Protein S100B after cardiac surgery: An indicator of long-term anxiety?

Abstract

Objectives. The aim of the study was to assess long-term state and trait anxiety in cardiac surgical risk patients. Design. Thirty two patients with serum S100B > 0.3 µg/l 48 hours after cardiac surgery with cardiopulmonary bypass were matched according to age, gender, type, date and length of surgery with 35 operated patients without elevated S100B. They completed Spielberger's Anxiety Inventory (STAI). Results. Patients with elevated S100B reported more state anxiety and trait anxiety. S100B was an independent predictor of both state and trait anxiety when controlling for perioperative variables. Conclusions. Patients with elevated S100B reported more anxiety 3–6 years after cardiac surgery. A postoperative blood sample can identify risk patients and facilitate appropriate follow-up.

• Cardiac surgery
• S100B
• anxiety

The influence of emotional factors on cardiac surgical outcome has been recognized over the past years. Several studies have described the incidence of anxiety and depression in cardiac surgical populations and given evidence of their clinical relevance.

Patients with preoperative anxiety and/or depression are predisposed to continuing symptoms 1–3. Emotional symptoms may also debut after surgery and thus negatively influence treatment outcome 1, 4. In general it seems that anxiety and depression are unrelated to disease severity in cardiac surgical patients 2. This suggests that there are other causal mechanisms of anxiety and depression for this group, and that relieving cardiac symptoms may not relieve emotional distress. Moreover, studies have shown that depression is an independent predictor of cardiac morbidity and mortality after coronary artery bypass surgery 2, 4, 5. Adequate treatment of emotional distress is thus motivated.

Several biochemical markers have been suggested and investigated as diagnostic and prognostic tools of outcome after cardiothoracic surgery. One of these, the glial protein S100B, has been studied extensively over the past decades. It has been proposed as a prognostic marker in, for example, traumatic brain injury 6, 7, cardiac arrest 8 and stroke 9, 10. Despite convincing results, the use of S100B as a surrogate marker of cerebral injury has been equated with considerable methodological difficulties. Contamination from extracerebral sources clouds interpretation of serum S100B values 11, 12, and the once-hoped for correlation with functional outcome, i.e. neuropsychological test results, has been hard to prove after cardiac surgery 13, 14.

Nonetheless the late release of protein S100B after cardiac surgery has been associated with postoperative stroke 15, 16 and mortality 17. Late release is defined as S100B measurable after the first 24 hours after surgery, at which time contamination from extracerebral sources is presumed to be insignificant 18.

Our research group has found that the late release of S100B two days after surgery (hereafter T48) has been associated with stroke 16. In addition patients with high S100B values at T48 had a significantly higher odds ratio for mortality in a two-year follow-up period 17. The aim of the present study was to expand the mortality study 17 and compare patients with and without elevated S100B. We examined state and trait anxiety in 67 patients 3 to 6 years after cardiac surgery. Perioperative variables related to morbidity and mortality were included retrospectively to investigate their possible relationship with subjective well-being.

Materials and methods

Subjects

From a consecutive series of 1 370 patients undergoing cardiac surgery with cardiopulmonary bypass at Lund University Hospital in Malmö, Sweden between 1995 and 1998, protein S100B values were routinely sampled after termination of cardiopulmonary bypass (T0), 5 hours postoperatively (T5), after 15 hours (T15) and on the second morning after surgery, range 38–42 hours (T48). The S100B results and other perioperative data were kept in two databases, of which one was specialized on central nervous system variables and the other on general surgical variables. The latter served as the official quality registry for the department. A patient group with complete records from both sources and with complete S100B samples was described in an earlier study 17. In this defined patient group of 678 patients, 67 patients had elevated (>0.3 µg/l) S100B values at T48.

Starting with the most recent operation, the first 32 patients with elevated S100B willing and able to participate were selected and matched with 35 subjects with S100B values under < 0.3 µg/l at T48. Control subjects were matched according to age, gender, perfusion time, type and date of surgery. For each patient, between two and four possible controls were identified and randomly selected one at a time as the need arose. Fifty three patients were contacted, of which 21 declined participation. Of 57 contacted control subjects, 22 declined participation. Demographic data for participating subjects are presented in Table I.

