departments and contained 1,074 of the 1,075 patients who underwent operation during the study period. The other was more specialized on central nervous system (CNS) variables, and was set up as a research project but continued after the research sampling period ended. Eight hundred ninety-nine patients were represented in both databases. All patients were followed up for survival as of July 1, 1999, resulting in a follow-up interval of 18 to 42 months. Follow-up was 100% complete and was performed through the Population and Welfare Statistics at Statistics Sweden, Statistiska Centralbyr
n, Stockholm, Sweden.
One patient was operated on twice, and because only the second operation was included in the study, the number of patients was reduced to 898. In 6 patients the operation was performed as an off-pump procedure and these patients were also excluded from analysis, leaving 892. All patients who died within 30 days (29 of 899; 3.2%) were excluded, as the aim was to evaluate intermediate- or long-term outcome. Sixteen of these patients died within 2 days and 13 during the interval of 2 to 30 days. In 8 of 13 patients (61.5%) the S100B level was elevated at 48 hours. As stated before, none of these patients was included in the study. The 863 remaining patients formed the primary investigation group in which a number of variables were identified for univariate correlation with death at follow-up. Due to missing S100B values at the designated sampling interval, another 96 patients had to be excluded, leaving 767. Finally for the Cox regression analysis, as no missing values can be accepted, another 89 patients with incomplete data had to be excluded. Therefore, 678 patients remained for the final analysis.
Blood for S100B analysis was sampled in patients according to the clinical routine at termination of CPB, 5 and 15 hours thereafter, and on the second morning after operation S100BT48 (38 to 42 hours after operation). All samples were centrifuged, cooled, and stored within 4 hours, and analyzed on a continuous basis upon arrival to the laboratory, generally during the next few days. The analyses were performed with a radio immunoassay (Sangtec100, AB Sangtec Medical, Bromma, Sweden), which has been described in detail earlier [8]. The lower detection limit of this analysis was 0.2 µg/L. The S100B values in this study have been entered as a dichotomous variable with a cutoff value of 0.3 µg/L to ensure a definite pathologic level.
Statistics
Values are given as mean ± standard deviation. Odds ratios are presented as OR with 95% confidence limits (CI) in parentheses. Statistical analysis was performed with SPSS (SPSS Inc, Chicago, IL) statistical package for PC. The graphs were produced with Statistica (StatSoft, Tulsa, OK) for PC. The significance level is p less than 0.05 if not otherwise stated.
A number of pre-, peri-, and postoperative clinical variables considered relevant for death/survival were chosen from the databases and are listed in Table 1. These variables were first tested for their relation to mortality in univariate analysis in the primary investigation group (n = 863) and, if significant (p < 0.05) or considered especially relevant (chronic obstructive pulmonary disease and gender), entered into a Cox regression analysis. Probability for entry and removal into the model was set to 0.05 and 0.10, respectively.
|
|
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
In 33 of 154 (21.4%) of the patients who had another operation than sole CABG (valve or combined with CABG or other operation) the S100BT48 more than 0.3 µg/L was 0.7 ± 0.79 µg/L. In the group who only had CABG, 65 of 613 (10.6%) of the patients had S100BT48 more than 0.3 µg/L, with a mean value of 0.7 ± 1.11 µg/L. It was significantly more common with elevated S100B if the operation was other than sole CABG (OR 2.3; 95% CI 1.6 to 3.2). There was, however, no difference between the two groups with respect to the mean increase of S100B.
Mortality was increased in the group with other operation compared to CABG alone. At follow-up, 20 of 154 (13%) of the patients compared to 29 of 613 (4.7%) had died (OR 3; 95% CI 2.1 to 4.4).
Actuarial analysis
To illustrate the longitudinal outcome (death/survival), a Kaplan Meier curve was drawn using all of the 767 patients with complete S100B values (Fig 1).
It shows the increasing divergence between the two patient groups denoted by the variable S100BT48 more than 0.3 µg/L (Wald = 43.22; p < 0.001).
|
|
2 increased 13.372; p < 0.001). In a second analysis the material was divided into two groups: those who died during the interval 30 to 90 days and those who died later than 90 days after the operation. Nine of 678 patients died during the first interval and 36 of 678 died during the second interval. The same patient variables were once again entered in Cox regression analysis and the results are shown in Tables 3 and 4 . The variables that significantly reflected increased mortality risk in the early period were preoperative left ventricular ejection fraction, prolonged need for ventilator assistance, and increased S100BT48. Significant predictive variables for death in the later period were preoperative chronic obstructive pulmonary disease and preoperative renal failure, emergency operation, cross-clamp time, severe CNS complication, prolonged need for ventilator assistance, and increased S100BT48. The prediction and power of S100BT48 was stronger in the early interval than in the later time. Six of 9 patients, who were dead at the interval 30 to 90 days, had elevated S100BT48 levels, but only 2 of 6 were discharged from hospital with a clinically identified CNS-related complication.
|
|
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
If patients sustained cerebral complications with elevated S100B values at 2 days after operation, the risk of an early death was more than 20 times higher than if cerebral outcome was uneventful and S100B was below the cutoff level of 0.3 µg/L. The cause of death in the individual patient is principally and unfortunately unknown, as postmortem examinations were not performed. Cause of death in Sweden is generally declared on criteria of likelihood if the patient has a known medical record. Therefore, patients who have undergone heart procedures are likely to have "cardiac death" as cause of death, where no other cause is obvious. We were unable to investigate whether S100B levels correlated with brain-related pathology, evident first after the original hospital admission.
