Effects of Experimental Psychological Stress on Distribution and Function of Peripheral Blood Cells Jos F. BROSSCHOT, PHD, ROBERT J. BENSCHOP, MSC, GUIDO L. R. GODAERT, PHD, MARTIN B. M. DE SMET, MIRANDA OLFF, PHD, COBI J. HEIJNEN, PHD, AND RUDY E. BALLIEUX, PHD Fifty male subjects (aged 24 to 55 years) were subjected to a mild and potentially uncontrollable interpersonal stress situation. They were asked to solve a difficult puzzle. Subsequently they were requested to explain their solution to "another subject," actually a confederate to the researchers. The confederate frustrated the subjects' explanation efforts. Care was taken that neither solving nor explaining of the puzzle was successful. The experimental situation induced mild psychological strain as documented by mood changes in the experimental group when compared with a control group of 36 male subjects. Peripheral blood was drawn by an indwelling catheter just before, directly after. 15 minutes after, and 30 minutes after the stress situation. Numbers of leukocytes, lymphocytes, monocytes, T-cell subsets, natural killer (NK) cells, and B-cells were determined. As functional assays we used in vitro proliferative responses of T- and B-cells to mitogenic stimulation (PHA and PWM) and to an antigen cocktail. The potential influences of health- and biobehavioral variables were taken into account in the analyses, as well as incidental differences in initial mood or immunological baseline. The results replicated and expanded on previous research. In contrast to controls, experimental subjects showed a significant increase in numbers of NK cells after the stress-period, returning to baseline values after 15 minutes of rest. A similar effect was shown on T-suppressor/ cytotoxic cells and, inversely, on T-helper/suppressor ratio, but these effects could be attributed to changes in the numbers of CD8+CD57+ cells. No effects were observed on proliferation. From the results we conclude that the effects of a short lasting mild psychological stressor are mainly restricted to cells of the NK cell population. Key words: immunology; psychology; psychoneuroimmunology; experimental (psychological) stress; peripheral blood cells; lymphocyte proliferation.
INTRODUCTION
mises firm conclusions on the causal relationship between psychological events Most of human psychoneuroimmunol- and immunological functioning. Furtherogical (PNI) studies are correlational or more, in the few experimental studies in prospective field studies. This compro- humans, psychological strains are mostly associated with "physical" stress like sleep deprivation and noise (1-5), a 20From the Department of Clinical Psychology and meter fall in a life boat (6), or a space flight (7, 8). Health, Faculty of Social Sciences. University of Utrecht (J.F.B., G.L.R.G., M.O.); the Department of To our knowledge, in only a few human
Pediatric Immunology, University Hospital for Children and Youth "Het Wilhelmina Kinderziekenhuis." (C.J.H.); and the Department of Clinical Immunology, University Hospital (R.J.B., MB.M.d.S., R.E.B.), Utrecht, The Netherlands. Address reprint requests to: Jos F. Brosschot, Department of Clinical and Health Psychology, Faculty
394 0033-3174/92/5401-O394SO3 00/0 Copyright © 1992 by the American Psychosomatic Society
of Social Sciences, University of Utrecht, P.O. Box 80140, 3508 TC Utrecht, The Netherlands. Received for publication April 23, 1991; revision received March 13, 1992
Psychosomatic Medicine 54:394-406 (1992)
STRESS EFFECTS O N BLOOD CELLS
experimental studies, a purely psychological experimental stressor has been applied, resulting in increases in CD8+ and natural killer (NK) cells and in NK cell activity (9, 10). In the present study psychological strain was induced by confronting human volunteers with a semirealistic, ambiguous, and potentially uncontrollable interpersonal situation. To control for the effects of the blood sampling procedure, control subjects were assigned to a neutral, relaxing condition. Stressor uncontrollability and unpredictability have been shown to be important determinants of psychological and physiological outcomes (11-15). Furthermore, these features have also been demonstrated to be of relevance to stress-induced changes in immune responses (5, 16-20). PNI field studies with human subjects, only yield indirect evidence for the importance of uncontrollability (6, 21, 22). Interpersonal aspects, like elicitation of feelings of guilt and shame, are also major determinants of adversity of stressors (2326). The stress situation employed in the present study is potentially uncontrollable as well as interpersonal. We have shown earlier that this paradigm can evoke psychological strains and changes in percentages of T-cell subsets (9). In the present study, the effect of the stressful episode on in vitro proliferation responses of peripheral blood lymphocytes was tested; in addition changes in distribution patterns of several lymphocytes subsets were analyzed. Double staining of CD8+CD57+ cells was used to define more specifically the characteristics of the affected subsets than had been done in previous research. As another expansion, immunological parameters were measured repeatedly after the termination of the stressor. Psychosomatic Medicine 54:394-406 (1992)
METHODS
Subjects Ninety-two male high school teachers participated in the study. Only male subjects were chosen because female endocrine cycles would complicate the interpretation of the results. To obtain a homogeneous group, it was assured that only teachers were selected who had a workload of over 20 teaching hours weekly. Six subjects were removed from the analyses. Three of these had chronic diseases with corresponding medication (Cushing's Disease, Colitis Ulcerosa, and Diabetes) that may interfere with immunocompetence. The three others had a tendency to faint during venapuncture. The final group consisted of 86 subjects. Ages ranged from 24 to 55 years (mean = 40.5 years; SD = 6.9). The subjects were randomly assigned to an experimental (N = 50) or control condition (N = 36). More subjects were included in the experimental condition to allow for some additional within group analyses for other purposes than reported here.
