1573

Abnormalities at Different Levels

of the Hypothalamic-Pituitary-Adrenocortical
Axis Early After Stroke
Tommy Olsson, MD; Niklas Marklund, MD; Yngve Gustafson, MD; and Birgitta Nasman, MD

Background and Purpose: Hypercortisolism is common in stroke patients. The aim of this study was to
investigate possible disturbances at different sites within the hypothalamic-pituitary-adrenal axis. We also
studied possible associations between hypercortisolism and clinical manifestations of brain dysfunction.
Methods: Patients with an acute ischemic stroke (n=16; mean±SD age, 71±11 years) were compared
with healthy elderly subjects (n=9). We performed a short adrenocorticotropic hormone (ACTH) test with
0.25 mg 1-24 ACTH injected intravenously and an overnight dexamethasone suppression test with 1 mg
dexamethasone given orally at 11 PM.
Results: Serum cortisol levels after dexamethasone at 8 AM were significantly higher in stroke patients
(p=0.003). The area under the curve for the cortisol response to ACTH was elevated in seven (47%) of
stroke patients, and the centered cumulative cortisol response was elevated in three (20%) patients. The
area under the curve response correlated significantly to the presence of an acute confusional state and
male sex in stroke patients (rs=0.63 and rs=0.62, respectively; p< 0.05), whereas the centered cumulative
cortisol response diminished with increasing age (rs=— 0.62;p<0.05). Postdexamethasone cortisol levels
were significantly correlated to the presence of an acute confusional state and to extensive limb paresis
(rs=0.66 and rs=0.62, respectively;p<0.05).
Conclusions: There are abnormalities in the cortisol axis both at the central level and at the adrenal level
early after stroke. Hypercortisolism is closely associated with cognitive disturbances and extensive motor
impairment. (Stroke 1992;23:1573-1576)
Downloaded from http://ahajournals.org by on January 18, 2021

KEY WORDS • cerebral ischemia • cerebrovascular disorders • cortisol

ism per se may contribute to cognitive disturbances.9

M any patients with acute stroke show a pro-

nounced hypercortisolism. 1 - 5 Increased
plasma and urinary cortisol levels are asso-
ciated with greater mortality and a poorer functional
outcome after stroke.1'24-6 This association has also
been demonstrated in other types of brain injury.7
Theoretically, dysregulation at several sites within the
hypothalamic-pituitary-adrenocortical axis may contrib-
ute. This includes an increased production rate of
cortisol, a change in metabolism and/or clearance rate
of cortisol, an increased sensitivity to stimulation of the
adrenal glands, and a decreased "shut-off" mechanism
of the cortisol axis.
In patients with chronic degenerative cerebral disease
such as Alzheimer's disease a correlation between cog-
nitive disturbances and hypercortisolism has been dem-
onstrated.8 It has been hypothesized that hypercortisol-
From the Departments of Internal Medicine (T.O., N.M.) and
Geriatric Medicine (Y.G., B.N.), Umea University Hospital,
UmeS, Sweden.
Supported by grants from King Gustav V's 80th Anniversary
Fund, Borgerskapets i UmeS Research Foundation, Fund for
Medical Research, Umea University, and 1987 Year's Foundation
for Stroke Research.
Address for correspondence: Tommy Olsson, MD, Department
of Internal Medicine, University Hospital, S-901 85 Umea,
Sweden.
Received May 13, 1992; final revision received July 16, 1992;
accepted August 7, 1992.
The acute confusional state commonly occurs in elderly
hospitalized patients10 and may be particularly frequent
among stroke patients.11
It is therefore of interest to study in more detail the
different sites within the cortisol axis where possible
abnormalities can lead to hypercortisolism early after
stroke. Furthermore, we wanted to examine whether
there was an association between cortisol levels and
acute confusional state and/or motor impairment in
these patients.
Subjects and Methods
Sixteen patients (11 men, five women; mean±SD age,
71 ±11 years) with an acute brain infarction were se-
lected for this study from our stroke unit. The median
delay from onset until admission was 11 hours (range,
1-100 hours). Based on clinical judgement and on the
results of computed tomographic (CT) scans, per-
formed in all patients, 14 patients had a supratentorial
brain infarction and two patients had a cerebellar
infarction. In the supratentorial brain infarction group,
eight patients had a probable nonembolic brain infarc-
tion, five patients had an embolic brain infarction, and
one patient had a lacunar infarction. Six of the patients
in this group had right-sided and eight had left-sided
brain lesions. These patients all had an extremity pare-
sis afflicting the contralateral arm or leg at admission.
Three patients with left-sided brain lesions also had a
 1574 Stroke Vol 23, No 11 November 1992

