Long-Term Impaired Quality of Life in Cushing’s
Syndrome despite Initial Improvement after
John R. Lindsay, Tonya Nansel, Smita Baid, Julie Gumowski, and Lynnette K. Nieman
Reproductive Biology and Medicine Branch (J.R.L., L.K.N., S.B.), Division of Epidemiology Statistics and Prevention
Research (T.N.), National Institute of Child Health and Human Development, and National Institute of Allergy and
Infectious Diseases (J.G.), National Institutes of Health, Bethesda, Maryland 20892
Context: Cushing’s syndrome (CS) is associated with symptoms that
may impair health-related quality of life (HRQL).
Methods: We used the short-form 36 survey to evaluate HRQL in 23
patients with Cushing’s disease before and after transsphenoidal
surgery (age, 42.7 ? 12.0 yr; 19 women and four men) and in a
cross-section of 343 CS patients (age, 48.2 ? 14.1 yr; 265 women and
78 men) in remission for up to 25.8 yr after surgery (adrenal, 5%;
ectopic, 6%). The z-scores were calculated for short-form 36 domains,
and physical (PCS) and mental (MCS) summary scores were com-
pared with those of age- and sex-matched controls (n ? 6742).
Results: Active Cushing’s disease was associated with low PCS and
MCS scores (P ? 0.05). Despite residual postoperative impairment,
primarily of physical domains, all HRQL parameters improved after
treatment with transsphenoidal surgery (3.2 ? 1.5 yr; P ? 0.05). In
the cross-section in remission at follow-up, there was a small, but
significant (P ? 0.05), impairment of both PCS and MCS. A longitu-
dinal postoperative analysis confirmed stable, but impaired, HRQL
over time. Logistic regression demonstrated that previous pituitary
radiation and current glucocorticoid use had little effect on HRQL
Conclusion: CS is associated with impaired HRQL, which partially
resolves after treatment. At longer-term follow-up, however, there is
residual impairment of HRQL. Determination of modifiable factors
that contribute to impaired HRQL may help reduce the physical and
psychosocial burden of this disease. (J Clin Endocrinol Metab 91:
may lead to impaired health-related quality of life (HRQL)
(1, 2). Important functional impairments arising from hy-
percortisolism include muscle weakness, fracture, and car-
diovascular events, which can adversely affect the survival
and well-being of these patients (3). Additional psycholog-
anxiety (4, 5), and cognitive decrements (6). Improved psy-
of hypercortisolism (7, 8). Short-term series have shown sig-
nificant improvements occurring within the first year of
treatment (9). However, longer studies suggest that pro-
longed exposure to high levels of glucocorticoids may cause
long-lasting deleterious effects (10–12).
Although the process of differential diagnosis and man-
agement of CS has been well validated (13), outcomes in
relation to quality of life have received less attention. Im-
standard short-form 36 (SF-36) health survey, which is an
integrated measure of physical and psychological well-being
HE MAJORITY OF patients with Cushing’s syndrome
(CS) have physical and psychological symptoms that
(3, 10, 14, 15). Previous studies were of relatively small
groups of patients with Cushing’s disease (CD) alone, with
short duration of postoperative follow-up and in the absence
of an age- and gender-matched control comparison.
We conducted a prospective study of HRQL using the
SF-36 health survey in a cohort of patients with CD before
and after transsphenoidal surgery (TSS). To characterize the
impact on HRQL of successful treatment in CS, we also
performed a cross-sectional analysis of HRQL in a cohort in
long-term clinical remission, which included patients with
The hypothesis was that we would observe improved HRQL
after treatment and that HRQL would be greater with in-
creasing duration of remission from hypercortisolism. We
anticipated lower HRQL in patients with ectopic compared
with adrenal- or pituitary-dependent CS due to associated
Patients and Methods
All patients were enrolled in a protocol approved by the institutional
review board of the National Institute of Child Health and Human
Development for investigation and management of CS. This study was
carried out from 1999–2004 at the National Institutes of Health Warren
CS were asked to complete the survey before and after treatment, and
they comprise group 1. Twenty-five of 33 patients who completed a
baseline pretreatment questionnaire also provided a follow-up ques-
tionnaire postoperatively. Twenty-three of these individuals who had
been followed for at least 6 months postoperatively were included in the
First Published Online November 8, 2005
Abbreviations: CD, Cushing’s disease; CI, confidence interval; CS,
Cushing’s syndrome; HRQL, health-related quality of life; MCS, mental
component scale; PCS, physical component scale; SF-36, short-form 36;
TSS, transsphenoidal surgery; XRT, irradiation.