Experimental Manipulation Before the experimental manipulation, the subjects were not aware of the existence of a control condition. Information given beforehand included that the present experiment was designed to study the effects of levels of task load upon hormonal patterns in peripheral blood. The experimental procedure started at 9:00 A.M., at which time a catheter was inserted, and lasted until 10:45 A.M. (see Scheme 1). After an initial 30mmute rest period, in which a neutral videofilm was shown, the subjects rated their initial mood state (time 1 (Tl)). Then a first blood sample was taken, and subsequently they were asked to solve a threedimensional puzzle in less than 8 minutes. The puzzle was shown in such a fashion that it looked solvable. Unknown to the subjects, one piece of the puzzle had been modified, resulting in an unsolvable puzzle. In this way care was taken to let every subject fail objectively. Next they were asked to explain a problem-solving strategy for the puzzle to another subject (actually a confederate (CF) to the experimenter). The puzzle had to be explained verbally, excluding any manipulation. The subjects (SS) incentive to succeed in explaining was partly assumed to be intrinsic (SS being teachers), and partly manipulated by: a) showing their 'teaching efficacy'
395
J. F. BROSSCHOT et al. decline from '10' to '0' (on a computer screen) during the time they were actually explaining; b) by showing the CF loosing money when the SS did not explain. Because the CF's performance did not improve despite the subject's efforts, the situation was potentially uncontrollable. This situation continued for 12 minutes. The stress period (solving the unsolvable puzzle and explaining it to the CF) took about 30 minutes, including instructions. On time 2 (T2) a second blood sample was taken, and mood was once more self-rated after the stress-situation. This period was followed by a 45-minute rest period, identical with the first, with another neutral film. On time 3 (T3) (after 15 minutes of rest) and time 4 (T4) (after 45 minutes of rest), respectively the third and fourth blood samples were taken. Before the fourth sample, mood was rated for the third and last time. The controls were subjected to the same procedure: only the problem solving and explication periods were replaced by reading popular magazines.
Blood Sampling and Immunological Analyses Peripheral blood rnononuclear cells (PBMC) were isolated by diluting peripheral blood with an equal volume of minimal essential medium (Gibco). supplemented with penicillin (100 IU/ml) and streptomycin (100 Mg/ml; MEM/PS), and separated using Ficoll/Paque (d=l.O77 g/cm3, Pharmacia, Sweden) gradient centrifugation (1000 x g, 20 minutes). The interphase containing the PMNC was collected and washed twice (400 X g, 10 minutes) with MEM/PS. Cells were resuspended in MEM/PS and counted in a hemocytometer. Viability of the isolated cells always exceeded 95% as determined by trypan blue dye exclusion. For lymphocyte subset analysis, cells were stained using simultest sets, containing directly F1TC- and PE-conjugated monoclonal antibodies against CD4 (Ieu3)/CD8 (Ieu2a), CD3 (Ieu4)/HLA-DR, and CD8 (Ieu2a)/CD57 (Ieu7). Also directly FITCconjugated antibodies against CD16 (leulla) were used (all monoclonals from Becton-Dickinson. USA). Cells (0.5*10'') were pelleted and incubated with the different monoclonal antibodies for 20 minutes on ice. Cells were washed twice using MEM/PS, supplemented with 0.5% BSA and 0.1% sodium azide. nfler which they were analyzed using a flowcytomcter (FACScan, Becton-Dickinson, USA). Proliferative responses of PMNC were tested by incubating 0.2*10c PMNC with different mitogens.
396
Mitogens used were phyto heamagglutinin (PHA) in two concentrations (40 and 60 Mg/ml, Wellcome diagnostics, the Netherlands); pokeweed mitogen (PVVM; 30 Mg/ml; Gibco), and an antigen cocktail (AG) containing varidase (40 lU/ml), tuberculin PPD (40 Mg/ml; Statens Serum Institute, Denmark), tetanus toxoid (10 LF/ml; RIVM, the Netherlands), and Candida (200 Mg/ml; Haarlems Allergenen Laboratorium, the Netherlands). Cells were cultured in 96 Uwell microtiter plates (NUNC, Denmark) using RPMI 1640 (Gibco) supplemented with 20% heat inactivated pooled human serum, penicillin (100 IU/ml), streptomycin (100 jig/ml), and 1-glutamin (2 HIM). After 5 days 3H-thymidine was added and the plates were harvested after an additional 16 to 18 hours of culturing on appropriate filters. Radioactive counts per minute (cpm) were determined after the addition of 3 ml of scintillation fluid in a LSC counter. Total numbers of leukocytes, monocytes, and lymphocytes per sample were determined in whole blood using an automated closed tube sampler (Technicon H-l System) at the Department of Heamatology, University Hospital Utrecht, the Netherlands.
