| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Medical, Clinical, and Experimental Psychology, University of Maastricht, Maastricht (M.L.P); Department of Health Psychology, (G.L.R.G.), Department of Medical Physiology and Sports Medicine (R.E.B., M.V.), and Wilhelmina Children’s Hospital (C.J.H.), University of Utrecht, Utrecht; Department of Clinical and Health Psychology, University of Leiden (J.F.B.); and Department of Chemical Endocrinology, Academic Hospital Nijmegen, St. Radboud, Nijmegen (F.C.G.J., L.M.J.W.S.), The Netherlands.
Address reprint requests to: Madelon Peters, Department of Clinical, Medical, and Experimental Psychology, University of Maastricht, P.O. Box 616, 6200 MD Maastricht, The Netherlands. Email: Madelon.Peters{at}DEP.unimaas.nl
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
METHODS: In a 2 x 2 design, effort and uncontrollability were manipulated independently of each other. Subjects participated in one of four experimental conditions, and their endocrine, immune, and sympathetic nervous system responses to the task were assessed.
RESULTS: Effort had a stimulating effect on enumerative immunological parameters (CD8+ and CD16+ cells) and on natural killer cell activity. The effect occurred immediately after the stressor and was transient. Regression models indicated that this effort effect may have been mediated by activation of the sympathetic nervous system. Uncontrollability influenced in vitro production of the cytokine interleukin-6, leading to decreased production 15 and 30 minutes after the stressor. Uncontrollability also led to an increased level of cortisol, but no evidence was found that the decrease in cytokine production was mediated by cortisol release.
CONCLUSION: The results suggest that two major stressor characteristics, effort and uncontrollability, may have differential effects on the immune system.
Key Words: effort • uncontrollability • stress • immune function • sympathetic-adrenal-medullary system • hypothalamic-pituitary-adrenal-cortical system
Abbreviations: DBP = diastolic blood pressure; DEX = dexamethasone; ELISA = enzyme-linked immunosorbent assay; HPAC =hypothalamic-pituitary-adrenal-cortical; IL = interleukin; LPS = lipopolysaccharide; MANOVA = multiple analysis ofvariance; NKCA = natural killer cell activity; SBP = systolicblood pressure.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Several researchers have tried to extend these findings to human subjects (8, 10, 17, 18), but the results of these studies have been inconclusive. In line with the prediction from animal studies, Sieber et al. (10) found decreased NKCA after uncontrollable stress but not after controllable stress, and Brosschot et al. (18) reported a decrease in the number of T-helper cells in blood after a stressor only if it was perceived as uncontrollable. However, Gomez et al. (17) found no effect of uncontrollability on various immune parameters (number of several types of immune cells, immunoglobulin level, and NKCA), and Weisse et al. (8) reported the unexpected finding of decreased lymphocyte proliferation after controllable but not uncontrollable stress.
One explanation for the contradictory results may be that in these experiments, the effects of uncontrollability could not be distinguished from the effects of differences in effort, which were also introduced by the manipulation. Effort and uncontrollability may both affect the immune system but by different mechanisms. Influential psychobiological models of stress state that situations requiring effort activate the sympathetic-adrenal-medullary system, whereas uncontrollable situations specifically activate the HPAC system (19–24). The fact that sympathetic nervous system activation influences the immune system is well established, and it has been demonstrated that the effects of most experimental stressors on the immune system depend largely on ß-adrenergic mechanisms (25, 26). The HPAC system, including the adrenal glucocorticoid cortisol, also has prominent immunomodulatory effects (27). However, because of the time course of the cortisol rise, cortisol effects have often been difficult to establish in short-term human stress studies.
The purpose of the present experiment was to identify the specific contributions of uncontrollability and mental effort to stress-induced immune changes. A 2 x 2 design was used in which subjects were asked to perform either a high- or low-effort task and in which they could or could not acquire control over the intensity of aversive noise. In a previous article (28), we presented the cardiovascular and endocrine outcomes of the study. That article confirmed the prediction that effort affects the sympathetic nervous system, whereas uncontrollability affects the HPAC axis. The high-effort condition led to more pronounced increases in heart rate, blood pressure, and norepinephrine levels than the low-effort condition. The uncontrollable condition led to an increased level of cortisol compared with the controllable condition. However, uncontrollability apparently also had sympathetic effects: Higher blood pressure and plasma norepinephrine levels were found in subjects assigned to uncontrollable conditions.
