Journal of Behavioral Medicine, Vol. 9, No. 1, 1986
Modulation of Cellular Immunity in Medical Students
Janice K. Kiecolt-Glaser, l's Ronald Glaser, 2'3
Eric C. Strain, 1 Julie C. Stout, 2 Kathleen L. Tarr, 2
Jane E. Holliday, 2 and Carl E. Speicher 4
Accepted for publication: December 8, 1984
This study assessed the psychosocial modulation of cellular immunity in 34
medical-student volunteers. The first blood sample was obtained 1 month
before examinations, and the second on the day of examinations. There were
significant declines in the percentage of helper~inducer T- lymphocytes, in
the helper/inducer-suppressor/cytotoxic-cell ratio, and in natural killer-cell
activity in the blood samples obtained on the day of examinations. Half of
the subjects were randomly assigned to a relaxation group which met be-
tween sample points; the frequency of relaxation practice was a significant
predictor of the percentages of helper~inducer cells in the examination sam-
ple. Three biochemical nutritional assays (albumin, transferrin, and total
iron-binding protein) were within normal limits on both samples. Data from
the Brief Syrup tom In ven tory sh o wed significantly increased global self-rated
distress associated with examinations in the no-intervention group, compared
to nonsignificant change in the relaxation group. Clinical and theoretical im-
plications are discussed.
KEY WORDS: stress; psychoimmunology; relaxation; cellular immunity.
This study was supported in part by funds from the State of Ohio Department of Mental Health,
the Bremer Foundation, and the Samuel J. Roessler Fund and by Ohio State University Com-
prehensive Cancer Center Grant CA-16058-09 from the National Cancer Institute, National
Institutes of Health.
~Department of Psychiatry, Ohio State University College of Medicine, Columbus, Ohio 43210.
2Department of Medical Microbiology and Immunology, Ohio State University College of
Medicine, Columbus, Ohio 43210.
3Comprehensive Cancer Center, Ohio State University College of Medicine, Columbus, Ohio
4Department of Pathology, Ohio State University College of Medicine, Columbus, Ohio 43210.
5To whom correspondence should be addressed at Department of Psychiatry, Ohio State Univer-
sity College of Medicine, 473 West 12th Avenue, Columbus, Ohio 43210.
o16o-7715/86/o2oo-ooo55o5.oo/o y 1986 Plenum Publishing Corporation
Kiecolt-Glaser, Glaser, Strain, Stout, Tarr, Holliday, and Speicher
There is growing evidence supporting the importance of relatively minor
life events as important moderators of physical and psychological health;
minor events appear to account for more of the variance in somatic and
psychological symptoms than major life events (DeLongis et al., 1982; Kan-
ner et al., 1981), even after controlling for initial psychological distress
(Monroe, 1983). Recent research suggests that one critical mediating link be-
tween both major and minor stressful events and physical health may be the
immune response, the body's defense against infectious and malignant disease.
Research from our laboratory has focused on the impact of relatively
commonplace events on immunity. We have found significant changes in
a number of different facets of the cellular immune response, in conjunc-
tion with relatively mild stressful events such as examinations. For example,
we found a significant decrease in natural killer (NK)-cell activity in
medical-student blood samples obtained during examinations, in comparison
to baseline samples obtained 1 month earlier. Moreover, lonelier students
had significantly poorer NK-cell activity (Kiecolt-Glaser et al., 1984a). These
data may have important implications for health, since recent evidence strong-
ly suggests an association between cancer incidence and NK-cell activity in
humans (Strayer et al., 1984). Consistent with this, data of Levy et al. (1984)
suggest that psychological factors are related to NK-cell activity and prog-
nosis in women with breast cancer.
Other immune functions were also affected in this medical-student
population; there were significantly higher antibody titers to three
herpesviruses, Epstein-Barr virus (EBV), herpes simplex virus (HSV), and
cytomegalovirus (CMV), in blood samples obtained from medical students
during final examinations, in comparison to samples obtained after their
return from summer vacation (Glaser et al., 1985). Higher antibody titers
suggest less control over latent virus by the cellular immune response, while
reestablishment of control over virus replication and latency is ultimately
followed by a concomitant drop in antibody titers (Glaser and Gottlieb-
Based on these data and others (Kiecolt-Glaser and Greenburg, 1984;
Kiecolt-Glaser et al., 1984b,c), we reasoned that interventions which reduc-
ed distress and/or loneliness might enhance cellular immunity. In a subse-
quent study, 45 geriatric residents of independent living facilities were
randomly assigned to one of three protocols: (1) progressive relaxation train-
ing, (2) social contact, or (3) no contact. Subjects in the relaxation and social-
contact conditions were seen individually three times a week for 1 month.