Biobehavioral Data On the day before the experiment, the General Health Questionnaire, developed by Ursin and coworkers (27) was administered. In a first part of this checklist, respondents can report amount and severity of health complaints during the month preceding the experiment. Only those scores that were considered relevant to the immune system (28) were used. From a second part of the same list, information on daily alcohol and coffee intake, smoking, physical exercise, averaged nightly sleeping hours, subjective fitness and health was obtained. In addition, the numbers of days of sick leave, days of illness, and visits to a physician over the last year were asked. Report of chronic disease and regular medicine intake was requested.
Assessment of Mood Levels Visual analogue scales were used to assess initial mood levels and mood changes during the experiment. The method has been used by several Scandinavian researchers (29-31). Each subject rated his perceived 'boredom,' 'ability to concentrate.' 'ef-
Psychosomatic Medicine 54:394-406 (1992)
STRESS EFFECTS O N BLOOD CELLS fort,' 'impatience,' 'interest,' 'irritation,' 'tenseness.' 'tiredness,' 'cheerfulness,' and 'sadness.' Mood was assessed three times during the experiment, just before blood sampling on Tl, T2 and, in a smaller subsample, T4.
Statistical Analyses The effect of the experimental manipulation on the immunological parameters was tested as interactions between stress condition (experimental versus control) and time (blood sampling moments), with analysis of variance (ANOVA) for repeated measures (four). Averaged tests of significance were used (equivalent to the split-plot approach), with Greenhouse-Geisser correction of degrees of freedom. Intergroup differences in linear and curvilinear trends were tested in these analyses with the use of polynomial contrasts. Student t tests were used to test for differences between the experimental and control group with respect to baseline of immunological parameters, initial mood, and health variables. Chi-square tests were applied when variables were noncontinuous, (like smokers versus nonsmokers). Despite random assignment, some differences existed on these variables between the groups. To test whether different immunological reaction patterns for the two conditions could be attributed to these differences, the ANOVA analyses were performed once more, using these variables as covariates. Differences between groups as to mood change scores were also tested as a two-way interaction ANOVA for repeated measures. Only the two first measurements (before and after the stressor) were used for the present purposes. For all the statistical analyses SPSS/PC+ subprograms were used (32). Differences in reported degrees of freedom are caused by missing values.
RESULTS
Biobehavioral Variables and Initial Mood Levels No differences on initial mood levels existed between the experimental and control groups. Despite random assignment to the conditions, a few statistically Psychosomatic Medicine 54:394-406 (1992)
significant differences (p < 0.05) between the experimental and control group were, however, observed in the biobehavioral variables. People of the experimental group were, on average, somewhat older than those of the control group (42.0 years versus 38.7 years]. Moreover, they reported less symptoms of colds/influenza and coughing/bronchitis over the last month, and less days of sick leave over the last year. Furthermore, experimental subjects reported fewer average daily coffee drinking than control subjects. Since these factors may be of importance with respect to the outcome of the immunological analyses (28), they were used as covariates in the analyses of covariance of immunological changes.
Stressor-induced Changes in Mood In Figure 1 mood changes are depicted as poststress (T2) values minus prestress values (Tl). ANOVA for repeated measures, with interaction between stress-condition and time, showed that most of the differences observed between the experimental and control group are significant. The results show that, in contrast to the control subjects, the experimental subjects report increases in 'effort' (F(l,71) = 30.84, p < 0.001), 'irritation' (NS), 'tenseness' (F(l,74) = 13.76, p < 0.001), 'impatience' (NS), 'concentration' (F(l,74) = 15.98, p < 0.001) and 'interest' (F(l,74) = 8.16, p < 0.01) during the stressor. Measurements at T4, which were only performed in a smaller subsample (about two thirds of the main sample), showed that all the moods that were changed at T2 had returned to baseline values at T4. Taking these data together, we can conclude that the experimental situation in397
J. F. BROSSCHOT et al. MOOD CHANGES (after - before stressor) Effort
m
m
m
m
^
—
< H H »
Irritation Tenseness
MfflgJ™
Impatience Concentration
mmmm
Boredom Interest
*** 8gjgg})SS38Sg
mmm•
Tiredness Cheerfulness
i
1
Sadness
experimental group (N=5 1)
control group (N=25)
Fig. 1. Differences in mood changes (before (Tl) minus after (T2) the stressor) between the experimental group and the control group. Positive numbers signify an increase in mood; negative numbers, a decrease in mood. Indicated are significances of ANOVA with condition x time: ***p < 0.001; "••p < 0.01.
duces a mild temporarily psychological arousal and distress.