In the present article, we report the immunological findings of the study. A large number of immunological parameters were assessed, including both enumerative parameters (number of CD3+ (total T), CD4+ (T-helper), CD8+ (T-cytotoxic/suppressor), and CD16+ (NK) cells) and functional parameters (NKCA, lymphocyte proliferation, and mitogen-induced cytokine production). We predicted that both mental effort and uncontrollability would affect the immune system but that the nature of these influences would differ. Mental effort is hypothesized to influence the immune system by sympathetic activation. Because sympathetic (ß-adrenergic) activation has previously been shown to lead to increased CD8+ cells, CD16+ cells, and NKCA and to decreased CD4+ cells and lymphocyte proliferation (1, 7, 25, 29–34), an effect on these parameters was specifically predicted. Moreover, changes in these parameters were predicted to be associated with the heart rate and blood pressure changes induced by the task.
We also hypothesized that uncontrollability would influence the immune system by HPAC axis activation, especially through the release of cortisol by the adrenal cortex. Cortisol has been shown to lead to suppression of lymphocyte proliferation and decreased production of macrophage- and lymphocyte-derived cytokines (35–40). Therefore, we predicted that uncontrollability would especially affect lymphocyte proliferation and the production of IL-4 and IL-6. Moreover, these changes were predicted to be associated with the cortisol response. Also, lymphocyte proliferation in the presence of different concentrations of the synthetic glucocorticoid DEX was determined. DEX mimics the effect of endogenous cortisol and suppresses the proliferative response in a dose-dependent manner (34). We tested whether uncontrollability affected the dose-response curve of the DEX-induced suppression.
Thus, the present study was aimed at unraveling the influence of different aspects of stress on the immune system. The demand characteristics of stressors may lead to sympathetically induced rapid but transient immune changes, which, at least in the short term, have little consequences for health. Uncontrollability, on the other hand, may be more detrimental for health, because cortisol has a generally suppressive effect on functional aspects of the immune system that may also be more prolonged.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Because of occlusion of the catheter, data from 15 subjects were lost for immunological determination. Two subjects were removed because of failure of the manipulation (see Results). For IL-4 and IL-6, more subjects were lost because of technical problems (eg, contamination of assays), leaving 78 and 49 subjects, respectively, for the analyses.
Procedure
A between-subject design was used, with 24 subjects participating in each of four experimental conditions. Subjects had been instructed to eat a light breakfast on the morning of the experiment; caffeine-containing beverages were not allowed. Subjects suffering from an infectious illness within 2 weeks before the experiment were rescheduled. Subjects were tested individually between 9:00 and 11:30 AM. The subject sat in a temperature-controlled, dimly lit room at a height-adaptable table with a PC screen in front of him. After the informed consent form was read and signed, electrodes for electrocardiographic measurement and a finger cuff for blood pressure recording were attached. Next, a cannula was inserted into a forearm vein of the left arm to allow for minimal disturbance during repeated blood draws. For the next 15 minutes, subjects filled out psychological questionnaires. Subjects were then allowed to relax while watching a documentary video film for 10 minutes. At the end of this relaxation period, the first blood sample (for endocrine and immunological assays) and first saliva sample (for cortisol) were taken (T0). Next, instructions to the test were presented on the PC screen, and subjects were allowed to practice the task. This took about 8 to 10 minutes. The test itself lasted for 15 minutes. Immediately after completion of the task, the second blood and saliva samples (T1) were taken, and subjects filled out a questionnaire on subjective appraisal of the task. Viewing of the video film was resumed, and exactly 15 minutes after completion of the stress task, the third blood and saliva samples were taken (T15). After the subjects had filled out an additional psychological questionnaire, viewing of the video film was resumed for the last time. Thirty minutes after task completion, the fourth and final blood and saliva samples were taken (T30). The electrodes, blood pressure cuff, and cannula were removed, and subjects were debriefed.
Stress Task
The stress task has been described in detail elsewhere (28). In short, it consisted of a mental task to be performed under continuous noise stimulation. Subjects were given one of two tasks requiring either high or low effort. Half of the subjects assigned to each task were given control of the intensity of noise stimulation by their performance, and the other half were not. This yielded four experimental conditions: high effort/controllable, high effort/uncontrollable, low effort/controllable; and low effort/uncontrollable.