At the end of the intervention period, the relaxation group showed a signifi-
cant increase in NK-cell activity and a significant decrease in antibody levels
to HSV with a concomitant decrease in self-reported distress, in contrast to
nonsignificant changes in the other two groups. Only the decrease in HSV
Psychosocial Modulation of Immunity
antibody titers persisted at the l-month follow-up, with little relaxation prac-
tice reported by subjects past the end of the intervention (Kiecolt-Glaser et
In this study, 34 first-year medical-student volunteers had blood drawn
twice, with the first (baseline) draw occurring 1 month before examinations
and the second during examinations. Half of the subjects were randomly
assigned to a hypnotic/relaxation group which met in the interval between
bleeds. Subjects provided self-report data each time blood was drawn.
The study was designed to meet four objectives. The investigation of
possible changes in the percentages of helper/inducer and suppressor/cytotox-
ic T lymphocytes and their ratio was of primary interest for this study, given
previous data from both our laboratory and others documenting the impact
of acute stressors on cellular immunity. Significant stress-related changes in
the percentages of helper/inducer and suppressor/cytotoxic cells could have
important consequences for health. Large reductions in the percentage of
helper/inducer cells can produce immunodeficiency, while severe disturbances
in the percentage of suppressor/cytotoxic cells are associated with autoim-
mune disorders (Reinherz and Schlossman, 1980).
Also of interest was the continued exploration of the effects of relaxa-
tion on immunity. We were particularly interested in testing the hypothesis
that relaxation would buffer stress-related changes in cellular immunity, with
the magnitude of the relationship dependent on the frequency of relaxation
Nutritional measures were included to provide information on possi-
ble changes in nutritional status. There are well-documented impairments
in various aspects of immunity found in undernourished individuals, and
moderate to severe protein-caloric malnutrition is associated with increased
frequency and severity of infection (Chandra, 1977). Since appetite and sleep
disturbances are found more frequently in distressed populations, nutritional
markers were included to assess the possibility that any changes in immunity
simply reflected underlying nutritional changes. Three plasma protein markers
were used, total iron-binding protein (TIBC), albumin, and transferrin.
Finally, NK-cell activity was assayed, in an attempt to replicate previous-
ly described stress-related changes (Kiecolt-Glaser et al., 1984a). The possi-
ble moderation of stress-related changes in this parameter by relaxation was
also of interest.
Subjects and Timing of Samples
The subjects were 35 first-year medical students in The Ohio State
University College of Medicine who volunteered for a research project on
Kiecolt-Glaser, Glascr, Strain, Stout, Tarr, Holliday, and Speicher
stress and the immune response. One man did not return for the second blood
draw. The average age of the remaining 22 men and 12 women was 23.5 years.
In the memorandum which described the study and solicited volunteers,
the students were told that half of the group would be assigned to a hyp-
notic/relaxation group which would meet during lunch hours; volunteers were
asked to be willing in principle to attend at least five sessions in the 2.5 weeks
before the second blood draw, with the option of discontinuing at any time
they wished. They were also told that those subjects who were not assigned
to the relaxation group would have the same opportunity the following
month. The only other inducement for participation was confidential in-
dividual and group feedback on changes in their immune response.
The first blood sample was drawn 1 month before examinations in mid-
November, 1 month after the first examination series. The second blood sam-
ple was obtained on the final day of the 3-day mid-December examination
The Brief Symptom Inventory (BSI; Derogatis and Spencer, 1982), ad-
ministered both times blood was drawn, provided information on changes
in global distress, as well as nine symptom dimensions. Subjects were asked
to rate the amount of distress in the last 2 weeks which was associated with
each of the 53 items from 0 (not at all) to 4 (extremely). The BSI also pro-
vided information on psychological symptomatology which might have con-
traindicated participation in the relaxation group, e.g., paranoid or psychotic
symptomatology; no subject reported symptoms which made such participa-
The brief state portion of the loneliness scale developed by Shaver et
al.(1984) was administered at both sample points in order to provide infor-
mation on changes in loneliness between the relaxation and the no-treatment
groups. These data were of interest because of the possibility that the in-
tervention might produce changes in loneliness as a result of potentially in-
creased interpersonal contact for the experimental subjects.
