Content uploaded by Ivan Berlin
Author content
All content in this area was uploaded by Ivan Berlin on Nov 18, 2015
Content may be subject to copyright.
BRIEF REPORT
International Journal of Neuropsychopharmacology (1999), 2, 105–110. Copyright # 1999 CINP
Serum thyroid-stimulating-hormone
concentration as an index of severity of major
depression
Ivan Berlin
1
, Christine Payan
1
, Emmanuelle Corruble
2
and Alain J. Puech
3
"
Department of Pharmacology, Ho
V
pital Pitie
T
-Salpe
V
trie
Z
re, Paris, France
#
Department of Psychiatry, Ho
V
pital Paul Brousse, Villejuif, France
$
Sanofi, Gentilly, France
Abstract
Alterations in thyroid axis are common in depression and subclinical hypothyroidism may predispose to
recurrent depressive episodes and resistance to antidepressants. The same normal reference ranges are used in
both depressive and non-psychiatric patients to detect hypothyroidism. We hypothesized that in depressive
patients, serum TSH (thyrotropin) elevation within the normal reference range ( upper 25th percentile) may
be related to patients ’ characteristics reflecting the severity of the depressive illness.
We analysed, in a cross-sectional study, the relationship between serum TSH and serum-free thyroxine (T4)
concentrations and different demographic and psychiatric characteristics in 94 depressive in-patients with
DSM-III-R criteria for major depression.
The frequency of subclinical hypothyroidism (normal serum T4, higher than normal serum TSH) was 5.3%.
In univariate analyses patients who had serum TSH concentrations upper 25th percentile of the normal
range were more likely to have recurrent depression, longer disease duration, higher number of episodes of
major depression, higher number of previous suicide attempts and higher body mass index than those patients
who had serum TSH concentrations upper 25th percentile of the normal range (age-adjusted p 0.05).
Stepwise logistic regression analysis showed that serum TSH upper 25th percentile of the normal range was
positively associated with recurrent depression (p l 0.0001), presence of somatic disease condition (p l 0.04),
marital status (p l 0.06) and number of suicide attempt (p l 0.1). On the other hand, significantly higher
serum TSH concentrations were observed in patients with recurrent depression, melancholia and associated
somatic disease conditions. Correspondence analysis showed that serum TSH in the higher 25th percentile of
the normal reference range projected together with the presence of melancholia, psychiatric and somatic
disease conditions, severe major depressive episodes, recurrence of depressive episodes, prescription of at least
two antidepressants or non-response to two antidepressants, and previous suicide attempts.
Our study suggests that serum TSH concentration in the upper 25th percentile of the normal reference range
may be associated with characteristics of severe major depression. Further prospective studies are needed to
establish whether serum TSH concentration in the upper 25th percentile of the normal reference range is a
contributory causal factor or a consequence of the severity of major depression.
Received 8 October 1988; Reviewed 6 January 1999 ; Revised 7 February 1999; Accepted 22 February 1999
Key words: Major depression, severity, serum TSH.
Introduction
Depression may be a clinical symptom of primary
hypothyroidism which is usually due to autoimmune
disease or the destructive therapy of Graves’ disease.
Alterations in the thyroid axis are common in depression
and several clinical symptoms are common to both
depression and hypothyroidism (Nemeroff, 1989). Clini-
Address for correspondence : Dr Ivan Berlin, Department of
Pharmacology, Ho
#
pital Pitie
!
-Salpe
#
trie
'
re, 47 Bd de l’Ho
#
pital, 75013
Paris, France.
Fax: 33 1 42 16 16 88
E-mail: ivan.berlin!psl.ap-hop-paris.fr
cal hypothyroidism is rarely associated with depression
(Ordas and Labbate, 1995) and the thyroid axis alterations
are generally subtle in depressive patients. The absence of
the normal nocturnal TSH surge which may result in a
lower circadian thyroid hormone secretion (Jackson, 1996 ;
Sullivan et al., 1997) may support the view that thyroid
axis alterations are of central origin (Nemeroff, 1989).
Subclinical hypothyroidism is characterized by serum
concentrations of thyroxine (T4) within the reference
range and a raised serum TSH concentration. This
condition which increases with age is higher in women
than men and has been found to be quite common in
people older than 55 yr in community surveys (Bagchi et
106 I. Berlin et al.
al., 1990; Parle et al., 1991; Sawin et al., 1985). Subclinical
hypothyroidism has been observed in 3.5–13.6% of
depressive patients (Gold et al., 1981; Joffe and Levitt,
1992; Sternbach et al., 1983). Further, exaggerated TSH
responses to TRH (thyrotropin-releasing hormone) occurs
more frequently in depressed patients with normal TSH
concentrations, although in the upper half of the range
considered normal (Kraus et al., 1997). Subclinical hypo-
thyroidism may predispose to higher frequency of
depressive episodes (Haggerty et al., 1993). Triiodo-
thyronine and T4 have been proposed as adjuvant therapy
in non-responders to antidepressants (Bauer et al., 1998 ;
Hickie et al., 1996; Howland 1993; Jackson, 1996; Prange,
1996; Sullivan et al., 1997), although the efficacy of such
treatments remains controversial.