Stressor-Induced Changes in Peripheral Blood Leukocytes, Monocytes, and Lymphocyte Subset Distribution Repeated measures ANOVAs over the four blood samples (see Scheme 1) show significant 2-way interactions (Time X Condition) (see Table 1). Experimental and control groups have different changes over time for numbers of lymphocytes, 398
CD8+ cells, CD57+ cells, CD16+ cells, and the CD4+/CD8+ ratio. The shift in the CD4+/CD8+ ratio appears to be mainly dependent on the increase in CD8+ cells. No significant interactions of Time X Condition were found for leukocytes, monocytes, CD3+, CD4+, and HLADR"1". The trends in the parameters that show significant responses to the stressor are illustrated in Figure 2. The changes in distribution of cell types are expressed as percentages of the individual baseline level (at Tl). In this way it can be seen that the mean number of CD57+ cells in the experimental group increased about 26%, whereas a decrease of about 11% can be observed in the control group. At time T3 the numbers of these cells return to baseline values. After T3, both groups diverge again in opposite directions. This pattern, which can be expressed as a cubic trend, is also visible in the responses of the CD16+ cells, the CD8+ cells, the lymphocyte numbers, and the CD4+/ CD8+ ratio. Polynomial contrast analyses reveal that the conditions differed significantly for this cubic trend with regard to CD57+, CD16+, and CD8+, and also for the linear and quadratic trends with regard to the same celltypes (p = < 0.01 at least). The lymphocyte numbers differed only for the linear trend. Since the markers CD8 and CD57 are shared by some T-suppressor/cytotoxic cells and NK cells, additional analyses were performed on the CD8+CD57+ cells (double positive cells), and on the remaining populations of CD8+CD57~ and CD8~CD57+ cells. The results are shown in Table 2 (and Figure 3). As demonstrated in Table 2, the CD8+CD57+ double positive cells are responsible for the change in the CD8+ population. The increase in CD8+CD57~ is not significant, while the trend in CD57+ (CD8~CD57+) is now Psychosomatic Medicine 54:394-406 (1992)
STRESS EFFECTS O N BLOOD CELLS Scheme 1. Overview of the experimental session.
Minutes 0
30
RESTl (neutral video)
60
45
PROBLEM SOLVING PERIOD
EXPLICATION PERIOD T2
Tl
even more 'peaked' (see Figure 3), and highly significant. The difference in the CD4+/CD8+ ratio (not shown) is also not significant anymore when CD8+CD57~ cells were used. Hence, the effect of the stressor appears to be most pronounced on cells with the NK cell phenotype expressing the CD57 marker, or expressing CD57 together with CD8.
Analyses of Covariance For CD57+ and CD8+57+ the baseline (Tl) values turned out to be significantly higher in the experimental group (p < 0.05). Therefore, ANOVA analyses were performed over T2, T3, and T4, using Tl as a covariate. This procedure, however, did not affect the results, meaning that the difference in reactivity patterns could not be attributed to baseline differences. Using as covariates the five potential confounders, which differ between the experimental and control groups (age, colds/influenza and coughing/bronchitis over the last month, days of sick leave during last year, daily coffee drinking, and initial "interest"), yielded no substantially different results. Hence, accidental differences in relevant variables between the experimental and control group could not explain the differences in reactivity between the experimental group and the control group. Psychosomatic Medicine 54:394-406 (1992)
105
75
REST 2 (neutral video) T3
T4
Proliferative Responses No significant interactions between time and condition were observed for the proliferative responses. A baseline difference existed between the conditions with respect to PHA-stimulated proliferation (40 jtg/ml). Additional covariate analysis, however, did not change the finding that there was no effect for this functional parameter.
DISCUSSION
In the present study it is shown that a short, mild psychological stressor can influence the distribution patterns of lymphocytes and NK cells in peripheral blood. No differences were observed when the effect of the stressor was tested on proliferative responses of T and B lymphocytes. The straining effect of the stressor was verified by significant changes in measures of reported effort and distress in the experimental subjects. We controlled for potential biobehavioral confounders (age, health symptoms, and health behavior variables). Where 'accidental' differences existed between the conditions, the analyses were corrected by using the potentially confounding variables as covariates in the analyses of variance. It was clearly shown that the results could not be attributed to these variables. Our results confirm and expand on pre399
J. F. BROSSCHOT et al. 9
TABLE 1. Cell Counts (x lO /Liter) on Times Tl, T2, T3, and T4 Exp
Leukocytes
Lymphocytes
Monocytes
CD3 +
CD4 +
CD8 +
CD4 + /CD8 + RATIO
CD57 +
CD16 +
HLADR+
T1 T2 T3 T4 Tl T2 T3 T4 Tl T2 T3 T4 Tl T2 T3 T4 T1 T2 T3 T4 T1 T2 T3 T4 Tl T2 T3 T4 Tl T2 T3 T4 T1 T2 T3 T4 Tl T2 T3 T4
Con
F (time
Mean
Std
Mean
Std