The Shaver et al., (1984) state loneliness scale was created by combining
eight items from the UCLA Loneliness Scale (Russell et al., 1980) and three
items from the NYU Loneliness Scale (Rubenstein and Shaver, 1980). Both
of these scales have adequate reliability and validity, which were established
through a number of studies (e.g., Russell, 1982). The resulting state
loneliness scale has coefficient alphas between 0.85 and 0.90 across various
samples (Shaver et al., 1984), correlates significantly with the UCLA
Loneliness Scale, and shows the expected changes when used with
undergraduates during the transition into college.
Psychosocial Modulation of Immunity 9
At the baseline sample point in November, subjects were asked to
describe any chronic health problems. They were also asked the type and
frequency of any regular self-relaxation practice (meditation, self-hypnosis,
On both occasions when blood was drawn, subjects were asked to
describe any acute health problems in the last month, medications taken in
the last 2 weeks, and caffeine intake in the last 48 hr. They were also asked
to specify their present weight, and the amount of weight change, if any,
during the last two weeks.
Questions added only at the second sample point included the number
of times subjects had practiced any type of relaxation in the last month (out-
side of group practice), the amount of sleep in the last 72 hr compared to
normal requirements, the number of hours of vigorous physical activity in
the last month, the quality of sleep in the last month, and any recent major
Two methods were used to evaluate the possibility that this volunteer sub-
ject sample might differ from their classmates in some important dimen-
sion(s), e.g., poorer students might volunteer in disproportionate numbers
because of concerns about the possible physiological consequences of
academic stress. The informed consent form included permission to check
grades, so that the academic performance of participants could be compared
with the nonparticipants (for whom grades were available without identify-
ing information). In addition, all first-year medical students had taken the
Sixteen Personality Factor Test (16PF) during orientation 2 months previous-
ly, and the scores of participants were compared to those of nonparticipants.
The first session, attended by all assigned subjects, provided an over-
view of hypnosis and its similarities to other forms of relaxation, as well as
an introductory group induction. The format of the remaining sessions was
quite similar, since it was anticipated that attendance would vary across the
10 possible sessions. Each session began with the same standard deepening
exercises. The middle portion of the session was varied and included self-
h~cpnosis, progressive relaxation, autogenic training, and various imagery ex-
ercises. Suggestions made near the end of each session included greater relax-
ation throughout the day, enhanced comprehension and retention of academic
material, and improved study habits. The subjects were strongly encouraged
to practice relaxation outside of the group sessions. All of the group ses=
sions were led by a clinical psychologist with substantial prior experience in
various forms of relaxation training.
Kiecolt-Giaser, Glaser, Strain, Stout, Tart, Holliday, and Speicher
At the end of each session, students rated the degree of relaxation
achieved during the session from 1 (none) to 9 (completely relaxed); they
were also asked for feedback on the session. The feedback sheets included
a space for their subject number, thus providing a record of attendance and
The group was designed to introduce the subjects to a number of dif-
ferent kinds of relaxation exercises, so that they might find the method(s)
which best suited their own needs. The hypnotic format was chosen because
the medical students were most familiar with it, and it was therefore most
likely to elicit volunteers. No single relaxation technique was promoted to
the exclusion of others, since the literature does not suggest that any one form
of relaxation produces more reliable physiological effects than any other (e.g.,
Benson et al., 1978; Edmonston, 1981; English and Baker, 1983; Lehrer et
al., 1980). Discussion during the relaxation sessions emphasized the com-
monalities among the various relaxation exercises. Sessions were 35-45 min
T-Lymphocyte Subset Assay
The percentage of helper/inducer T cells, suppressor/cytotoxic T cells,
and helper/suppressor-cell ratios were determined using the monoclonal an-
tibodies OKT-4 and OKT-8, respectively (Ortho), as previously described
(Moll et al., 1982).
Briefly, lymphocytes isolated on Hypaque-Ficoll gradients were wash-
ed with trypsin diluent, then resuspended in complete RPMI 1640 medium
supplemented with 20% fetal bovine serum. Monocytes were removed by
placing the cell suspensions in plastic tissue culture flasks and incubating at
37~ in a CO2 incubator for 1-3 hr. This adsorption was performed in order
to reduce cell clumping and, therefore, to increase the accuracy of the counts.