Several evidences confirm the association of thyroid
axis deficiency with depression, and routine screening for
thyroid axis alterations has been advised (Szabadi, 1991).
However, only a few studies evaluated the association of
clinical characteristics with thyroid status (Kraus et al.,
1998; Sullivan et al., 1997). Furthermore, to establish
subclinical hypothyroidism in depressive patients, the
reference ranges used are based on observations in non-
psychiatric, endocrinological patients with hypothyroid-
ism. It may, however, be hypothesized that more subtle
alterations in the thyroid axis may be related to clinical
characteristics in depressive patients and that even small
elevations in serum TSH levels, usually considered to be
within the normal range, may have clinical consequences.
Materials and methods
Subjects
A total of 151 patients meeting DSM-III-R (American
Psychiatric Association, 1987) criteria for major de-
pression as their primary diagnosis and necessitating
hospitalization were screened in seven hospitals of the
Assistance Publique-Ho
#
pitaux de Paris. Among these
patients 97 had serum TSH and serum-free T4 determined
within 8 d of commencing hospitalization; three patients
were excluded from the analyses because they were
treated with T4 at the time of serum TSH and T4
determinations. Patients were classified for diagnostic
criteria by the MINI International Psychiatric Interview
(Lecrubier et al., 1997). Patients with bipolar disorder,
pregnancy, and older than 75 yr were not included in the
study. Investigators classified patients as responders or
non-responders when they were discharged from hospital.
Patients were considered as non-responders if they did
not respond to one antidepressant and required at least
two different types of antidepressant in succession, at
doses generally accepted as sufficiently high to elicit a
therapeutic response. The mean age of the 94 patients
included in data analysis was 44p13 yr; 66 (70%) were
women, 49 (52%) were married, 64 (68%) had recurrent
depression and 60 (64 %) patients presented melancholia
according to DSM-III-R criteria for major depressive
episode melancholic type. The mean duration of major
depressive disorder was 7.6p8.5 yr (median 3.7 yr) and
that of the current episode 8.3p8.6 months (median 4
months). Twenty (21%) patients were receiving neuro-
leptics, 62 (66%) benzodiazepines, 8 (8.5 %) anxiolytics
other than benzodiazepines, and 1 patient was on lithium
and another on carbamazepine (2%). Fifty (53%) patients
had associated somatic disease conditions. The most
frequent somatic illnesses were: hypertension (13
patients) endocrinological disease, mainly diabetes
mellitus (7 patients), allergic manifestations (8 patients),
and rheumatologic disease conditions (6 patients). None
of the patients were treated by drugs known to modify
thyroid function and all were free of iode-containing
radiopaque agents. Informed written consent was
obtained from each patient and the study was approved
by the Ethics Committee of the Cochin Hospital, Paris.
TSH and T4 assays
Serum TSH concentration was measured by third gen-
eration chemiluminescent assay (Spencer et al., 1990) and
serum T4 by BeriLux FT4 solid phase antigen luminescent
technique (Behringwerke AG, Marburg, Germany).
Statistical analyses
Serum TSH and serum-free T4 concentrations were
determined in seven different centres. There was no
difference between the centres for either serum TSH or
serum-free T4 concentrations. As normal ranges of each
laboratory were somewhat different, TSH data were
transformed as percentages of the respective normal
ranges of each laboratory. Further, because TSH and T4
data were not normally distributed [especially TSH data
which were negatively (left) skewed] data were normal-
ized by logarithmic transformation. Statistical analyses
were thus performed on log-transformed data. Normal
laboratory ranges for serum TSH (and serum-free T4) are
determined to exclude overt hypo- or hyperthyroidism.
We hypothesized that in psychiatric conditions more
subtle thyroid axis changes may play a role, and
dichotomized this population of patients with major
depression into two categories : (i) patients who had
serum TSH concentrations upper 25th percentile and
(ii) those who had serum TSH concentrations upper
25th percentile of the reference range.
Although for age no significant difference occurred,
age was included in each statistical analyses as covariate.
ANCOVA was used for between-group comparisons of
107Serum TSH and severity of depression
continuous variables. Frequencies were compared by the
Mantel–Haenszel χ
#
test. Stepwise forward logistic
regression analysis was used to identify variables associ-
ated with serum TSH status. Correspondence analysis was
used to identify aggregates of different variables. Stat-
istical analyses were performed by statistical software
BMDP (release 7, 1992, Los Angeles, CA). All tests were
two-tailed. Differences were considered significant if
p 0.05.