In the high-effort conditions, subjects performed mental arithmetic. In 16 successive trials (each trial lasting 50–70 seconds), a series of multiple choice sums were presented by the computer one by one in a self-paced manner. In each trial, subjects had to give the correct answer to a specified number of sums. This number varied according to the condition (controllable or uncontrollable) and the capability of each individual subject (as initially determined in the practice trial and refined by performance in subsequent trials). In the high-effort/controllable condition, the number of sums that had to be solved correctly was set at a level enabling successful performance for a particular subject. In the high-effort/uncontrollable condition, the number of sums was set at a level normally leading to failure for a particular subject. Failure in all trials of the uncontrollable condition was guaranteed by occasionally providing only incorrect multiple choice alternatives.
In the low-effort condition, subjects were presented 16 trials (50–70 seconds) in which they had to find the right key to press, of four alternatives, to stop the noise. The screen changed color every minute, and subjects were told that there was a system in the key presses that they had to discover. During each trial, only one key press was allowed, and only when the screen turned green. In the low-effort/controllable condition, the correct key to press was always the same, and after a few attempts, this was clear to all subjects. In the low-effort/uncontrollable condition, subjects received feedback on 14 trials that they had chosen the wrong key regardless of which key they had actually chosen.
In all conditions, industrial noise was delivered to subjects during task performance via headphones at three different intensities (average volume, 73, 81, and 90 dB; peak volume, maximum of 2 dB higher). Successful performance allowed a subject to choose the intensity of the noise to be presented on the next trial. However, as described above, success and failure were manipulated, and only subjects in the controllable condition could actually choose noise intensity on most trials (with the restriction that the two louder intensities had to be chosen on three occasions each). To match the amount of aversive stimulation between conditions, a subject in the uncontrollable condition was yoked to the previous subject in the controllable condition with respect to noise duration and order of intensities.
Measurements
Manipulation check.
To determine whether our manipulation had the desired effect on subjective perceptions of effort and uncontrollability, subjects filled out a checklist specifically constructed for the study (28). The checklist contained nine items on perceived uncontrollability (eg, I felt helpless during the experiment; I felt I could gain control over the noise) and five items on perceived effort (eg, I had to invest a great deal of effort in finding the right answers; I felt that the task as a whole required a lot of effort). Items were rated on a seven-point scale, with the two extremes labeled as "completely disagree" and "completely agree." Subscales were confirmed by factor analysis. Reliability analyses yielded a Cronbach’s 
value of 0.87 for the effort subscale and 0.54 for the control subscale.
Cardiovascular measurements.
The electrocardiogram was recorded by means of three electrodes, which were attached at the sternum and lower right and left ribs. A trigger signal was generated at the peak of the R wave, and interbeat interval times were stored on the computer in milliseconds. SBP, DBP, and mean arterial pressure were measured continuously using the Finapres method (41).
Endocrine measurements.
Blood samples (5 ml) for catecholamine analysis were collected in precooled tubes containing 0.25 mol/liter ethyleneglycol-bis-(ß-aminoethylether)-N,N,N',N'-tetraacetic acid and 0.2 mol/L glutathione in distilled water (pH 7.4). Blood samples were placed on melting ice, and plasma was separated by refrigerated centrifugation (10 minutes at 1500g) and immediately frozen at -20°C. Plasma samples were analyzed for concentrations of norepinephrine and epinephrine using high-performance liquid chromatography (42).
Cortisol was measured in saliva. Salivary cortisol is considered to reflect the concentration of unbound cortisol in plasma and has been found to be independent of salivary flow rate (43). Saliva samples were collected by using citric acid–containing Salivettes (Sarstedt, Etten-leur, The Netherlands). Cortisol was measured by radioimmunoassay after extraction and paper chromatography (44). The sensitivity of the radioimmunoassay was 0.03 nmol/liter.
Immunological determinations.
Blood samples for immunological determinations were collected in heparinized tubes. The total number of lymphocytes per sample was determined by standard means. Subset analyses for T cells, T-helper cells, T-suppressor/cytotoxic cells, monocytes, and NK cells were performed in whole blood using Simultest sets (Becton-Dickinson, Woerden, The Netherlands) containing monoclonal antibodies against CD3, CD4, CD8, CD14, and CD16/56 according to the manufacturer’s protocol. A flow cytometer (FACScan, Becton-Dickinson) was used for analysis. The percentage of positive lymphocytes was determined and used to calculate absolute numbers of specific subsets (percentage subset x total number of lymphocytes).