The nonadeherent cells were washed off and used to determine the percen-
tage of T-cell subsets. Lymphocytes (106) were incubated in 0.01 ml of either
OKT-4 or OKT-8 monoclonal antibody for 30 min on ice. Cells were wash-
ed with cold RPMI 1640 medium, resuspended in goat anti-mouse IgG con-
jugated to fluorescein isothiocyanate (Cappel Laboratories), and incubated
for an additional 30 min on ice. The cells were washed and assayed, using
an Ortho System 50 fluorescence-activated cell sorter (FACS). The normal
range for OKT-4 (helper/inducer)-positive cells is approximately 33 to 53 %,
and the normal range for OKT-8 (suppressor/cytotoxic)-positive cells is 11
to 30%. The normal range for OKT-4/OKT-8 cell ratios is 1.1.-3.5.
Psychosocial Modulation of Immunity
A microtiter chromium-51 release cytotoxicity assay was used to deter-
mine NK-cell activity, as previously described (Kiecolt-Glaser et al., 1984c).
Briefly, mononuclear cells were obtained by Ficoll-Hypaque separation of
whole blood. The target cells used in the assay were K-562 cells, a myeloid
Triplicate (0.1-ml) aliquots of lymphocyte and labeled target-cell suspen-
sions were placed in wells of 96-well plates (Limbro, Conn) resulting in
effector- to target-cell ratios of 40:1, 20:1, and 10:1. In addition, triplicate
wells with target cells and medium only and with target cells and detergent
(1% socium dodecyl sulfate) were prepared in order to determine spon-
taneously released radioactivity and maximal lysis. Supernates were counted
in a Beckman 9000 gamma counter. Results are reported as the percentage
lysis using the formula
experimental cpm - spontaneous release cpm
maximal cpm - spontaneous release
% lysis = x 100
Using this procedure for the measurement of NK-cell lysis, differences
are expected among the three effector:target-cell ratios, with larger ratios
producing more lysis. The 40:1 ratio is the most sensitive, and the 10:1 the
least, based on the larger supply of effector cells in the former.
Protein assays provide better information on global nutritional status
than those for carbohydrates and fats, since the former have varied nutri-
tional building blocks, as well as very complex synthetic pathways. Different
protein markers were used because of the differences in their half-lives; the
half-life of albumin is 2 to 3 weeks, in comparison to 8 days for total iron-
binding protein (TIBC) and transferrin. Transferrin and TIBC assays are
different ways to measure the same protein.
The procedure use to measure albumin is an adaptation of the brom-
cresol green dye-binding method of Rodkey (1965), later modified by Doumas
(1971). This procedure is recognized as a particularly good procedure as com-
pared to other dye-binding techniques because of its specificity and freedom
To assay TIBC, ferric chloride was added to each plasma sample to
saturate transferrin. The excess iron was removed with magnesium carbonate.
12 Kiecolt-Glaser, Glaser, Strain, Stout, Tarr, Holliday, and Speicher
After centrifugation the supernatant was measured for plasma iron by com-
plexing the ferrous ions with ferrozine in a sodium acetate buffer solution.
The absorbance of the violet-red complex was read at 560 nm in a spec-
trophotometer (Tietz, 1976).
Transferrin is an iron-transporting protein. The concentration in plasma
is affected by the dietary intake of iron. Nutritionally deficient but calorie-
rich diets are generally lacking in iron, and as a result, plasma iron levels
tend to be low, and transferrin levels high. It has been shown that the estima-
tion of transferrin levels may be used to assess the effectiveness of the total
parenteral nutrition (Keyser, 1979).
A rate nephelometry procedure using a Beckman human im-
munoglobulin reagent kit and a Beckman immunochemistry analyzer system
was used to analyze transferrin levels. Antibody to human transferrin was
used in the assay, in which the peak rate signal caused by the antigen/an-
tibody complex is proportional to the increase in light scatter which is read
by the instrument (Buffone, 1980).
The data were analyzed using a repeated-measures analysis of variance
design, with one between-subjects variable (group assignment) and one within-
subjects variable (change from the first to the second sample). Additional
multiple regression analyses within the relaxation group were used to evaluate
the hypothesis that the frequency of relaxation practice was a significant
predictor of immunologic competence, after controlling for baseline levels.