JCEM is published monthly by The Endocrine Society (http://www.
endo-society.org), the foremost professional society serving the en-
Printed in U.S.A.
The Journal of Clinical Endocrinology & Metabolism 91(2):447–453
Copyright © 2006 by The Endocrine Society
first author (J.R.L.) used an existing database of approximately 900
patients who had been enrolled in a range of protocols for investigation
and excluded them from a mailing list sent to patients with confirmed
CS. The survey was commenced in 1999, with two subsequent mailings
in 2001 and 2004 to a total of 775 patients. Four hundred eighteen
individuals returned 688 questionnaires; of these, 159 contributed two
or more responses. Group 2 comprised a cross-sectional analysis of the
most recent/latest questionnaire response from 343 subjects who were
deemed to be in postoperative clinical remission. Seventy-five patients
were excluded from this analysis, comprising eight patients who com-
pleted pretreatment responses alone, seven patients with insufficient
follow-up data, and 60 individuals with possible/previously confirmed
recurrence who are described below.
Quality of life
We used the SF-36R (version 2) for assessment of HRQL (16). The
SF-36 is a generic health survey that has been widely validated and used
in an international context to evaluate self-reported domains of health
status (16). The SF-36 survey consists of a 36-item questionnaire that
includes eight components of HRQL: physical functioning, role limita-
tions due to physical health, bodily pain, general health, vitality, social
functioning, role limitations due to emotional health, and mental health.
These eight domains formed two broader health dimension scales: a
physical component scale (PCS) and a mental component scale (MCS).
two health dimensions by setting the general population mean at 50 ?
10 (mean ? 1 sd). To control for any difference in age and gender
between the population norms and our sample, we compared our sam-
studied by Ware et al. (16). To do this, we created a deviation score for
each patient, calculated as the absolute difference between the individ-
ual’s score and the appropriate age/gender norm. This score was then
standardized to a z-score by dividing it by the sd for that age/gender
sds the patient deviated from his/her own age/gender population
norm. In our analyses, we present data using both the SF-36 standard-
ized score as well as this age/gender-controlled deviation score.
In addition to the SF-36 quality of life dimensions, cognitive function
was assessed with the Medical Outcomes Study cognitive functioning
scale (17). The Medical Outcomes Study quality of life assessment was
the precursor to the SF-36; however, the cognitive functioning scale was
omitted from the SF-36 in the interest of brevity. We elected to include
the cognitive functioning scale, because this may be a particularly rel-
evant dimension for this population.
Demographic variables and symptom checklist
The survey instrument included demographic data (age, gender,
marital status, and highest grade of educational attainment), diagnosis,
primary treatment, and subsequent treatments (surgery, medicine, or
radiation) since discharge. The subjects were asked to indicate whether
they had experienced a recurrence or were currently cured of CS and to
indicate whether they were currently taking hydrocortisone, pred-
nisone, dexamethasone, mitotane, or ketoconazole. All patients were
requested to provide details of the most recent laboratory screening for
objective assessment of remission status. Patients also were asked to
indicate whether they were currently experiencing any of 42 items on a
checklist containing signs and symptoms associated with CS.
Evaluation of long-term remission
The posttreatment cross-sectional analysis excluded patients with
previously confirmed recurrence of CS as determined from our internal
records (n ? 34) and patients who indicated that they were not cured
(n ? 19). Patients with persistent hypercortisolism (n ? 2) or who
required ketoconazole (n ? 4) or mitotane (n ? 1) treatment were
excluded. These responses were cross-checked with our records and
with returned results of biochemical screening for examination of re-
Descriptive statistics were calculated for continuous variables. Fre-
quencies were generated for other variables. Significance was assumed
at P ? 0.05. All data including the SF-36 domains are expressed as the
mean ? sd, except where noted. ?2and Fisher’s exact tests were used
to examine differences in demographic variables. Only patients with
both pre- and posttreatment responses were included in the intraindi-
of z-scores for each group and domain were calculated. A mean z-score
(effect size) less than 0.20 was considered clinically nonsignificant, a
z-score in the range 0.2–0.5 was classified as small, a z-score in the range
0.5–0.8 as moderate, and a z-score greater than 0.8 as large (18, 19).