To determine the proliferative responses of lymphocytes, heparinized blood was diluted 10 times with RPMI-1640 (Gibco, Grand Island, NY) supplemented with antibiotics, and 100 µl of the diluted blood was incubated with 50 µl of phytohemagglutinin (final concentration, 25 µg/ml). Background proliferation was determined by incubating cells in medium only. Quadruplicate cultures were performed. In addition, proliferative responses in the presence of different concentrations (2 x 10-9, 5 x 10-9, 1 x 10-8, 2 x 10-8, 5 x 10-8, 1 x 10-7, 2 x 10-7, 5 x 10-7, and 1 x 10-6 M) of the synthetic glucocorticosteroid DEX were determined. Culturing took place for 3 days at 37°C, after which 3H-thymidine was added. After an additional 24 hours of culturing, cells were harvested. Radioactivity (cpm) was determined with a liquid scintillation counter.
NKCA was also determined in whole blood (100 µl of undiluted, 1:2 diluted, and 1:4 diluted blood). Plates were incubated for 3.5 hours at 37°C, with 100 µl of 51Cr-labeled K562 cells (2.5 x 106/ml). After centrifugation, 100 µl of the supernatant was counted in a gamma counter. Spontaneous lysis and total lysis were determined by 51Cr release in wells containing medium (RPMI-1640 supplemented with 5% fetal calf serum) or 1% Triton X-100, respectively. Specific lysis was determined by the following formula: % Lysis = (cpm of specific sample - cpm of spontaneous lysis)/(cpm of total lysis - cpm of spontaneous lysis). The capacity to produce cytokines was also determined in whole blood. For IL-6 production, blood was diluted 10 times with medium (RPMI-1640), and 150 µl of the diluted blood was incubated with 50 µl of LPS (final concentrations, 10, 1, 0.1, 0.01, and 0.001 ng/ml) or with no LPS added. After 18 hours of incubation at 37°C, supernatant was harvested, and the level of IL-6 was determined by ELISA (Pelikine, CLB, Amsterdam, The Netherlands).
To determine IL-4 production, whole-blood samples were diluted with an equal amount of medium. Phagocytes were removed by the use of ferrocarbonyl and magnetic extraction. One hundred microliters of the phagocyte-depleted sample was plated, and 50 µl of -CD28 (1:2500), 50 µl of -CD21 (1:4500), and 50 µl of -CD22 (1:4500) were added. Culturing took place at 37°C for 72 hours, in triplicate. Supernatant was harvested, and the level of IL-4 was determined by ELISA.
Statistical Analyses
For each immunological parameter, four measurements at various time points (T0, T1, T15, and T30) were available. All parameters were subjected to repeated-measures analyses (SPSS 7.5, procedure GLM, method repeated) with the within-subject factor of time (four levels) and the between-subject factors of control (two levels) and effort (two levels). Post hoc contrasts for time were specified, using T0 as the reference category and contrasting the three other time points to this reference (method: simple). A significant time effect indicates changes in time in immunological parameters regardless of condition, whereas the time-by-effort and time-by-control effects indicate a differential change in immunological parameters in the high- vs. low-effort or high- vs. low-control conditions. The time contrasts clarify at which moment in time the (differential) task effect becomes apparent (immediately after the task or 15 or 30 minutes later, all compared with baseline).
Multivariate testing was performed in which the percentage of lymphocyte subsets CD3+, CD4+, CD8+, CD14+, and CD16+, lymphocyte proliferation, and NKCA were included in one single analysis. This analysis was repeated using the calculated absolute number of subsets instead of the percentage of subsets. Significant multivariate effects were followed up by univariate tests of individual immunological parameters.
Separate analyses were performed for IL-4 and IL-6. The cytokine assays had a substantial number of missing values, and including them in the same analysis would have compromised group size for the analyses of other parameters.
Degree of suppression of lymphocyte proliferation by DEX was analyzed using a repeated-measures MANOVA with the additional within-subject factor of dose (nine levels). The dose-by-time interaction and dose-by-time-by-effort (or control) interactions were assessed. For the dose effect, polynomial contrasts were specified to test for linear, quadratic, etc., dose effects.
To test whether immune changes were also related to subjective perceptions of effort and control, independent of condition, Pearson product moment correlations were calculated between checklist scores on perceived effort, perceived uncontrollability, and changes from baseline in immune parameters at T1, T15, and T30.