Baseline Comparisons and Information
Baseline comparisons revealed no significant pretreatment differences
between the intervention and the no-treatment group on the immunological
or nutritional measures or in caffeine intake. Similarly, the groups did not
differ significantly either in self-reported distress on the BSI scales or in state
No subject reported any relaxation practice at baseline. No subject
reported any acute or chronic health problems which might have a signifi-
cant immunological component, aside from mild allergies. No subject reported
taking any medication which had known immunological consequences.
The BSI data are presented in Table I. Post hoc analyses (Waller and
Duncan, 1969) of the significant interactions showed nonsignificant change
Psychosoeial Modulation of Immunity 13
Table I. BSI Means and Standard Deviations for Relaxa-
tion and Control Groups at Baseline and During Ex-
BSI scale Relaxation Control
*Interaction between group and trial variables, P
**Change over sample points, P < 0.05.
***Change over sample points, P < 0.01.
Kiecolt-G|aser, Glaser, Strain, Stout, Tarr, Holliday, and Speicher
within the relaxation group, compared to significant increases in anxiety (P
< 0.01), in obsessive-compulsive symptomatology (P < 0.05), and in the
global severity index (P < 0.05) in the no-intervention group.
There was a significant decrease in state loneliness from the first to the
second bleed IF(l, 32) = 13.22, P < 0.001). The mean at baseline was 29.77,
compared to 26.02 during examinations. Neither the main effect for group
membership nor the interaction between group membership and change across
samples was significant (Fs < 1).
One subject reported a 2-1b weight loss in the 2 weeks before the baseline
sample point, while three subjects reported weight gains of 2 to 4 lb. Four
subjects reported weight losses of 1 to 4 pounds in the 2 weeks prior to ex-
aminations, while eight subjects reported weight gains of 1 to 6 lb.
The students were asked how many hours they had slept in the last 3
days before the second blood draw, in comparison to their normal sleep re-
quirements. One subject reported sleeping more than usual in the 72 hr before
the second blood draw. Five subjects reported normal amounts of sleep, while
the remaining 28 subjects reported an average deficit of 5.61 hr (SD = 3.90).
The size of the sleep deficit was not significantly correlated with any of the
immunological measures taken during examinations.
There was not a significant difference in the quality of sleep between
the intervention and the no-treatment groups, and they did not differ in the
size of any sleep deficits (Fs < 1). The number of hours of vigorous physical
activity in the previous 3 weeks also failed to distinguish significantly be-
tween groups (F < 1), with a group mean of 6.00 (SD = 4.21).
Caffeine usage in the 48 hr before the blood draw increased significantly
from the first to the second sample [F(1, 32) = 4.89, P < 0.05] but did not
differ as a function of group membership or the interaction between the in-
dependent variables (Fs < 1). The mean intake (equivalent to the number
of cups of coffee) increased from 4.45 to 6.15. Caffeine intake was not
significantly correlated with any of the immunological measures on either
the first or the second sample, however.
In order to evaluate the effects of the relaxation intervention on
academic performance, additional comparisons were made within the ex-
perimental group using the mean standard scores (across three disciplines)
on the first and second tests in a 2 • 2 repeated-measures analysis of variance.
There were no significant differences as a function of group membership (F
< 1), change from the first to the second test (F < 1), or the interaction
between group membership and change from the first to the second test [F(1,
32) = 1.80].
The academic performances of participants and nonparticipants were
not significantly different. The participants' mean standard score (across three
disciplines) on the first examination did not differ from that of students who
did not participate in this research project (F < 1).
Psychosocial Modulation of Immunity
The scores of participants and nonparticipants were compared on each
of the 16PF standard scales. There were no significant differences as a func-
tion of experimental participation.
The percentage of helper/inducer cells decreased significantly in the ex-
amination sample compared to the baseline sample [F(1, 29) = 10.27, P <
0.003], as shown on the left side of Fig. 1. There was not a significant main
effect as a function of group membership (F< 1), and the interaction bet-
ween group membership and change over time did not reach significance [F(1,
29) = 3.40].
The percentage of suppressor/cytotoxic cells did not change significantly
[F(1, 29) = 3.38]. The percentage of suppressor cells was 20% at baseline
and 13.7% during examinations. There were no significant effects as a func-
tion of group membership (F< 1) or the interaction between group member-
ship and change over trials [F(1, 29) = 1.79].