slight-to-moderate dysphasia at admission. Both pa- 1—p=0.003—i

tients with cerebellar infarctions had vertigo, one of So 700
them having a right-sided arm paresis as well. None of
•S- 600
the patients had a pronounced decrease in conscious-
ness, i.e., more than drowsiness, high fever (>38.5°C), | 500
renal failure (plasma creatinine level >200 /xmol/1),
known extensive weight loss or malnutrition, hypothy- | 400
roidism/hyperthyroidism, pituitary insufficiency, uncon-
trolled diabetes mellitus, obvious abstinence reactions fe 300
from alcohol or other central nervous stimulants, or CO
epilepsy. None was treated with medications known to g 200
interfere with the test results such as glucocorticoids, t
estrogens, anticonvulsants, high-dose benzodiazepines, 8 100
or ephedrine. Two of the patients had non-insulin-
dependent diabetes mellitus.
As control subjects, nine healthy elderly people were
I Controls Stroke
FIGURE 1. Scatterplot of serum cortisol levels at 8 AM after
selected from an ongoing study of hormone changes in dexamethasone. Controls, healthy elderly subjects (n=9);
the elderly (six men, three women; mean±SD age,
Stroke, stroke patients (n=15). One patient was excluded
71 ±9 years). All were thoroughly investigated by a
because of lack of data. Mann-Whitney U test was used for
resident in geriatric medicine. CT scan of the brain was
normal in all individuals, and none were taking drugs. statistical analysis.
The patients were examined between the third and
seventh day after admission. They were investigated in a 179+170 versus 31±7 nmol/1; p=0.003). Six of the
standardized manner, with repeated clinical assess- patients (38%) were nonsuppressors to dexamethasone
ments. The extent of extremity paresis (i.e., of the by conventional criteria; i.e., they had a serum cortisol
extremity mostly afflicted) was quantified using a four- level >138 nmol/1 after dexamethasone.15 The healthy
point scale.3 Acute confusional state was diagnosed control subjects were all suppressors to dexamethasone;
using criteria from the Diagnostic and Statistical Manual the highest postdexamethasone cortisol level in the
of Mental Disorders, edition 3, revised (DSM-III-R).1112 control group was 47 nmol/1.
A diagnosis of a major depressive episode was based on Stroke patients had a higher responsiveness to
criteria from DSM-III-R.12 All tests and interviews with ACTH, although not significantly so (serum cortisol
patients, relatives, and staff regarding the diagnoses of after 30 minutes, 802±226 versus 688±115 nmol/1,
Downloaded from http://ahajournals.org by on January 18, 2021

acute confusional state and major depression were p=0.23). The area under the curve for the cortisol
made by one of the authors. response after ACTH was clearly elevated in seven
A short adrenocorticotropic hormone (ACTH) stim- (47%) of the stroke patients (Figure 2) compared with
ulation test was performed in the fasting state between the control group. The centered cumulative response
8 and 9 AM. In this test 0.25 mg ACTH (1-24 ACTH; was elevated above the upper limit for control subjects
Synacthen, CIBA-Geigy) was administered slowly intra- in three (20%) of the patients. When the highest
venously as a bolus dose. Blood was collected for
cortisol analyses before the injection and 30 and 60
minutes after the injection. This was followed by an £ 70 ooo
overnight dexamethasone suppression test in which 1 o
mg dexamethasone (Decadron, Merck Sharp & Dohme o
International) was given orally at 11 PM. Blood was S 60 000
drawn on the following day at 8 AM for serum cortisol 8
analysis. Serum cortisol was analyzed with a radioimmu-
noassay kit (Farmos Diagnostica, Turku, Finland) with OO
i 50 000
an interassay coefficient variation <8% for the analysis. o OO
The area under the curve for the cortisol response to .2
o
ACTH was calculated by the trapezium rule.13 The o
OO 00 O OOC

centered cumulative cortisol response to ACTH was I 40 000

1

OO
a
calculated using the following formula: area under the ooo
curve—(basal serum cortisol levelx60).13 3
The data were analyzed using the computerized sta-
a
f 30 000
8
1

tistical program SYSTAT.14 Differences in cortisol levels a

•a
between groups were analyzed with the Mann-Whitney c o
3 o
U test. Spearman correlation coefficients (rs) were used a
£
for the calculation of correlations. < 20 000
Controls Stroke
Results FIGURE 2. Scatterplot of serum cortisol response to adreno-
Basal serum cortisol levels were nonsignificantly corticotropic hormone expressed as area under the curve
higher in stroke patients compared with control subjects response. Area under the curve was calculated by the trape-
(mean±SD serum cortisol levels, 387±195 versus zium rule. Controls, healthy elderly subjects (n=9); Stroke,
361 ±82 nmol/1). Serum cortisol levels were significantly stroke patients (n=15). One patient was excluded because of
higher in stroke patients after dexamethasone (Figure 1; lack of data.