5.32 5.57 5.53 5.71 1.61 1.66 1.57 1.57 0.400 0.399 0.401 0.393 1.11 1.10 1.08 1 12 0.717 0.697 0.708 0.765 0.529 0.568 0 503 0.470 1.51 1.41 1.59 1.84 0.299 0.354 0.275 0.233 0.236 0 262 0.193 0.159 0.322 0.341 0.344 0.348
1.7 1.8 1.9 2.0 0.46 0.49 0.45 0.49 0.16 0.16 0.18 0.17 0.32 0.35 0.34 0.35 0.22 0.23 0.23 0.25 0.21 0.24 021 0.21 0.10 0.11 0.12 0.13 0.17 0.19 0.14 0.13 0.16 0.16 0.12 0.12 0.13 0.15 0.14 0.15
5.44 5.51 5.56 5.81 1.54 1.53 1.55 1.64 0.360 0.355 0.372 0.401 1.05 1.04 1.06 1.11 0.714 0.719 0.726 0.769 0.470 0.435 0.453 0.472 1.66 1.75 1.75 1.75 0.217 0.185 0.197 0.210 0.171 0 146 0.147 0.147 0.335 0.345 0.349 0.368
1.8 1.9 1.8 1.9 0.43 0.41 0.40 0.43 0.16 0.17 0.17 0.18 0.32 0.34 0.30 0.34 0.25 0.23 0.22 0.27 0.17 0.13 0.15 0.16 0.12 0.10 0.11 0.11 0.11 0.09 0.11 0.10 0.10 0.10 0.08 0.09 0.14 0.15 0.13 0.15
X cond.) 1.18
3,249
11.50**
3,243
2.83
3,189
2.25
3,234
1.17
3,234
17.09**
3,234
12.84"
3,234
18 17"
3,234
6.68"
3,135"
0.59
3,234
'' Degrees of freedom before Creenhouse-Ceisser correction; significancies shown are after correction. h CDl 6+ was only determined in a part of the sample, hence the smaller DF here. " * p < 0.001. "p
INTRODUCTION
mises firm conclusions on the causal relationship between psychological events Most of human psychoneuroimmunol- and immunological functioning. Furtherogical (PNI) studies are correlational or more, in the few experimental studies in prospective field studies. This compro- humans, psychological strains are mostly associated with "physical" stress like sleep deprivation and noise (1-5), a 20From the Department of Clinical Psychology and meter fall in a life boat (6), or a space flight (7, 8). Health, Faculty of Social Sciences. University of Utrecht (J.F.B., G.L.R.G., M.O.); the Department of To our knowledge, in only a few human
Pediatric Immunology, University Hospital for Children and Youth "Het Wilhelmina Kinderziekenhuis." (C.J.H.); and the Department of Clinical Immunology, University Hospital (R.J.B., MB.M.d.S., R.E.B.), Utrecht, The Netherlands. Address reprint requests to: Jos F. Brosschot, Department of Clinical and Health Psychology, Faculty
394 0033-3174/92/5401-O394SO3 00/0 Copyright © 1992 by the American Psychosomatic Society
of Social Sciences, University of Utrecht, P.O. Box 80140, 3508 TC Utrecht, The Netherlands. Received for publication April 23, 1991; revision received March 13, 1992
Psychosomatic Medicine 54:394-406 (1992)
STRESS EFFECTS O N BLOOD CELLS
experimental studies, a purely psychological experimental stressor has been applied, resulting in increases in CD8+ and natural killer (NK) cells and in NK cell activity (9, 10). In the present study psychological strain was induced by confronting human volunteers with a semirealistic, ambiguous, and potentially uncontrollable interpersonal situation. To control for the effects of the blood sampling procedure, control subjects were assigned to a neutral, relaxing condition. Stressor uncontrollability and unpredictability have been shown to be important determinants of psychological and physiological outcomes (11-15). Furthermore, these features have also been demonstrated to be of relevance to stress-induced changes in immune responses (5, 16-20). PNI field studies with human subjects, only yield indirect evidence for the importance of uncontrollability (6, 21, 22). Interpersonal aspects, like elicitation of feelings of guilt and shame, are also major determinants of adversity of stressors (2326). The stress situation employed in the present study is potentially uncontrollable as well as interpersonal. We have shown earlier that this paradigm can evoke psychological strains and changes in percentages of T-cell subsets (9). In the present study, the effect of the stressful episode on in vitro proliferation responses of peripheral blood lymphocytes was tested; in addition changes in distribution patterns of several lymphocytes subsets were analyzed. Double staining of CD8+CD57+ cells was used to define more specifically the characteristics of the affected subsets than had been done in previous research. As another expansion, immunological parameters were measured repeatedly after the termination of the stressor. Psychosomatic Medicine 54:394-406 (1992)
METHODS
Subjects Ninety-two male high school teachers participated in the study. Only male subjects were chosen because female endocrine cycles would complicate the interpretation of the results. To obtain a homogeneous group, it was assured that only teachers were selected who had a workload of over 20 teaching hours weekly. Six subjects were removed from the analyses. Three of these had chronic diseases with corresponding medication (Cushing's Disease, Colitis Ulcerosa, and Diabetes) that may interfere with immunocompetence. The three others had a tendency to faint during venapuncture. The final group consisted of 86 subjects. Ages ranged from 24 to 55 years (mean = 40.5 years; SD = 6.9). The subjects were randomly assigned to an experimental (N = 50) or control condition (N = 36). More subjects were included in the experimental condition to allow for some additional within group analyses for other purposes than reported here.