The helper/suppressor-cell ratio decreased significantly from the first
to the second sample [F(1, 29) -- 5.57, P < 0.03], as shown in the right
side of Fig. 1. There were no significant effects attributable to either group
membership or the interaction between the two independent variables (Fs
< 1). The normal ranges for the percentages of helper and suppressor cells
are important in this context. Using these monoclonals (OKT-4 and OKT-8),
the normal percentage of helper cells is from 34 to 54~
for suppressor cells is 20 to 37%, and the normal range for the helper/sup-
pressor ratio is 1.1 to 3.7.
After obtaining the low values for the medical students at the first sam-
ple point, additional assays were made using blood from 11 university staff
The normal range
40- _0 3
BASELINE EXAM BASELINE EXAM
Fig. 1. Means ( _+_ SE) at baseline and during examinations for the
percentage of helper cells (left) and the helper/suppressor-cell ratio
16 Kiecolt-Glaser, Glaser, Strain, Stout, Tarr, Holliday, and Speicher
NK EFFECTOR: TARGET CELL RATIO
Fig. 2. Means (_+ SE) for the three NK-cell effector- to target-
cell ratios at baseline and during examiations.
members to assess the possibility that the low values reflected artifacts in
the laboratory technique and/or in the monoclonal stock. These assays were
run by the same technicians, and the monoclonal antibodies came from the
same stock. The percentages of helper and suppressor cells from these sub-
jects were well within normal limits, a finding not consistent with a laboratory-
induced artifact(s). The very low values found in medical-student samples
during examinations compared to their already low baseline values further
raised the possibility of stress-related impairments at baseline.
The NK-cell analysis included the three effector- to target-cell ratios
as an additional variable. There was a significant decrease in NK-cell lysis
of target cells as shown in Fig. 2 [F(1, 32) = 11.07, P < 0.003]. There was
also the expected significant main effect for differences among the three NK
effector- to target-cell ratios, [F(2, 64) = 45.05, P < 0.0001]. There was
not a significant main effect attributable to group assignment, and the in-
teraction between change over trials and group membership was not signifi-
cant (Fs < 1).
Multiple regression analyses were used to evaluate the hypothesis that
the frequency of relaxation practice was a significant moderator variable in
the stress-related immunological changes within the relaxation group; no sub-
ject in the no-treatment group reported any relaxation practice. There were
large differences in the frequency of relaxation practice within the group;
the sum of group attendance and home practice sessions ranged from 5 to
50, with a mean of 12.07 and a standard deviation of 7.19. The frequency
of relaxation practice was a significant predictor of the precentage of
helper/inducer cells in the examination period after correcting for baseline
Psychosocial Modulation of Immunity 17
Table II. Regression Analysis for Relaxation Frequency and Immunocompetence
R 2 (Step 2 with
R 2 (Step 1 with
correlation R 2 change
0.05 0.43 0.38* 0.64
0.03 0.18 0.15 0.34
0.03 0.16 0.13 0.33
*P < 0.01.
levels; more frequent practice was clearly associated with higher
helper/inducer-cell percentages, as shown in Table II. In contrast, the fre-
quency of practice was not a significant predictor of the NK-cell activity.
Nutritional data are shown in Table III. Neither albumin nor TIBC
plasma levels showed significant changes over time, and neither appeared
to be affected by the relaxation intervention, with all Fs < 1. Plasma transfer-
rin levels, in contrast, decreased significantly from the first to the second
sample [F(1, 30) = 8.89, P < 0.01], although the second sample was still
well within normal limits. There was not a significant difference between the
two groups, nor was the interaction significant (Fs < 1).
Significant declines were found in the percentage of helper/inducer cells,
helper/suppressor-cell ratios, and NK-cell activity in the examination sam-
Table II1. Normal Ranges, Means, and Standard Deviations for Plasma
Protein Markers at Baseline and During Examinations
Normal range 3.8-5.1
Baseline 4.90 (0.34) 363.33 (60.48) 346.57 (53.61)
Examination 4.91 (0.24) 370.40 (21.27) 318.73 (50.49)
18 Kiecoit-Glaser, Glaser, Strain, Stout, Tarr, Holliday, and Speicher
pie, compared to the baseline sample. The significant decrease in the percen-
tage of helper/inducer cells and in the helper/suppressor ratio may have
important implications for other aspects of the immune response. Both
helper/inducer and suppressor/cytotoxic cells perform extremely critical
regulatory functions for the immune response (Reinherz and Schlossman,
1980). Helper/inducer cells have an inducer function in the B-lymphocyte
proliferation and differentiation sequence which is necessary for the synthesis
of immunoglobulins; immunoglobulins provide a critical defense against in-
fectious agents. The optimal development of cytotoxicity in sup-
pressor/cytotoxic cells requires the presence of helper/inducer cells.