This study was designed in retrospect on the basis of prospective information sampled at the time of hospitalization. No retrospective search for missing data or reconstructions from stored patient records was undertaken. We had to choose between a smaller cohort with more variables, or a larger cohort and fewer variables. We used the primary investigation group of 863 patients for the univariate analysis to identify variables to enter in the Cox regression analysis. The missing S100BT48 samples in 96 patients were probably either never sampled, misplaced, or not analyzed. Despite our efforts to register and keep records of database forms and samples as part of the clinical routine at the time of operation, a substantial number were missing at analysis and resulted in patient exclusions. For instance, left ventricular ejection fraction was calculated and reported for the majority of patients but missing left ventricular ejection fraction values account for 63 patients of the 767 patients in which we had complete S100BT48 values. However, the missing values and lacking S100B samples occurred at random. The remaining patients available for analysis still form a large and very significant cohort. The rationale for exclusion of the patients who died during the first 30 days after operation was that S100B samples at 48 hours would be missing in more than half of the patients (16 of 29) who died within the first 2 days. In the 13 other patients S100B level was elevated in 8 and would have contributed to further increase the association between S100B and mortality. Our aim was to investigate the long-term effect, and for this reason it was also reasonable to exclude these patients in whom death was more or less caused by intraoperative or permanently reduced cardiac output after operation.
In this study we aimed to compare the association between S100B and death after operation. This was performed in multiple regression analysis (also) including other clinically relevant variables. One could raise the questions why such obvious risk factors as age did not remain significant in regression analysis, or why the predictive value of variables changed between the two compared time intervals. It should be realized that by excluding the mortality that occurred during day 0 to 30, we have lost patients in which, for instance, the variable age may have had more importance. In addition there were other predictors in the Cox regression that correlated with death, sharing the same variance and therefore, suppressing age as a predictor. It is also quite obvious that with time, other variables than those we studied may play a significant part, and the influence of typical operative variables ought to decrease.
The majority of patients with increased S100B levels in this study left the hospital without any clinical signs or suspicion of neurologic impairment. The association between S100B and brain injury has been confirmed in studies of both global and focal brain injury in stroke, trauma, and cardiac operation [5, 13, 15–17]. The S100B release from injured brain cells is considered to leak through a permanently or reversibly damaged blood–brain barrier. As mentioned earlier, there are methodological difficulties involved with blood samples collected in conjunction with CPB or shortly after autotransfusion of shed blood. Because our sampling took place more than 38 hours after operation, it is unlikely that contamination was the cause of the elevated concentrations of S100B. However, Anderson and co-workers [18] have suggested that mediastinal S100B, at least theoretically and in small amounts, may be absorbed through the lymphatic system after removal of the chest drainage tubes, and is therefore not necessarily of cerebral origin. This general uptake has not yet been proven. Should it have occurred to a significant degree in the present study, one may argue that it should have caused S100B elevations in more than 11% of patients. On the other hand, Anderson and co-workers have used a different method for the analysis of S100B with an improved sensitivity at low range values. Furthermore, the concentration gradient between the pericardium and serum is unknown. Because the biological half-life of S100B in serum is as short as 25 minutes [12], it would still take a continuous and very substantial transport to result in measurable serum levels.
In this study we have handled the S100B value as a dichotomous variable with a defined cutoff level. It could be argued that if used as a continuous variable more information perhaps would have emerged. However, as the range close to and below 0.2 µg/L is uncertain or not defined at all, this would have led to other methodological difficulties, which were avoided with a dichotomous variable. In addition, we have not made any attempt to correlate the actual level of S100B with outcome. Protein leakage from the brain should be related to the volume of damaged or involved tissue, and it is well known that patients with a large stroke seem to fare worse than patients with small strokes, a comparison that from this point of view would seem logical. However, the nature of S100B and brain involvement in patients undergoing CPB is still poorly understood and therefore it is controversial to make inferences to such a defined lesion as stroke. To further emphasize the complexity of interpretation it is not only a question of how large a CNS lesion may be, but perhaps even more important is the exact location of the lesion. S100B does not differentiate between different anatomic structures of the brain.