Experimental Manipulation Before the experimental manipulation, the subjects were not aware of the existence of a control condition. Information given beforehand included that the present experiment was designed to study the effects of levels of task load upon hormonal patterns in peripheral blood. The experimental procedure started at 9:00 A.M., at which time a catheter was inserted, and lasted until 10:45 A.M. (see Scheme 1). After an initial 30mmute rest period, in which a neutral videofilm was shown, the subjects rated their initial mood state (time 1 (Tl)). Then a first blood sample was taken, and subsequently they were asked to solve a threedimensional puzzle in less than 8 minutes. The puzzle was shown in such a fashion that it looked solvable. Unknown to the subjects, one piece of the puzzle had been modified, resulting in an unsolvable puzzle. In this way care was taken to let every subject fail objectively. Next they were asked to explain a problem-solving strategy for the puzzle to another subject (actually a confederate (CF) to the experimenter). The puzzle had to be explained verbally, excluding any manipulation. The subjects (SS) incentive to succeed in explaining was partly assumed to be intrinsic (SS being teachers), and partly manipulated by: a) showing their 'teaching efficacy'
395
J. F. BROSSCHOT et al. decline from '10' to '0' (on a computer screen) during the time they were actually explaining; b) by showing the CF loosing money when the SS did not explain. Because the CF's performance did not improve despite the subject's efforts, the situation was potentially uncontrollable. This situation continued for 12 minutes. The stress period (solving the unsolvable puzzle and explaining it to the CF) took about 30 minutes, including instructions. On time 2 (T2) a second blood sample was taken, and mood was once more self-rated after the stress-situation. This period was followed by a 45-minute rest period, identical with the first, with another neutral film. On time 3 (T3) (after 15 minutes of rest) and time 4 (T4) (after 45 minutes of rest), respectively the third and fourth blood samples were taken. Before the fourth sample, mood was rated for the third and last time. The controls were subjected to the same procedure: only the problem solving and explication periods were replaced by reading popular magazines.
Blood Sampling and Immunological Analyses Peripheral blood rnononuclear cells (PBMC) were isolated by diluting peripheral blood with an equal volume of minimal essential medium (Gibco). supplemented with penicillin (100 IU/ml) and streptomycin (100 Mg/ml; MEM/PS), and separated using Ficoll/Paque (d=l.O77 g/cm3, Pharmacia, Sweden) gradient centrifugation (1000 x g, 20 minutes). The interphase containing the PMNC was collected and washed twice (400 X g, 10 minutes) with MEM/PS. Cells were resuspended in MEM/PS and counted in a hemocytometer. Viability of the isolated cells always exceeded 95% as determined by trypan blue dye exclusion. For lymphocyte subset analysis, cells were stained using simultest sets, containing directly F1TC- and PE-conjugated monoclonal antibodies against CD4 (Ieu3)/CD8 (Ieu2a), CD3 (Ieu4)/HLA-DR, and CD8 (Ieu2a)/CD57 (Ieu7). Also directly FITCconjugated antibodies against CD16 (leulla) were used (all monoclonals from Becton-Dickinson. USA). Cells (0.5*10'') were pelleted and incubated with the different monoclonal antibodies for 20 minutes on ice. Cells were washed twice using MEM/PS, supplemented with 0.5% BSA and 0.1% sodium azide. nfler which they were analyzed using a flowcytomcter (FACScan, Becton-Dickinson, USA). Proliferative responses of PMNC were tested by incubating 0.2*10c PMNC with different mitogens.
396
Mitogens used were phyto heamagglutinin (PHA) in two concentrations (40 and 60 Mg/ml, Wellcome diagnostics, the Netherlands); pokeweed mitogen (PVVM; 30 Mg/ml; Gibco), and an antigen cocktail (AG) containing varidase (40 lU/ml), tuberculin PPD (40 Mg/ml; Statens Serum Institute, Denmark), tetanus toxoid (10 LF/ml; RIVM, the Netherlands), and Candida (200 Mg/ml; Haarlems Allergenen Laboratorium, the Netherlands). Cells were cultured in 96 Uwell microtiter plates (NUNC, Denmark) using RPMI 1640 (Gibco) supplemented with 20% heat inactivated pooled human serum, penicillin (100 IU/ml), streptomycin (100 jig/ml), and 1-glutamin (2 HIM). After 5 days 3H-thymidine was added and the plates were harvested after an additional 16 to 18 hours of culturing on appropriate filters. Radioactive counts per minute (cpm) were determined after the addition of 3 ml of scintillation fluid in a LSC counter. Total numbers of leukocytes, monocytes, and lymphocytes per sample were determined in whole blood using an automated closed tube sampler (Technicon H-l System) at the Department of Heamatology, University Hospital Utrecht, the Netherlands.
Biobehavioral Data On the day before the experiment, the General Health Questionnaire, developed by Ursin and coworkers (27) was administered. In a first part of this checklist, respondents can report amount and severity of health complaints during the month preceding the experiment. Only those scores that were considered relevant to the immune system (28) were used. From a second part of the same list, information on daily alcohol and coffee intake, smoking, physical exercise, averaged nightly sleeping hours, subjective fitness and health was obtained. In addition, the numbers of days of sick leave, days of illness, and visits to a physician over the last year were asked. Report of chronic disease and regular medicine intake was requested.
Assessment of Mood Levels Visual analogue scales were used to assess initial mood levels and mood changes during the experiment. The method has been used by several Scandinavian researchers (29-31). Each subject rated his perceived 'boredom,' 'ability to concentrate.' 'ef-
Psychosomatic Medicine 54:394-406 (1992)
STRESS EFFECTS O N BLOOD CELLS fort,' 'impatience,' 'interest,' 'irritation,' 'tenseness.' 'tiredness,' 'cheerfulness,' and 'sadness.' Mood was assessed three times during the experiment, just before blood sampling on Tl, T2 and, in a smaller subsample, T4.