Helper/inducer cells also play an inducer role in the interactions between
T lymphocytes and macrophages. Therefore, if helper/inducer-cell function
is disrupted, immunodeficiency may result (Reinherz and Schlossman, 1980).
The finding that the frequency of relaxation practice was a significant
predictor of helper/inducer-cell percentages in the second sample suggests
that it may be possible to affect helper/inducer-cell percentages through com-
mon stress-reduction interventions. While further work is needed to confirm
these data, such effects may have important clinical implications for some
types of immunodeficient conditions.
The NK-cell data confirm our previous demonstration of stress-related
changes in medical students. The absence of a significant relationship bet-
ween the frequency of relaxation practice and NK-cell activity have been
reported in mice exposed to various physical stresses, with such declines oc-
curring within 24 to 48 hr of stressor exposure (Aarstad et al., 1983; Herber-
man, 1982). In contrast, similarly rapid changes of that magnitude in the
percentage of helper/inducer cells are unlikely. Thus, by the end of the 3-day
examination period, any moderating effects in NK-cell activity attributable
to relaxation may have been overshadowed by the acute stress of ex-
While plasma transferrin levels declined significantly in this well-
nourished population for unknown reasons, the final values were still well
within the normal range, as were those for albumin and TIBC. Since TIBC
measures the same protein as transferrin, it is unlikely that the decrease in
transferrin indicates nutritional deprivation. These data suggest that the
observed changes in cellular immunocompetence were not induced by poor
nutrition. The minor changes in sleep also appear unrelated to the large im-
The selection of medical students is based in large part on their ability
to reliably produce excellent examination scores. Despite their repeated suc-
cessful performances in this situation, we still find consistent and significant
changes in cellular immunity, both in these data and in our previous research
with this population (Glaser et al., 1985; Kiecolt-Glaser et al., 1984a,c). It
Psychosocial Modulation of Immunity
is not known whether this lack of adaptation occurs with repeated exposure
to commonplace stressors in other populations; it provides an important ques-
tion for future investigations, since it could have a significant long-term im-
pact on health.
Neither the medical students in this study nor the geriatric subjects in
our previous study (Kiecolt-Glaser et al., 1985a) were members of clinical
populations seeking treatment for stress-related disorders. Since the subjects
were not, as a group, significantly distressed or otherwise symptomatic before
the intervention, their level of involvement and incentive for intensive prac-
tice may have been less than those found in many clinical populations;
moreover, brief training in progressive relaxation is frequently not associated
with significant autonomic changes among less anxious subjects (Lehrer et
al., 1980). Therefore, these data may understate the magnitude of im-
munological change which is possible in more distressed populations.
These helper/inducer T-lymphocyte relaxation data have important
clinical implications. Added to the previous study using a geriatric popula-
tion (Kiecolt-Glaser et al., 1985a), these data provide further evidence that
relaxation may be able to enhance at least some components of cellular im-
munity and, thus, perhaps ultimately might be useful in influencing the in-
cidence and course of disease. Relaxation has already been shown to have
a number of positive physiological effects, including recent data which
demonstrate its impact on glucose tolerance in non-insulin-dependent diabetes
(Surwit and Feinglos, 1983).
The medical-student and geriatric relaxation data provide new evidence
for central nervous system medication of immune function. There are data
suggesting that acute stress is immunosuppressive (e.g., Ader, 1981; Coe
et al., 1985) and may be associated with important dysfunctions at the
molecular level in DNA repair (Kiecolt-Glaser et al., 1985b). In contrast, it
appears that relaxation may enhance immunity and provide a valuable "buf-
fer" for at least some immunologic functions during more stressful times.
Consistent with this, there has been considerable anecdotal and theoretical
speculation concerning the possible enhancement of health through the elicita-
tion of positive emotions (e.g., Cousins, 1976; Lazarus, 1982). It appears
that such speculation may have some preliminary empirical support.
We thank Allison Post and Carol Chung for laboratory assistance and
Marta Berger, Daniel Georges, Perry Nystrom, and Tamara Wood for ex-
perimental assistance. We also wish to thank Angie Grieshop, Connie Gillam,