Table I.  Descriptive statistics

	Total	S100B	Control
n	67	32	35
Age*†	73.4 (7.5)	74.3 (6.7)	72.5 (8.2)
Years since operation*	4.2 (1.1)	4.2 (1.1)	4.2 (1.1)
Cross clamp time, min*	46.8 (21.7)	48.1 (22.3)	45.6 (21.4)
Perfusion time, min*	85 (30.0)	88.8 (32.4)	81.5 (27.6)
Men/Women	54/13	26/6	28/7
CABG/valve‡/other	59/6/2	28/3/1	31/3/1
S100B T48	0.35 (0.21)	0.51 (0.22)	0.21 (0.02)

*Mean (Standard deviation).

†Age at time of interview.

‡Two patients with elevated S100B underwent combined valve surgery and CABG. Matching was performed on the basis of valve surgery.

Procedure

Approval from the local ethics committee was rendered (LU633-99, December 1999; LU53-02, March 2002). The research protocol adhered to the principles stated in the World Medical Association Helsinki Declaration. In batches of four or five, letters explaining the studies were sent to patients and controls. A week or two later, the psychologist (CDB or KA) contacted patients by telephone to obtain consent. Subjects either completed the questionnaire in the psychologist's office, at the hospital or primary care center closest to the patient's home or in their own home in cases where the questionnaire was sent by mail. Data collection was performed between 2000–2002, and patients were enrolled in one of two parallel studies.

Data for other variables found to be associated with long-term mortality in an earlier study 17 were retrieved from the department quality registry, namely emergency surgery, chronic obstructive pulmonary disease, preoperative left ventricular ejection fraction, cross-clamp time, perfusion time, postoperative use of intraaortic balloon pump, postoperative central nervous system complication, postoperative prolonged ventilator use and postoperative renal failure. Mortality as of February 2006 was confirmed through the regional computerized patient administrative system (PASIS). The variables age, gender, and elapsed time since surgery were also included for analysis.

Instruments

Patients completed Spielberger's Anxiety Inventory (STAI) 19.

Statistical analysis

Correlational analysis, independent samples t-tests and multiple regression analysis were performed with SPSS 12.0 Statistical Package. Statistical significance was accepted at p < 0.05.

Results

STAI

Patients with elevated S100B at T48 reported significantly higher levels of state and trait anxiety. Age and lapsed time from surgery were not related to anxiety. With gender as a covariate, differences in anxiety levels between the groups were accentuated. Women had significantly higher scores on trait anxiety. Men with elevated S100B had significantly higher state and trait anxiety than male controls. Mean anxiety ratings for women with elevated S100B levels were over 44, the cut-off score for clinically relevant anxiety 20. In sum, patients with elevated S100B 2 days after surgery report more state and trait anxiety, with scores close to the clinical cut-off. Anxiety scores and t-test comparisons are presented in Table II.

Table II.  State and trait anxiety: mean scores and t-test comparisons

	n	Mean	Std dev	t-value	df	p-value		n	Mean	Std dev	t-value	df	p-value
State Anxiety							Trait Anxiety
S100B > 0.3	29	40.88	10.22	3.31	62	0.002	S100B > 0.3	25	42.97	12.31	2.66	55	0.010
Controls	35	32.52	9.94				Controls	32	35.19	9.81
Total	64	36.31	10.83				Total	57	38.61	11.55
Women	12	41.20	11.77			ns	Women	12	45.75	12.78	2.53	55	0.014
Men	52	35.18	10.4				Men	45	36.7	10.55
Women S100B > 0.3	5	48	9.67			ns	Women S100B > 0.3	5	53	11.43			ns
Female controls	7	36.36	11.22				Female controls	7	40.43	11.49
Men S100B > 0.3	24	39.4	9.88	2.9	50	0.006	Men S100B > 0.3	20	40.42	11.38	2.2	43	0.033
Male controls	28	31.57	9.57				Male controls	25	33.73	9