Finally, to test for mediation, regression analysis was used according to the method described by Baron and Kenny (45). Changes in heart rate, blood pressure, and norepinephrine were taken as indicators of sympathetic activation and tested as potential mediators of the effort effect on immune parameters. The change in cortisol level was tested as a mediator of the effect of uncontrollability on immune parameters.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Total scores for perceived effort and perceived uncontrollability were calculated by summing the scores on individual items. As expected, the perceived effort subscale showed a significant main effect of effort (F(1,90) = 58.3, p < .001), demonstrating that subjects in the high-effort condition reported more perceived effort after the task than subjects in the low-effort condition. However, there was also a significant main effect of control (F(1,90) = 12.0, p = .01) and a significant control-by-effort interaction (F(1,90) = 14.5, p < .001). The mean effort scores indicated that this was due to a relatively high perceived effort in the low-effort/uncontrollable condition compared with the low-effort/controllable condition. The uncontrollable and controllable condition of the high-effort task did not differ in perceived effort. Moreover, there were no differences in the actual number of sums solved in the controllable and uncontrollable high-effort conditions.
For the perceived uncontrollability subscale, only the control main effect reached significance (F(1,90) = 54.5, p < .001), demonstrating that subjects in the uncontrollable condition did indeed perceive the task as less controllable than subjects in the controllable condition.
Lymphocyte Subset Analyses, Lymphocyte Proliferation, and NKCA
Percentage of lymphocyte subsets (CD3+, CD4+, CD8+, CD14+, and CD16+) in peripheral blood, lymphocyte proliferation (no DEX added), and NKCA (1:2 dilution) were analyzed multivariately in a single analysis. (When the analysis was repeated using the 1:1 or 1:4 dilution of NKCA, results were identical.) Eighty-two subjects were included in the analysis. Data for the four conditions at T0, T1, T15, and T30 are shown in Table 1.
|
Post hoc contrasts for time further clarified the results: For percentage of CD3+, CD4+, CD8+, and CD16+ cells, the T0 vs. T1 contrast showed significant time and time-by-effort effects. Thus, immediately after the task, the percentage of subsets in the blood changed from baseline, and this change was more prominent in the high-effort condition. CD8+ and CD16+ cells also showed a significant contrast between T0 and T15 and between T0 and T30 for time but not for time by effort. Thus, the percentages of CD8+ and CD16+ cells remained elevated compared with baseline until 30 minutes after task completion, but this prolonged task effect was not different for the high- and low-effort conditions. For NKCA, all time contrasts (T0 vs. T1, T0 vs. T15, and T0 vs. T30) reached significance for the time effect, whereas the contrast between T0 and T1 and between T0 and T30 reached significance for the time-by-effort interaction. NKCA remained elevated for 30 minutes after the task, but whereas the increase immediately after the task was most prominent in the high-effort condition, at T30 the low-effort condition had the largest increase from baseline (Fig. 1, top left). Finally, lymphocyte proliferation showed a significant contrast for the time effect between T0 and T15 only. Thus, proliferation decreased 15 minutes after the task, but this was the same for all conditions (no time-by-effort or time-by-control interactions; Fig. 1, bottom left).
|
|
Cytokine Production
Complete IL-6 determinations were available for 49 subjects. Repeated-measures multiple analyses of variance were performed separately for IL-6 production at each concentration of LPS. Table 3 shows results for the sample with 1 ng/ml LPS added. Both the time effect and time-by-control interaction were significant (F(3,43) = 5.33, p = .003 and F(3,43) = 3.28, p = .030, respectively). The post hoc contrast showed that for the time effect, the contrast between T0 and T1 reached significance (demonstrating the overall increase in IL-6 production immediately after the task), whereas for the time-by-control interaction, the contrasts between T0 an T15 and between T0 and T30 reached significance. As can be seen from Figure 1 (bottom, right), 15 and 30 minutes after the stressor, IL-6 production decreased from baseline in the two uncontrollable conditions only. For the assays with 10, 0.1, and 0.01 ng/ml of LPS added, the results were similar but less strong: The time and time-by-control effects had a p value between 0.05 and 0.10. The post hoc contrasts between T0 and T15 and between T0 and T30 did reach significance for most of the assays, confirming the decreased production of IL-6 15 and 30 minutes after the task. No differences between conditions were observed only for the assay with the lowest concentration of LPS or no LPS. Neither time-by-effort nor time-by-effort-by-control interaction effects were found for any of the analyses.