It has been inferred from in vitro studies that high concentrations of S100B may induce neuronal apoptosis [19]. However, it seems unlikely that circulating S100B in the postoperative course would impose damage to the brain.
The elimination of S100B would be decreased by renal dysfunction and could have resulted in a higher S100B concentration in patients with preoperative and postoperative renal failure. The higher mortality risk in these patients may then "positively" affect the association between high S100B and mortality. Similarly, the patients with only CABG had a lower mortality and less often increased S100B values. However, it is important to remember that each of the variables with a predictive value for death independently contributed to the model when it was included stepwise in the Cox regression analysis. Any co-association between variables entered has the contrary effect of reducing the possibility of gaining enough power to remain significant in the model. S100BT48 turned out to be the variable with the highest predictive power.
A patient with an elevated S100B concentration, even without clinical signs of a neurologic complication, had an increased risk for early death. This risk could have been explained by subclinical brain injury, which perhaps could have been diagnosed by magnetic resonance imaging or by prospective neurologic or neuropsychological examinations, had such been undertaken. This study, therefore, does not prove that S100B is a surrogate marker for subclinical brain injury, but it clearly shows that there is a definite association between high postoperative S100B values and a shortened life expectancy. This association is moreover a strong relationship if the patient in fact has brain injury.
S100B may be a covariate or marker for other processes not known or investigated in this study. Brooker and colleagues [20] showed that the number of small capillary arteriolar densities was increased in the brain if cardiotomy suction blood was reintroduced to the perfusion circuit. Therefore, in speculation, if S100B is present in blood from the cardiotomy suction and one of its sources is fat, S100B and small capillary arteriolar densities may be related. If so, S100B would measure the degree of embolization or the degree of damage, depending on the circumstances. However, small capillary arteriolar densities are not suggested as a mechanism of brain injury in global anoxia as in cases of cardiac arrest or in macroembolic stroke. S100B, however, has been shown in these patients to be a valuable prognostic marker [15, 21].
One may question the clinical value of a "prediction" that becomes known to us almost 2 days after the event, without any possibility of undoing the procedure or avoiding the harmful effects the operation itself may have caused. If a preoperative variable serves as a warning and sometimes makes us more aware of particular problems, or even makes us decide against operation in a particular patient, you may say that it is for the benefit of that patient. As an isolated laboratory result after an otherwise successful procedure, it could be argued an unnecessary test that should be avoided for ethical or humanitarian reasons. On the contrary, however, we would like to emphasize the importance of resolving the unsolved issues concerning the mechanisms that cause the elevation of S100B in these patients, and why it is related to their outcome.
In conclusion, high S100B level, 2 days after cardiac operations, carries a very bad prognosis for long-term outcome, and especially so in combination with any kind of CNS complication.
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  N. Nagdyman, P. Ewert, K. Schmitt, M. Hubler, B. Stiller, C. Muller, and F. Berger Protein S-100 is present in extracerebral fluids before and after cardiac surgery in children Ann Clin Biochem, July 1, 2008; 45(4): 409 - 412. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Carrier, A. Denault, J. Lavoie, and L. P. Perrault Randomized controlled trial of pericardial blood processing with a cell-saving device on neurologic markers in elderly patients undergoing coronary artery bypass graft surgery. Ann. Thorac. Surg., July 1, 2006; 82(1): 51 - 55. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. C. Lases, M. A. Schepens, F. J. Haas, L. P. Aarts, H. T. ter Beek, E. P. van Dongen, H. P. Siegers, I. van der Tweel, and E. H. Boezeman Clinical prospective study of biochemical markers and evoked potentials for identifying adverse neurological outcome after thoracic and thoracoabdominal aortic aneurysm surgery Br. J. Anaesth., November 1, 2005; 95(5): 651 - 661. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K.-J. Wang, H.-H. Wu, S.-Y. Fang, Y.-R. Yang, and A. C.-C. Tseng Serum S-100 {beta} Protein During Coronary Artery Bypass Graft Surgery With or Without Cardiopulmonary Bypass Ann. Thorac. Surg., October 1, 2005; 80(4): 1371 - 1374. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Baufreton, P. Allain, A. Chevailler, F. Etcharry-Bouyx, J. J. Corbeau, D. Legall, and J. L. de Brux Brain Injury and Neuropsychological Outcome After Coronary Artery Surgery Are Affected by Complement Activation Ann. Thorac. Surg., May 1, 2005; 79(5): 1597 - 1605. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. I. Hall Serum S-100{beta} protein and postoperative neurological dysfunction - ready for prime time?/La proteine serique S-100{beta} et la dysfonction neurologique postoperatoire - correlation prete pour une large diffusion? Can J Anesth, August 1, 2004; 51(7): 645 - 648. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
0.001. Note that the hospital mortality (patients who died within 30 days from the operation) was excluded from the study.