Results
Among the 94 patients with major depression only five
had serum TSH concentrations higher than the normal
reference range and only one had a serum TSH concen-
Table 1. Relationship between serum TSH concentration [upper 25th percentile of the
normal range (or higher n l 5)] and demographic and selected psychiatric
characteristics of patients hospitalized for major depression [n (%) or meanp..]
Serum TSH upper Serum TSH upper
25th percentile 25th percentile
(n l 70) (n l 24) p value‡
Age (yr) 43p13 47p12 0.2
Sex (females) 47 (67) 19 (79) 0.27
Body mass index
(kg\m
#
)
23p4.4 26p6.4 0.03
Marital status
(married)
32 (46) 17 (71) 0.07
Disease duration (yr) 6p811p10 0.04
Present episode
duration 3
months
36 (51) 14 (58) 0.68
Recurrent depression 40 (58) 23 (96) 0.002
Number of previous
episodes of major
depression
0.0003
None 28 (40) 1 (4)
1 22 (31) 11 (46)
2 2 (3) 6 (25)
3 18 (26) 6 (25)
Melancholia present* 42 (60) 18 (75) 0.37
Severity† 52 (74) 17 (71) 0.9
Previous suicide
attempt(s)
25 (36) 14 (57) 0.12
Number of previous
suicide attempt(s)
0.76p1.2 1.68p2.3 0.009
Other psychiatric
diagnoses present
38 (55) 14 (58) 1
Somatic disease
condition associated
33 (47) 17 (71) 0.09
* According to DSM-III-R criteria for major depressive episode melancholic type.
† Number of patients with DSM-III-R severity, grade 3 (severe, without psychotic
features) j4 (with psychotic features).
‡ Age-adjusted.
tration twice as high as the upper limit of the reference range.
None of the patients had serum-free T4 concentrations be-
yond the limits of the reference range. Thus, the frequency
of subclinical hypothyroidism in this sample was 5.3%.
Table 1 shows the effect of the normal range of serum
TSH concentration upper 25th percentile when
compared with that upper 25th percentile. Patients
who had serum TSH concentrations upper 25th
percentile of the normal range, or higher, were more likely
to have recurrent depression, longer disease duration,
higher number of previous suicide attempts, higher
number of episodes of major depression, and higher body
mass index. Variables having a p value lower than 0.1
were included in a stepwise age-adjusted logistic re-
108 I. Berlin et al.
Figure 1. Correspondence analysis. Each variable is represented by two factor scores (factor 1 l axis 1, factor 2 l axis 2). The
higher the absolute value of the factor score the higher the contribution of the variable to this factor. The distance of the
coordinate scores between variables expresses the intensity of their relationship: the smaller the distance the stronger the
relationship. Abbreviations: TSH ‘High’, serum TSH concentration upper 25th percentile of the normal range; TSH ‘Low ’,
serum TSH concentration upper 25th percentile of the normal range ; Recurrent depr &, recurrent depression present; Recurrent
depr ', recurrent depression absent; Som dis &, somatic disease condition present; Som dis ', somatic disease condition absent;
Resp 2AD, response to 2 antidepressants or non-response, Resp 1 AD, response to 1 antidepressant; Melancholia &,
melancholia present; Melancholia ', melancholia absent; Suicide att &, previous suicide attempt(s) ; Suic att ', no previous
suicide attempt; Severe depr ep &, severe depressive episode (as defined by DSM-III-R) present; Severe depr ep ', severe
depressive episode absent; SSRI ', no administration of serotoninergic antidepressant; SSRI &, administration of serotoninergic
antidepressant; TCA &, administration of tricyclic antidepressant; TCA ', no administration of tricyclic antidepressant.
gression analysis. Four variables remained in the final
model with a model fit of 97%: recurrent depression (p l
0.0001), and presence of associated somatic disease
condition (p l 0.04), marital status (p l 0.06) and number
of previous suicide attempts (p l 0.11). Disease duration
and body mass index did not enter in the final model. This
model correctly classified 81% of the patients.
The relationship between variables can be further
illustrated by correspondence analysis. Figure 1 gives
projections of the variables on the first two axes and
shows that serum TSH concentrations upper 25th
percentile of the normal reference range are projected on
the upper side of and close to axis 2, along with the
following variables : presence of melancholia, presence of
109Serum TSH and severity of depression
Table 2. Comparison of serum-free thyroxine (T4) and TSH
concentrations according to some selected patients’
characteristics (means of raw datap..)