|
Correlations Between Immune Changes, Perceived Effort, and Perceived Uncontrollability
The responses of the various immune parameters (at the time of greatest change from baseline) were correlated with scores on perceived effort and perceived uncontrollability. The change from baseline at T1 in percentage of CD3+ and CD4+ cells showed a significant negative correlation with perceived effort (-0.21 and -0.29), and the change from baseline at T1 in percentage and absolute numbers of CD16+ cells showed a significant positive correlation with perceived effort (0.29 and 0.25). The increase from baseline in NKCA at T1 showed a significant positive correlation with perceived effort for the undiluted and the 1:4 diluted assays (0.26 and 0.41, respectively; with a correlation of 0.20, the 1:2 diluted assay just failed to reach significance). Thus, more subjectively perceived effort was related to a larger decrease in percentage of CD3+ and CD4+ cells and a larger increase in percentage and number of CD16+ cells and NKCA. The change in percentage and absolute number of CD8+ cells and the change in lymphocyte proliferation at either T1, T15, or T30 did not significantly correlate with perceived effort.
The decrease from baseline at T15 and T30 in IL-6 (1 ng/ml LPS) significantly correlated with perceived uncontrollability (both 0.33). Also, the IL-6 assays with 10 and 0.1 ng/ml LPS showed a significant correlation with perceived uncontrollability at T15 and T30. The more uncontrollability was reported, the greater the decrease in IL-6 15 and 30 minutes after the stressor. The correlation between the proliferative response at T15 (compared with baseline) and perceived uncontrollability just failed to reach significance (r = 0.21, p = .059).
Mediation of Effort and Control Effect by Sympathetic vs. Cortisol Mechanisms
Our previous article (28) reported that effort had clear sympathetic effects, influencing the magnitude of the heart rate, blood pressure, and norepinephrine responses, whereas control had an effect on the HPAC axis, influencing plasma levels of cortisol. To investigate whether the immune changes induced by effort and uncontrollability are dependent on these sympathetic and cortisol responses, respectively, tests of mediation were performed (45). To establish mediation, three successive regression analyses were performed. The first regression should demonstrate that the independent variable (eg, effort) significantly affects the mediator (eg, heart rate). The second regression should demonstrate that the independent variable affects the dependent variable (eg, percentage of CD8+ cells). Finally, in the third regression, both the mediator and independent variable are entered as predictors of the dependent variable. The mediator should show up as a significant predictor of the dependent variable, whereas the independent variable should no longer be significantly associated with the independent variable, or at least its ß coefficient should be substantially reduced in comparison to the second equation (without the mediator entered).
We applied these models to test for mediation of the effort effect for variables showing a significant effort effect (percentage of CD3+, CD4+, CD8+, and CD16+ cells, total number of CD8+ and CD16+ cells, and NKCA). Heart rate, SBP, DBP, and plasma norepinephrine at T1 - T0 were entered as mediators in successive analyses. The results for these regression equations are shown in Table 4. For percentage of CD8+ cells, heart rate appeared as a mediator of the effort effect. The first regression equation showed that perceived effort significantly predicted the increase in heart rate at T1 (left column, ß = 0.46, p < .001). The second equation showed that perceived effort was also significantly associated with the increase in CD8+ cells (right column, in parentheses; ß = 0.31, p = .001). Finally, in the third equation, it was demonstrated that when both effort and heart rate response were entered as predictors, only heart rate response was significantly associated with the increase in percentage of CD8+ cells (middle column; ß = 0.33, p = .005), and the association between effort and change in percentage of CD8+ was no longer significant (right column; ß = 0.16, NS). Similar mediation models were applied to the other parameters correlating with perceived effort. For percentage of CD3+ and CD4+ cells, no mediation could be established. CD8+ and CD16+ cells (percentage and absolute numbers) and NKCA did seem to be mediated by sympathetic effects: Heart rate and especially SBP and DBP appeared as mediators in the analyses.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
We predicted that the amount of mental effort would specifically be related to changes in the number of CD3+, CD4+, CD8+, and CD16+ cells, NKCA, and lymphocyte proliferation and that these changes would be mediated by sympathetic activation. These predictions were largely confirmed. The change from baseline in the percentages of CD3+, CD4+, CD8+, and CD16+ cells were all significantly affected by effort condition, and perceived effort was significantly correlated with the CD3+, CD4+, and CD16+ percentage responses. However, when absolute number of cells was considered, an effort effect was present only for the increase in CD16+ and CD8+ cells. The decrease in percentage of CD3+ and CD4+ cells therefore is probably the result of the elevated number of CD8+ and CD16+ cells in the blood. The same divergence in results for percentage vs. absolute number of cells was reported by Naliboff et al. (3).