Spearman’s rank test was used to examine correlations between SF-36
outcomes and continuous demographic variables.
An analysis of the effect of time from treatment on HRQL was con-
ducted within group 2, which included serial observations from indi-
vidual respondents. However, if a subject contributed more than one
in that time period was used. Data were analyzed using mixed model
analysis of repeated measures, followed by post hoc testing for pairwise
comparisons among the times from treatment, using Statistical Analysis
System (SAS) version 8 software (SAS Institute, Inc., Cary, NC).
Logistic regression analysis was used to evaluate the relationship of
the physical symptom score, time from treatment, glucocorticoid use,
previous pituitary irradiation (XRT) and deviation of greater than 0.5 sd
in PCS and MCS domains below the normative mean. Testing was
performed using the statistical package for social science (SPSS version
12 for Windows, release 09.04, 2003, SPSS, Inc., Chicago, IL) and Graph-
Pad PRISM version 4.00 for Windows (GraphPad, San Diego, CA;
Group 1: HRQL in CD before and after TSS
Patients with active CD (Table 1) had low PCS, MCS, and
low individual SF-36 domain z-scores compared with age-
and MCS scores improved from 32.6 ? 10.5 and 38.8 ? 12.5
to 45.8 ? 12.7 and 50.5 ? 9.6, respectively (P ? 0.001), at
postoperative follow-up (mean, 3.2 ? 1.5; range, 0.7–5.4 yr;
Table 2). All individual SF-36 domains improved after TSS
function after treatment (56.2 ? 25.8 vs. 73.2 ? 29.4; P ? 0.02)
Postoperatively, scores for bodily pain (P ? 0.5), vitality
(P ? 0.4), social functioning (P ? 0.1), role limitations due to
emotional health (P ? 0.6), general health (P ? 0.4), mental
health (P ? 0.5), and the mental summary scores (P ? 0.8),
improved to levels similar to U.S. population norms. In con-
trast, physical functioning (P ? 0.04), role limitations due to
physical health (P ? 0.049) and physical summary (P ? 0.04)
scores remained significantly below U.S. population norms.
Group 2: cross-sectional posttreatment survey of patients in
clinical remission from CS
(CD, 89%; adrenal, 5%; ectopic, 6%) in the posttreatment
cross-sectional sample were similar in terms of gender, ed-
ucation, and marital status to those in the posttreatment arm
of group 1 (Table 1). Compared with group 1, the patients
were older by approximately 6 yr (P ? 0.06), a smaller pro-
portion was treated by TSS alone (P ? 0.09), and they had a
longer period of remission (P ? 0.0001). The SF-36 domain
of group 1 were not statistically different.
In the cross-sectional group, PCS scores were higher in
J Clin Endocrinol Metab, February 2006, 91(2):447–453 Lindsay et al. • Quality of Life in CS
men (P ? 0.02) and inversely associated with age (r ? ?0.36;
P ? 0.001). Both PCS and MCS scores were independent of
diagnostic category and educational status.
Postoperatively at the longest follow-up point (11.8 ? 4.9
yr; range, 0.7–25.8 yr), z-scores for SF-36 domains were low
(P ? 0.05) compared with age- and sex-matched population
these patients had low PCS (45.7 ? 0.7) and MCS (47.0 ? 0.7)
scores (mean ? sem; P ? 0.0001) compared with the general
population controls (50 ? 0.1). Corresponding z-scores for
PCS and MCS were ?0.4 ? 1.2 and ?0.3 ? 1.2.