Statistical Analyses The effect of the experimental manipulation on the immunological parameters was tested as interactions between stress condition (experimental versus control) and time (blood sampling moments), with analysis of variance (ANOVA) for repeated measures (four). Averaged tests of significance were used (equivalent to the split-plot approach), with Greenhouse-Geisser correction of degrees of freedom. Intergroup differences in linear and curvilinear trends were tested in these analyses with the use of polynomial contrasts. Student t tests were used to test for differences between the experimental and control group with respect to baseline of immunological parameters, initial mood, and health variables. Chi-square tests were applied when variables were noncontinuous, (like smokers versus nonsmokers). Despite random assignment, some differences existed on these variables between the groups. To test whether different immunological reaction patterns for the two conditions could be attributed to these differences, the ANOVA analyses were performed once more, using these variables as covariates. Differences between groups as to mood change scores were also tested as a two-way interaction ANOVA for repeated measures. Only the two first measurements (before and after the stressor) were used for the present purposes. For all the statistical analyses SPSS/PC+ subprograms were used (32). Differences in reported degrees of freedom are caused by missing values.
RESULTS
Biobehavioral Variables and Initial Mood Levels No differences on initial mood levels existed between the experimental and control groups. Despite random assignment to the conditions, a few statistically Psychosomatic Medicine 54:394-406 (1992)
significant differences (p < 0.05) between the experimental and control group were, however, observed in the biobehavioral variables. People of the experimental group were, on average, somewhat older than those of the control group (42.0 years versus 38.7 years]. Moreover, they reported less symptoms of colds/influenza and coughing/bronchitis over the last month, and less days of sick leave over the last year. Furthermore, experimental subjects reported fewer average daily coffee drinking than control subjects. Since these factors may be of importance with respect to the outcome of the immunological analyses (28), they were used as covariates in the analyses of covariance of immunological changes.
Stressor-induced Changes in Mood In Figure 1 mood changes are depicted as poststress (T2) values minus prestress values (Tl). ANOVA for repeated measures, with interaction between stress-condition and time, showed that most of the differences observed between the experimental and control group are significant. The results show that, in contrast to the control subjects, the experimental subjects report increases in 'effort' (F(l,71) = 30.84, p < 0.001), 'irritation' (NS), 'tenseness' (F(l,74) = 13.76, p < 0.001), 'impatience' (NS), 'concentration' (F(l,74) = 15.98, p < 0.001) and 'interest' (F(l,74) = 8.16, p < 0.01) during the stressor. Measurements at T4, which were only performed in a smaller subsample (about two thirds of the main sample), showed that all the moods that were changed at T2 had returned to baseline values at T4. Taking these data together, we can conclude that the experimental situation in397
J. F. BROSSCHOT et al. MOOD CHANGES (after - before stressor) Effort
m
m
m
m
^
—
< H H »
Irritation Tenseness
MfflgJ™
Impatience Concentration
mmmm
Boredom Interest
*** 8gjgg})SS38Sg
mmm•
Tiredness Cheerfulness
i
1
Sadness
experimental group (N=5 1)
control group (N=25)
Fig. 1. Differences in mood changes (before (Tl) minus after (T2) the stressor) between the experimental group and the control group. Positive numbers signify an increase in mood; negative numbers, a decrease in mood. Indicated are significances of ANOVA with condition x time: ***p < 0.001; "••p < 0.01.
duces a mild temporarily psychological arousal and distress.
Stressor-Induced Changes in Peripheral Blood Leukocytes, Monocytes, and Lymphocyte Subset Distribution Repeated measures ANOVAs over the four blood samples (see Scheme 1) show significant 2-way interactions (Time X Condition) (see Table 1). Experimental and control groups have different changes over time for numbers of lymphocytes, 398
CD8+ cells, CD57+ cells, CD16+ cells, and the CD4+/CD8+ ratio. The shift in the CD4+/CD8+ ratio appears to be mainly dependent on the increase in CD8+ cells. No significant interactions of Time X Condition were found for leukocytes, monocytes, CD3+, CD4+, and HLADR"1". The trends in the parameters that show significant responses to the stressor are illustrated in Figure 2. The changes in distribution of cell types are expressed as percentages of the individual baseline level (at Tl). In this way it can be seen that the mean number of CD57+ cells in the experimental group increased about 26%, whereas a decrease of about 11% can be observed in the control group. At time T3 the numbers of these cells return to baseline values. After T3, both groups diverge again in opposite directions. This pattern, which can be expressed as a cubic trend, is also visible in the responses of the CD16+ cells, the CD8+ cells, the lymphocyte numbers, and the CD4+/ CD8+ ratio. Polynomial contrast analyses reveal that the conditions differed significantly for this cubic trend with regard to CD57+, CD16+, and CD8+, and also for the linear and quadratic trends with regard to the same celltypes (p = < 0.01 at least). The lymphocyte numbers differed only for the linear trend. Since the markers CD8 and CD57 are shared by some T-suppressor/cytotoxic cells and NK cells, additional analyses were performed on the CD8+CD57+ cells (double positive cells), and on the remaining populations of CD8+CD57~ and CD8~CD57+ cells. The results are shown in Table 2 (and Figure 3). As demonstrated in Table 2, the CD8+CD57+ double positive cells are responsible for the change in the CD8+ population. The increase in CD8+CD57~ is not significant, while the trend in CD57+ (CD8~CD57+) is now Psychosomatic Medicine 54:394-406 (1992)
STRESS EFFECTS O N BLOOD CELLS Scheme 1. Overview of the experimental session.