Also consistent with the prediction, the increase in NKCA immediately after the stressor was significantly affected by effort condition, and the magnitude of the increase was positively correlated with perceived effort. However, NKCA was assessed in whole blood, and the increase in NKCA may depend on the increased number of NK cells in blood. Therefore, analysis of NKCA was repeated using the increase in absolute number of NK cells at T1 as a covariate (results not shown). Again, both the time effect as well as the time-by-effort effect reached significance, also for the T0/T1 contrast. Thus, the greater increase in NKCA immediately after the stressor in the high-effort condition probably cannot be attributed entirely to the increase in the number of NK cells immediately after the stressor.
Finally, the prediction that effort would affect lymphocyte proliferation could not be confirmed. Task performance per se did have an effect on lymphocyte proliferation: Fifteen minutes after the task, there was a significant overall decrease in proliferation from baseline. However, the magnitude of the decrease was not related to differences in effort between the tasks or to self-rated perceived effort.
We further tested whether the effort-induced immune changes were mediated by changes in sympathetic nervous system activity. The increases in heart rate, blood pressure, and plasma norepinephrine were considered as indicators of sympathetic activation. Heart rate and especially blood pressure responses appeared to be associated with the changes in CD8+and CD16+ cells and NKCA. For the changes in percentage of CD3+ and CD4+ cells, no evidence of sympathetic mediation was found, confirming that these changes may be secondary to the changes in number of CD8+ and CD16+ cells.
The second prediction was that uncontrollability would specifically affect cytokine production (IL-4 and IL-6) and lymphocyte proliferation and that this would be mediated by cortisol release. For lymphocyte proliferation, no control main effect was found. We did find an effect of uncontrollability on mitogen induced IL-6 production. In the two uncontrollable conditions, but not in the controllable conditions, in vitro IL-6 production was decreased 15 and 30 minutes after task completion. Moreover, perceived uncontrollability was significantly correlated with the magnitude of the decrease in IL-6 15 and 30 minutes after the stressor. For the mitogen-induced production of IL-4, no control effect was found.
The second part of the prediction said that immune changes induced by uncontrollability are mediated by cortisol. As presented in our previous article (28), the uncontrollable tasks led to a significantly higher plasma level of cortisol than the controllable tasks. Moreover, the delayed effect on IL-6 production (15 and 30 minutes after the stressor) is more congruent with a more slowly evolving endocrine process rather than with the rapid sympathetic nervous system effects. However, in our regression models, cortisol did not show up as a mediator of decreased production of IL-6. There are several possible reasons for this result. First, the latency and time course of the cortisol response probably shows a high degree of interindividual variation. Therefore, it is difficult to relate the magnitude of the cortisol response at a single point in time to the magnitude of the immune change. In fact, overall salivary cortisol decreased from T0 on to T30, probably reflecting the circadian rhythm of cortisol release. Uncontrollability thus did not lead to an increase in cortisol but to a lesser decline. (In another study (46), it was demonstrated that when the same stressor (only high effort) was applied in the afternoon, when the circadian rhythm is less prominent, the high-effort/uncontrollable condition led to a significant increase in cortisol from baseline.) Second, both the relatively higher level of cortisol and lower level of cytokine production could be related to a third factor, an increased level of adrenocorticotropic hormone, which stimulates release of cortisol from the adrenal cortex and has an inhibitory effect on cytokine production (47).
An indirect way to test whether uncontrollability affects immune system function through the release of cortisol was performed by assessing the control effect on the dose-response curve for DEX-induced suppression of lymphocyte proliferation. It was predicted that if proliferation would become suppressed by endogenous cortisol released in uncontrollable situations, there would be less opportunity for DEX to induce additional suppression. Immediately after the task, as well as 15 minutes later, suppression of proliferation by DEX was significantly less than before the task, when low doses of DEX were added. Fifteen minutes after the task, but not immediately after the task, there was also a task-induced decrease in lymphocyte proliferation, and one may speculate that this was mediated by increased levels of cortisol after the task that prevented DEX from further suppression. However, the expected differential effect of uncontrollability on the proliferative response or suppression by DEX could not be confirmed.