Subjective health rating and satisfaction scale among cross-
Patients compared their current health to that before de-
veloping CS. Results from patients in the remission group
were as follows: much better, 38%; somewhat better, 12%;
about the same, 20%; somewhat worse, 20%; much worse,
7%; or missing, 3%. Similarly, when asked to rate their level
of satisfaction with the outcome of treatment for CS, re-
sponses were: extremely satisfied, 57%; quite satisfied, 26%;
moderately satisfied, 9%; slightly satisfied, 4%; not at all
satisfied, 2%; or missing, 3%.
HRQL by time from treatment
We observed long-term higher PCS (P ? 0.0001) and MCS
(P ? 0.01) scores in the cross-sectional analysis compared
with the pretreatment arm of group 1 (Fig. 3). There were no
significant changes in PCS or MCS scores in the posttreat-
ment period with increasing time from treatment (P ? 0.05).
Symptoms and relationship to SF-36 summary scores
Table 3 demonstrates the prevalence of active symptoms
in the cross-sectional series at the most recent follow-up.
FIG. 1. Individual SF-36 domains before (Pre) and after (Post) treat-
ment in 23 patients with CD treated with TSS. Results are expressed
as the mean ? 95% CI for the pre- and posttreatment z-scores com-
pared with age- and sex-matched controls from the U.S. population.
SF-36 domains are PCS, MCS, physical functioning (PF), role limi-
tations due to physical health (RP), body pain (BP), general health
(GH), vitality (VT), social functioning (SF), role limitations due to
emotional health (RE), and mental health (MH). P ? 0.0001 for dif-
ferences between all domains in the pre- and posttreatment arms,
except for BP (P ? 0.03).
TABLE 1. Characteristics of subjects treated for CS in the SF-36 analysis
(n ? 343)
(n ? 23)
(n ? 23)
Age range (yr)
Gender (% female)
Marital status (%)
Educational attainment (%)
8th grade or less
Some high school; did not graduate
High school or GED
Some college or 2-yr degree
4-yr college degree
4? college degree
Time interval since treatment (yr)
Glucocorticoid dependency (%)
Previous treatment (%)
3.2 (1.5) 11.8 (4.9)a
Group 1: Before and after TSS in CD; group 2: cross-sectional posttreatment analysis of patients in apparent remission from CS. Results
are means (SD). ADX, Adrenalectomy; GED, General Educational Development; Rx, treatment.
aP ? 0.0001 for comparison of demographic data of patients in the posttreatment arm of group 1 and those in the cross-sectional series.
Lindsay et al. • Quality of Life in CSJ Clin Endocrinol Metab, February 2006, 91(2):447–453
Fatigue (41%), forgetfulness (35.7%), and trouble sleeping
(33.3%) were the most common symptoms. The presence of
any symptom was associated with lower PCS/MCS in all
cases in which a significant difference was observed. The
presence of symptoms measured using the total numerical
score of positive symptoms was inversely related to SF-36
summary domains: PCS: r ? ?0.5; P ? 0.0001; and MCS: r ?
?0.6; P ? 0.0001.
Effect of glucocorticoids
Although patients requiring glucocorticoids after remis-
sion of CS had marginally lower physical summary z-scores
(?0.5 ? 1.3 vs. ?0.3 ? 1.2; P ? 0.05), the remainder of the
SF-36 domains were similar (P ? 0.05) compared with those
who were glucocorticoid independent.
Predictors of HRQL
The use of glucocorticoids and previous pituitary XRT
were not significantly associated with impaired PCS and/or
MCS scores in a model that examined their relationship to
z-score deviation of more than 0.5 sd below age- and sex-
matched population means. Physical symptoms scores (de-
fined in Table 3) were significantly associated with both
PCS was 14.4 (6.4–32.7) and for MCS was 6.1 (2.9–12.8) in
those with six or more physical symptoms compared with
those with none or one symptom. Increasing time from treat-
ment was identified as a significant contributor to the ob-
ratios (95% CI) for impaired PCS was 0.4 (0.2–1.2), 0.36 (0.2–
0.9), and 0.3 (0.1–0.6) for patients 0–5, 5–10, and 10–15 yr
from treatment, respectively, compared with those who had
been treated more than 15 yr previously.