Somatic variables

In order to investigate the possibility that patients’ reported anxiety in fact could be explained by other perioperative variables or that S100B was a covariate of morbidity, variables previously shown to correlate with mortality in univariate analysis 17 were tested for association with anxiety. State anxiety was significantly correlated with chronic obstructive pulmonary disease, and trait anxiety in turn with chronic obstructive pulmonary disease and gender. Gender was included as a predictor for both state and trait anxiety because the women in the sample, and the women with elevated S100B in particular, exhibited clinical anxiety levels. In a stepwise multiple regression analysis with gender and chronic obstructive pulmonary disease as predictors of anxiety, chronic obstructive pulmonary disease and gender were predictors of long-term state (R² change = 0.18, p = 0.004) and trait anxiety (R² change = 0.27, p < 0.000). S100B was thereafter added to the models and contributed significantly to both predictions (state anxiety R² change = 0.09, p = 0.009; trait anxiety R² change = 0.07, p = 0.024). As of February 2006 nine participants with elevated S100B were deceased, as was the case with five controls. The difference did not reach statistical significance.

Discussion

Patients with elevated S100B had significantly higher levels of both state and trait anxiety 3 to 6 years after surgery. The magnitude of differences in scores reflects a considerable effect size (>1SD), and the mean scores approached the cut-off for clinically significant anxiety on both state and trait measures. Earlier research 21 has demonstrated that women often report higher levels of emotional distress after cardiac surgery, as was the case in our sample.

In order to investigate the possibility that patients’ reported anxiety in fact could be due to other factors during surgery or that S100B was a covariate of such possible morbidity, available perioperative data were collected for analysis. Johnsson's earlier results 17 indicated that S100B was significantly related to a number of perioperative morbidity variables, but here again the objective was to determine the unique contribution of S100B after controlling for these covariates. Chronic obstructive pulmonary disease and gender (for trait anxiety only) were predictors of anxiety in multiple regression analysis, but S100B gave a unique contribution after controlling for these variables. Accordingly the role of S100B as a covariate of anxiety cannot be determined within the bounds of this study, but it is an interesting possibility.

In our model age and time elapsed since surgery could not predict anxiety. The retrospective analysis of the relationship between comorbidity on the one hand and S100B and long-term anxiety on the other was limited to the available pre- and postoperative data. Obtaining data on objective current morbidity and, where applicable, cause of mortality was not feasible within the frame of this study.

An obvious limitation of the study was the limited sample. Successful matching procedures for patients and control subjects were employed in order to minimize variability due to age and gender as well as type, date and length of surgery. Some patients found the wording of Spielberger anxiety inventory items difficult to comprehend, and a loss of data occurred as a result.

The results obtained are nonetheless remarkable: a blood test taken 2 days after surgery can predict emotional state 3 to 6 years after surgery. Speculating about the cause of this elevated anxiety is important. Anxiety has been found to be a predictor of cardiac events 2, is elevated in various patient groups 22 and is a covariate of personality traits, i.e. neuroticism 22. Measures of personality were not assessed in this study, wherefore we cannot draw any conclusions about the personality makeup of our patients. However, studies have shown that cardiac patients who perceive their health as poor also experience more emotional distress 20, 23–25. S100B has recently been found to be a biochemical marker of stress in experimental studies 26. Applying these experimental findings to our clinical data, one can speculate whether patients with elevated S100B experienced more stress in the perioperative period.

In addition to increasing the risk of late mortality17, elevated S100B serum levels two days after surgery are a predictor of emotional well-being several years after surgery. The cause of the elevated S100B levels cannot be ascertained from the present study, but the potential of early postoperative identification of patients risking an unfavorable long-term outcome is promising for the facilitation of appropriate postoperative care, follow-up and rehabilitation.

This study was supported by grants from Anna and Edwin Berger's Foundation and Siegvald's Fund, Lund University. This paper was presented at the 3^rd International Conference on Biochemical Markers for Brain Damage (BMBD) in Lund, Sweden September 25–27, 2003.