The finding of the present study that both effort and uncontrollability influence different parameters of immune system activity may explain some of the seemingly contradictory findings of previous studies on uncontrollability and immune function in humans. In these studies, not only controllability differed between conditions; there were also variations in effort. In the study of Weisse et al. (8), it seemed that subjects in the controllable condition had made more frequent responses (ie, invested more effort) than subjects in the uncontrollable condition. Sieber et al. (10) introduced differences in the possibility to respond as part of their manipulation. They had three experimental conditions. In one condition, subjects could control loud noise by giving the correct response; in another condition, they were given the suggestion of control but actually the responses had no effect. In the last condition, subjects could not respond at all. It was found that only subjects who could not respond at all showed decreases in NKCA after the stressor. The authors suggested that the results can be explained by the absence of potential control in this condition, but at the same time this was the only condition that did not require effort from the subjects. Finally, in the Gomez et al. study (17), subjects in the no-control condition gave higher ratings of subjective effort, and subjective controllability was negatively correlated with subjective effort. Thus, in none of these previous studies can the effects of uncontrollability be differentiated from the effects of effort.
It may be argued that the effects of uncontrollability were also confounded by effort in the present study. In the low-effort condition, subjects who did not have control reported more perceived effort than subjects who had control. Subjects in the low-effort/controllable condition soon discovered that they had to press the same key on each trial. Subjects in the uncontrollable condition had to keep on searching for the supposed system throughout all 16 trials, which required more effort. Still, we are confident in ascribing the differences in the immune effects between the high- and low-control conditions to uncontrollability and not to differences in effort. First, within the high-effort condition, the control manipulation did not have an effect on perceived effort; nevertheless, uncontrollability influenced the immunological parameters in this condition. Second, uncontrollability affected a different immunological parameter than effort; if the effect was due to differences in perceived effort, enumerative parameters and NKCA, rather than cytokine production, would have been affected. Finally, the decrease in IL-6 production was correlated with perceived uncontrollability but not perceived effort. Therefore, we conclude that effort and uncontrollability indeed have independent effects on the immune system.
One limitation of this study is that because of procedural errors, data on one of our most prominent parameters, mitogen-induced IL-6 production, were lost for the first 29 subjects, leaving only 49 subjects in the final analyses. Another immunological parameter we had planned to include, mitogen-induced production of the cytokine interferon-
, had to be dropped altogether because the reliability of the assays was doubtful. Therefore, the generalizability of the suppressive effect of uncontrollability on cytokine production is limited. Moreover, there is always a limitation to the generalizability of laboratory findings to real life. The profound disruptive effect that naturally occurring stressful experiences may have, and the possible moderating influences of social support and coping responses, can never be captured in the laboratory.
Nevertheless, this study does prove that careful experimental elucidation of specific psychological influences on immune, and associated endocrine and cardiovascular, responses is possible and contributes to a better understanding of the intricacies of stress-immune relations. Two distinct features of stress situations were highlighted, effort and uncontrollability, and both were shown to affect the immune system response in different ways. Effort mainly affects cell numbers (especially CD8+ and CD16+ cells) and NKCA, and this effect is probably dependent on sympathetic nervous system activation. Uncontrollability inhibits cytokine production (IL-6), which may be dependent on activation of the HPA axis. Moreover, whereas the sympathetically mediated effort effect was transient and only observable immediately after the stressor, the uncontrollability effect seemed to be more prolonged. A stressor as brief as 15 minutes already led to an inhibition of IL-6 production lasting 30 minutes or more (no more blood samples were taken after 30 minutes). Moreover, by interfering with cytokine production, uncontrollability may indirectly affect a wide range of immunological functions, because cytokines play an essential role in activation of various aspects of immunological defense. Therefore, it might be expected that whether a stressor is controllable will have more consequences for health than the demand characteristics of a stressor. Future studies should be aimed at the generalizability of the effects of uncontrollability to other functional immune parameters and at its long-term effects. However, as this study has proven, experimental designs should always control for differences in effort between conditions.
Received for publication June 23, 1998.
Revision received April 28, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. Kern and T. Ziemssen Review: Brain immune communication psychoneuroimmunology of multiple sclerosis Multiple Sclerosis, January 1, 2008; 14(1): 6 - 21. [Abstract] [PDF]
|
|
|
|
 |
|
 E. Chen Impact of Socioeconomic Status on Physiological Health in Adolescents: an Experimental Manipulation of Psychosocial Factors Psychosom Med, May 1, 2007; 69(4): 348 - 355. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Griffin and E. Chen Perceived Control and Immune and Pulmonary Outcomes in Children With Asthma Psychosom Med, May 1, 2006; 68(3): 493 - 499. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. K. Kiecolt-Glaser, L. McGuire, T. F. Robles, and R. Glaser Psychoneuroimmunology and Psychosomatic Medicine: Back to the Future Psychosom Med, January 1, 2002; 64(1): 15 - 28. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