14001
p=C .16
1200"
p=0.02
1000"
1r
8 /?<0.05), whereas only a moderate correlation with
E 800"
I p=0. 19 / yc
600"
— ° — ACS
400"
/ •— Non-ACS
200
Before 30 60
Time (minutes)
FIGURE 3. Line graph of serum cortisol before and after the
administration of 0.25 mg i.v. adrenocorticotropic hormone
(1-24 ACTH) to stroke patients (n=16). ACS, stroke patients
with an acute confusional state according to DSM-III-R;
Non-ACS, stroke patients without ACS. Mann-Whitney U test
was used for statistical analysis.
Downloaded from http://ahajournals.org by on January 18, 2021
postdexamethasone serum cortisol level (47 nmol) in
the healthy elderly group was used as an arbitrary cutoff
point with the highest area under the curve response to
ACTH in the control group (43,020 nmol/lxmin) as the
other cutoff point there was a significant association
between nonsuppressibility to dexamethasone and ex-
cessive area under the curve response to ACTH
(/?=0.005; Fisher's exact test). This association did not
remain significant when the centered cumulative corti-
sol response was used instead of the area under the
curve response (p=0.22). The stroke patients with a
"normal response" to ACTH, i.e., an area under the
curve response <43,020 nmol/lxmin, were still more
resistent to dexamethasone compared with control sub-
jects (/?=0.003).
None of the patients were judged to have a major
depression. However, five of the 16 patients were diag-
nosed as having an acute confusional state when the
tests were made. Patients with acute confusional state
had a significantly increased cortisol response 30 min-
utes after ACTH compared with patients without con-
fusion (Figure 3). The cortisol response after 60 minutes
was slightly higher in patients with acute confusional
state. However, there was only a tendency toward a
higher area under the curve and centered cumulative
cortisol response in this group of patients (55,196±7,711
versus 39,171±9,352 and 28,241±11,648 versus
19,142±7,509 nmol/lxmin;/?=0.16 andp=0.46, respec-
tively). Seven patients (44%) had some degree of limb
paresis.
The response to ACTH after 30 minutes in stroke
patients was significantly correlated to male sex, exten-
Olsson et al Stroke and the Cortisol Axis 1575
sive limb paresis, and acute confusional state (rs=0.63,
0.63, and 0.55, respectively;p<Q.O5). Furthermore, limb
paresis and acute confusional state were strongly posi-
tively correlated (T- S =0.86;/J<0.001). The area under the
curve response for cortisol to ACTH was significantly
associated with acute confusional state and male sex in
stroke patients (rs=0.63 and r s =0.62, respectively;
p<0.05), whereas a moderate correlation to limb paresis
was seen (rs=0.45; NS). The centered cumulative corti-
sol response diminished with increasing age (/-s=-0.62;
acute confusion was seen (rs=0.42; NS).
Serum cortisol levels after dexamethasone in stroke
patients were significantly correlated to the presence of
acute confusional state and to limb paresis (rs=0.66 and
0.62, respectively; /?<0.05). Four of five patients with
acute confusional state had a profoundly decreased
suppression after dexamethasone, i.e., a serum cortisol
level >200 nmol/1 after dexamethasone, whereas none
of the nonconfused patients had a postdexamethasone
cortisol level above this value.
Discussion
This study indicates abnormalities in the hypothalam-
ic-pituitary-adrenal axis both at a central level, i.e., a
decreased suppressibility, and in the feed-forward,
stimulatory arm of the cortisol axis early after stroke.
An increased cortisol production rate has been shown in
stroke patients, using both urinary free cortisol levels4
and isotope dilution methods.5 Based on our findings,
an increased cortisol production rate early after stroke
can, in some patients, be caused by an increased adrenal
responsiveness to ACTH.
Repeated stresses are common for stroke patients,
such as various cardiovascular complications, infections,
and emotional reactions. These repeated stresses may
increase the adrenal sensitivity to ACTH 16 and there-
fore prolong the hypercortisolism. Furthermore, in a
study of patients undergoing major abdominal surgery it