FIG. 2. Individual domains of SF-36 in 343 patients in long-term
apparent remission after treatment for CS. Data are presented as the
mean ? 95% CIs. All data crossing the zero line (U.S. population age-
and sex-matched control values; n ? 6742) are not significant.
FIG. 3. PCS and MCS domains for the pretreatment arm of group 1
(pre) and the cross-sectional posttreatment responses, expressed as
z-scores, compared with age- and sex-matched controls. The timing of
each patient’s posttreatment response was categorized according to
time (years) from treatment: 0–4.9 (n ? 44), 5–9.9 (n ? 131), 10–14.9
(n ? 171), and more than 15 (n ? 94) yr. In the event of multiple
was selected for analysis. Results are expressed as the mean ? SEM.
***, P ? 0.0001; **, P ? 0.01 (for comparison with the posttreatment
TABLE 2. SF-36 domains in patients treated for CS: before and after TSS in CD (n ? 23) and in a cross-sectional posttreatment (post-
Rx) analysis from those in apparent long-term remission from CS at latest follow-up (n ? 343)
(n ? 23)
(n ? 23)
(n ? 343)
Pre-Rx CD vs.
Post-Rx CD vs.
All statistical comparisons between groups are as indicated, and controls were derived from the 1998 U.S. general population (16). Results
are means (SD).
aP ? 0.0001 vs. population controls.
bP ? 0.05 vs. population controls.
J Clin Endocrinol Metab, February 2006, 91(2):447–453 Lindsay et al. • Quality of Life in CS
Differences between subjects in remission and those with a
confirmed previous recurrence or persistent hypercortisolism
at most recent mailing
We examined differences in SF-36 responses for those in
apparent clinical remission compared with others who had
scores were lower among those who had a previous recur-
rence (37.6 ? 12.2), those with persistent hypercortisolism
(38.3 ? 3.7), and those who believed they were not cured
(34.2 ? 11.8) compared with the remission group detailed
above (by ANOVA, P ? 0.0001). Similarly, when we exam-
ined MCS scores, patients with a previous recurrence (41.9 ?
16.4), with persistent hypercortisolism (44.9 ? 3.7), or who
believed they were not cured (39.6 ? 13.3) had lower scores
than the remission group (P ? 0.0001).
The present series confirms that active CS is associated
with impaired HRQL. The magnitude of difference com-
pared with age- and sex-matched controls was large in all
SF-36 domains, except for mental health and bodily pain,
which were moderately impaired. The magnitude of these
changes also was similar to previous reports that HRQL in
active CS is significantly impaired compared with that in
subjects with acromegaly, prolactinoma, or nonfunctioning
Postoperative HRQL values within the first 3 yr of fol-
ever, despite apparent clinical remission, when compared
with the U.S. population, there was residual impairment
primarily of physical scores, comprised of physical function-
TABLE 3. Prevalence of symptoms and relationship to SF-36 physical (PCS) and mental (MCS) component scores in patients comprising
the posttreatment cross-sectional series
Active symptoms at follow-up
% of patients
Mean norm-based PCS score (SD) Mean norm-based MCS score (SD)
Decreased muscle strength or weaknessa
Feelings of sadness
Decreased ability to exercisea
High blood pressurea
Decreased attention span
Feelings of frustration
Feelings of being fat and ugly
Thinning or balding of haira
Women and men: lack of sexual drivea
Unable to worka
Feelings of hopelessness
Cuts/abrasions heal slowlya
Women: loss of menstrual periodsa
Purple stretch marksa
Women: irregular menstrual periodsa
Ruddy or red complexiona
High blood sugara
Feelings of being left out of the family
Fat pads around neck and shouldersa
Column 2 demonstrates the percentage of patients who indicated the presence of active symptoms.
aSymptoms that comprised physical symptoms score.
Mean (?SD) norm-based PCS and MCS scores were calculated for each group with (?ve) or without (?ve) symptoms and analyzed for
significant differences using Student’s unpaired t test;bP ? 0.0001,cP ? 0.01,dP ? 0.05.
Lindsay et al. • Quality of Life in CS J Clin Endocrinol Metab, February 2006, 91(2):447–453