Minutes 0
30
RESTl (neutral video)
60
45
PROBLEM SOLVING PERIOD
EXPLICATION PERIOD T2
Tl
even more 'peaked' (see Figure 3), and highly significant. The difference in the CD4+/CD8+ ratio (not shown) is also not significant anymore when CD8+CD57~ cells were used. Hence, the effect of the stressor appears to be most pronounced on cells with the NK cell phenotype expressing the CD57 marker, or expressing CD57 together with CD8.
Analyses of Covariance For CD57+ and CD8+57+ the baseline (Tl) values turned out to be significantly higher in the experimental group (p < 0.05). Therefore, ANOVA analyses were performed over T2, T3, and T4, using Tl as a covariate. This procedure, however, did not affect the results, meaning that the difference in reactivity patterns could not be attributed to baseline differences. Using as covariates the five potential confounders, which differ between the experimental and control groups (age, colds/influenza and coughing/bronchitis over the last month, days of sick leave during last year, daily coffee drinking, and initial "interest"), yielded no substantially different results. Hence, accidental differences in relevant variables between the experimental and control group could not explain the differences in reactivity between the experimental group and the control group. Psychosomatic Medicine 54:394-406 (1992)
105
75
REST 2 (neutral video) T3
T4
Proliferative Responses No significant interactions between time and condition were observed for the proliferative responses. A baseline difference existed between the conditions with respect to PHA-stimulated proliferation (40 jtg/ml). Additional covariate analysis, however, did not change the finding that there was no effect for this functional parameter.
DISCUSSION
In the present study it is shown that a short, mild psychological stressor can influence the distribution patterns of lymphocytes and NK cells in peripheral blood. No differences were observed when the effect of the stressor was tested on proliferative responses of T and B lymphocytes. The straining effect of the stressor was verified by significant changes in measures of reported effort and distress in the experimental subjects. We controlled for potential biobehavioral confounders (age, health symptoms, and health behavior variables). Where 'accidental' differences existed between the conditions, the analyses were corrected by using the potentially confounding variables as covariates in the analyses of variance. It was clearly shown that the results could not be attributed to these variables. Our results confirm and expand on pre399
J. F. BROSSCHOT et al. 9
TABLE 1. Cell Counts (x lO /Liter) on Times Tl, T2, T3, and T4 Exp
Leukocytes
Lymphocytes
Monocytes
CD3 +
CD4 +
CD8 +
CD4 + /CD8 + RATIO
CD57 +
CD16 +
HLADR+
T1 T2 T3 T4 Tl T2 T3 T4 Tl T2 T3 T4 Tl T2 T3 T4 T1 T2 T3 T4 T1 T2 T3 T4 Tl T2 T3 T4 Tl T2 T3 T4 T1 T2 T3 T4 Tl T2 T3 T4
Con
F (time
Mean
Std
Mean
Std
5.32 5.57 5.53 5.71 1.61 1.66 1.57 1.57 0.400 0.399 0.401 0.393 1.11 1.10 1.08 1 12 0.717 0.697 0.708 0.765 0.529 0.568 0 503 0.470 1.51 1.41 1.59 1.84 0.299 0.354 0.275 0.233 0.236 0 262 0.193 0.159 0.322 0.341 0.344 0.348
1.7 1.8 1.9 2.0 0.46 0.49 0.45 0.49 0.16 0.16 0.18 0.17 0.32 0.35 0.34 0.35 0.22 0.23 0.23 0.25 0.21 0.24 021 0.21 0.10 0.11 0.12 0.13 0.17 0.19 0.14 0.13 0.16 0.16 0.12 0.12 0.13 0.15 0.14 0.15
5.44 5.51 5.56 5.81 1.54 1.53 1.55 1.64 0.360 0.355 0.372 0.401 1.05 1.04 1.06 1.11 0.714 0.719 0.726 0.769 0.470 0.435 0.453 0.472 1.66 1.75 1.75 1.75 0.217 0.185 0.197 0.210 0.171 0 146 0.147 0.147 0.335 0.345 0.349 0.368
1.8 1.9 1.8 1.9 0.43 0.41 0.40 0.43 0.16 0.17 0.17 0.18 0.32 0.34 0.30 0.34 0.25 0.23 0.22 0.27 0.17 0.13 0.15 0.16 0.12 0.10 0.11 0.11 0.11 0.09 0.11 0.10 0.10 0.10 0.08 0.09 0.14 0.15 0.13 0.15
X cond.) 1.18
3,249
11.50**
3,243
2.83
3,189
2.25
3,234
1.17
3,234
17.09**
3,234
12.84"
3,234
18 17"
3,234
6.68"
3,135"
0.59
3,234
'' Degrees of freedom before Creenhouse-Ceisser correction; significancies shown are after correction. h CDl 6+ was only determined in a part of the sample, hence the smaller DF here. " * p < 0.001. "p