was shown that the high levels of cortisol in the first 2
days after surgery were associated with increased levels
of ACTH and corticotropin-releasing hormone, indicat-
ing an increased central drive.17 Thereafter the levels of
ACTH and corticotropin-releasing hormone were sup-
pressed to subnormal values, but the adrenal respon-
siveness to ACTH was increased.17 Thus, an increased
adrenal responsiveness to ACTH may contribute to
maintaining high circulating cortisol levels after stroke
in some patients. Stimulating effects of cytokines on the
cortisol axis and direct effects of a rise in intracranial
pressure may also influence cortisol levels.18'19 These
changes may be aggravated in the elderly, in whom the
threshold for cortisol suppression seems to be elevated
in various diseases such as major depression and after
femoral neck fractures.20-21
There seems to be a subgroup of patients who have a
disturbance of the hypothalamic-pituitary-adrenocorti-
cal axis both centrally and peripherally. In our study
cortisol axis abnormalities were confined mostly to
patients with cognitive disturbances, preferably those
with an extensive functional impairment. This is in line
with earlier studies from our unit. 34 Cognitive distur-
bances late after stroke have also been associated with
high postdexamethasone cortisol levels.22 Furthermore,
the occurrence of postoperative delirium was in one
 1576 Stroke Vol 23, No 11 November 1992
study associated with a significant increase in circulating
levels of cortisol.23 There also seems to be an association
between hypercortisolism and major depression late
after stroke,22 but this is not seen early after stroke.3
Glucocorticoid receptors are present in the human
brain,24 and high glucocorticoid levels, endogenous
and/or exogenous, have been suggested to be toxic to
neurons, especially in the hippocampus.25-26 In addition,
the hippocampus seems to be very sensitive to ischemic
brain damage, with experimental lesions causing impair-
ments of learning and memory in rats.27 Theoretically,
hypercortisolism could reinforce the ischemic damage
to hippocampal neurons and thereby contribute to
cognitive disturbances after stroke.28 Hippocampal ac-
tivity normally inhibits the cortisol axis, as demon-
strated by electrical stimulation in humans.29 Damage to
hippocampal neurons can therefore be followed by
hypercortisolism. This may be due to either a lesion-
induced loss of a feedback site in the hippocampal
system or an enhanced stimulation causing basal hyper-
secretion.30 In stroke patients these abnormalities may
coexist.
In summary, we have shown both an increased re-
sponsiveness of the adrenal glands to ACTH and a
decreased suppressibility to dexamethasone in sub-
groups of patients early after acute stroke. This indi-
cates disturbances at several sites within the hypotha-
lamic-pituitary-adrenocortical axis early after stroke.
These changes seem to be associated mainly with cog-
nitive disturbances in patients with pronounced neuro-
logical deficits.
Downloaded from http://ahajournals.org by on January 18, 2021
Acknowledgments
The authors wish to thank Dr. Kjell Asplund and Dr. Erik
Hagg for constructive criticism of the manuscript.
References
1. Oka M: Effect of cerebral vascular accident on the level of 17-hy-
droxycorticosteroids in plasma. Ada Med Scand 1956;156:221-226
2. Feibel JH, Hardy PM, Campbell RG, Goldstein MN, Joynt RJ:
Prognostic value of the stress response following stroke. JAMA
1977;238:1374-1376
3. Olsson T, Astrom M, Eriksson S, Forssell A: Hypercortisolism
revealed by the dexamethasone suppression test in patients with
acute ischemic stroke. Stroke 1989;20:1685-1690
4. Olsson T: Urinary free cortisol excretion shortly after ischaemic
stroke. J Intern Med 1990;228:177-181
5. Korsic M, Brinar V, Plavsic V, Mihajlovic D, Giljevic Z: Some
aspects of adrenocortical stress response following stroke. Ada
Med lugosl 1990;44:137-145
6. Mulley GP, Wilcox RG, Harrison MJG: Plasma cortisol as a mea-
sure of stress response in acute stroke, (letter) Stroke 1989;20:1593
7. Woolf PD, Cox C, Kelly M, Nichols D, McDonald JV, Hamill RW:
The adrenocortical response to brain injury: Correlation with the
severity of neurological dysfunction, effects of intoxication, and
patient outcome. Alcohol Clin Exp Res 1990;14:917-921
8. Gurevich D, Siegel B, Dumlao M, Perl E, Chaitin P, Bagne C,
Oxenkrug G: HPA axis responsitivity to dexamethasone and cog-
nitive impairment in dementia. Prog Neuropsychopharmacol Biol
Psychiatry 1990;14:297-308
9. Wolkowitz OM, Reus VI, Weingartner H, Thompson K, Breier A,
Doran A, Rubinow D, Pickar D: Cognitive effects of corticoste-
roids. Am J Psychiatry 1990;147:1297-1303
10. Levkoff SE, Evans DA, Liptzin B, Cleary PD, Lipsitz LA, Wetle
TT, Reilly CH, Pilgrim DM, Schor J: Delirium: The occurrence
and persistence of symptoms among elderly hospitalized patients.
Arch Intern Med 1992;152:334-340
11. Gustafson Y, Olsson T, Eriksson S, Asplund K, Bucht G: Acute
confusional states (delirium) in stroke patients. Cerebrovasc Dis
1991;l:257-264
12. American Psychiatric Association Committee on Nomenclature
and Statistics: Diagnostic and Statistical Manual of Mental Disorders,
ed 3, revised. Washington, DC, American Psychiatric Assoc, 1987,
pp 99-107
13. Altman DG: Practical Statistics for Medical Research. London,
Chapman and Hall, 1991, pp 431-433
14. Wilkinson L. SYSTAT: The System for Statistics. Evanston, III,
SYSTAT Inc, 1988
15. Carroll BJ, Feinberg M, Greden JF, Tarika J, Albala AA, Haskett
RF, James NM, Kronfol Z, Lohr N, Steiner M: A specific labora-
tory test for the diagnosis of melancholia. Arch Gen Psychiatry
1981;38:15-22
16. Gaillard RC, Al-Damluji S: Stress and the pituitary-adrenal axis.
Baillieres Clin Endocrinol Metab 1987;l:319-354
17. Naito Y, Fukata J, Tamai S, Seo N, Nakai Y, Mori K, Imura H:
Biphasic changes in hypothalamo-pituitary-adrenal function dur-
ing the early recovery period after major abdominal surgery. J Clin
Endocrinol Metab 1991;73:111-117
18. Bateman A, Singh A, Krai T, Solomon S: The immune-hypo-
thalamic-pituitary adrenal axis. Endocr Rev 1989;10:92-112
19. Feibel J, Kelly M, Lee L, Woolf P: Loss of adrenocortical suppres-
sion after acute brain injury: Role of increased intracranial pres-
sure and brain stem function. J Clin Endocrinol Metab 1983;57:
1245-1250
20. Stangl D, Pfohl B, Zimmerman M, Coryell W, Corenthal C: The
relationship between age and post-dexamethasone cortisol: A test
of three hypotheses. J Affect Disord 1986;11:185-197
21. Roberts NA, Barton RN, Horan MA, White A: Adrenal function
after upper femoral fracture in elderly people: Persistence of stim-
ulation and the roles of adrenocorticotrophic hormone and immo-
bility. Age Ageing 1990;19:304-310
22. Lipsey JR, Robinson RG, Pearlson GD, Rao K, Price TR: The
dexamethasone suppression test and mood following stroke. Am J
Psychiatry 1985;142:318-323
23. Mclntosh TK, Bush HL, Yeston NS, Grasberger R, Palter M, Aun
F, Egdahl RH: Betaendorphin, cortisol and postoperative delir-
ium: A preliminary report. Psychoneuroendocrinology 1985;10:
303-313
24. Seckl J, Dickson KL, Yates C, Fink G: Distribution of glucocorti-
coid and mineralocorticoid receptor messenger RNA expression in
human postmortem hippocampus. Brain Res 1991;561:332-337
25. Sapolsky RM, Krey LC, McEwen BS: The neuroendocrinology of
stress and aging: The glucocorticoid cascade hypothesis. Endocr
Rev 1986;7:284-301
26. De Reuck J, Vandekerckhove T, Bosma G, De Meulemeester K,
Van Landegem W, De Waele J, Tack E, De Koninck J: Steroid
treatment in acute ischaemic stroke: A comparative retrospective
study of 556 cases. Eur Neurol 1988;28:70-72
27. Auer RN, Jensen ML, Whishaw IQ: Neurobehavioral deficit due
to ischemic brain damage limited to half of the CA1 sector of the
hippocampus. J Neurosci 1989;9:1641-1647
28. Sapolsky RM, Pulsinelli WA: Glucocorticoids potentiate ischemic
injury to neurons: Therapeutic implications. Science 1985;229:
1397-1400
29. Mandell A, Chapman LF, Rand RW, Walter RD: Plasma corti-
costeroids: Changes in concentration after stimulation of hippo-
campus and amygdala. Science 1963;139:1212
30. Sapolsky RM, Zola-Morgan S, Squire LR: Inhibition of glucocor-
ticoid secretion by the hippocampal formation in the primate.
J Neurosci 1991;ll:3695-3704