Serum-free
T4
(pmol\l)
TSH
(mU\l)
Recurrent depression
Yes 13.4p3 1.8p1.2***
No 13.7p2.6 0.9p0.6
Melancholia present
Yes 13.4p2.9 1.7p1.2*
No 13.6p2.7 1.1p0.7
Previous suicide attempt(s)
Yes 13.1p2.6 1.5p0.9
No 13.8p3 1.5p1.2
Other psychiatric diagnoses
Yes 13.1p2.7* 1.6p1.2*
No 14p3 1.4p1
Number of antidepressants
used
1 14.1p2.7 1.3p1
213p2.9 1.6p1.2
Somatic disease condition
associated
Yes 13.5p2.7 1.7p1.3**
No 13.5p3 1.2p0.8
Statistical analysis was performed on log-transformed data.
* p 0.05, ** p 0.01, *** p 0.0001
somatic disease condition, recurrence of depressive
episodes, prescription of at least two antidepressants,
presence of severe major depressive episodes and a
previous history of a suicide attempt.
When serum-free T4 and TSH levels were directly
compared according to several characteristics of the
depressive illness, a higher serum TSH concentration was
observed in patients with recurrent depression, mel-
ancholia and associated psychiatric and somatic disease
conditions. Patients with other associated psychiatric
diagnoses also had a lower serum-free T4 concentration
(Table 2). The presence of associated generalized anxiety
disorder (n l 29) did not influence serum-free T4
(12.4p2.6 vs. 13.9p2.9 pmol\l, p l 0.25) or serum TSH
(1.5p0.9 vs. 1.5p1.2 pmol\l, p l 0.57) concentrations.
Discussion
This study shows that ‘ high-normal’, i.e. upper 25th
percentile of the normal range of serum TSH concen-
tration is associated with greater severity of major
depression. According to multivariate analyses, depress-
ive patients with ‘ high-normal’ TSH concentrations were
more likely to have recurrent depression, previous suicide
attempts, associated somatic and psychiatric disease con-
ditions, melancholia and were treated by two or more anti-
depressants than those with ‘low-normal’ ( upper 25th
percentile of the normal range) serum TSH concentrations.
The relationship between reduced thyroid axis function
and depression has long been suspected. However, overt
hypothyroidism is very rare among depressive patients
(Fava et al., 1995 ; Gold et al., 1981). No specific study
compared the prevalence rate of subclinical hypo-
thyroidism among depressive patients with that in the
general population. However, prevalence rates seem to be
similar: subclinical hypothyroidism has been observed in
3.5–13.6% of depressive patients (Gold et al., 1981; Joffe
and Levitt, 1992; Sternbach et al., 1983) and it is estimated
to be 5.6–11.1% for women and 2.7–3.7 % for men
between the age range of 35 and 65 yr in the normal
population (Danese et al., 1996). Subclinical hypothyroid-
ism increases with age and is approximately twice as
frequent among women than among men (Bagchi et al.,
1990; Parle et al., 1991) as is the prevalence of depression.
The definition of subclinical hypothyroidism is based
on the normal reference range of serum TSH con-
centrations. This reference range has been established in
healthy subject populations (generally blood donors)
without clinical signs of endocrinological thyroid dys-
function. It is not known whether this ‘ endocrinological’
reference range can be applied to patients with depression.
Recent studies suggest that ‘ high-normal’ serum TSH
concentration is associated with exaggerated TSH re-
sponse to TRH in depressed patients (Kraus et al., 1997).
Circadian difference in serum-free T4 concentration
distinguishes between depressed and control subjects;
depressed subjects have a higher mean TSH response to
TRH than those with single episodes (Sullivan et al.,
1997). It seems, therefore, that subtle thyroid under-
function, with serum TSH concentrations in the upper
third or quarter of the usual ‘endocrinological ’ reference
range may be a contributory factor in patients with
depression. A similar phenomenon has been found for
high serum cholesterol levels. In persons with high serum
cholesterol concentration T4 administration significantly
reduced total and LDL-cholesterol only in subjects
with ‘high-normal’ serum TSH concentrations and it had
no effect in those persons who had ‘low-normal’ serum
TSH levels (Michalopoulou et al., 1998).
Subtle thyroid axis modifications may account for up to
36% of the variance in antidepressant treatment outcome
(Sullivan et al. 1997). Tricyclic antidepressant non-
responders have a higher TSH response to TRH and a
greater reduction in morning to evening difference in
serum T4 levels (Sullivan et al., 1997). Alterations in
thyroid function may influence not only central β-
adrenergic (Whybrow and Prange, 1981) but also sero-
110 I. Berlin et al.
tonergic activities (Cleare et al., 1995) and this may lead
to reduced treatment response. In further studies assessing
predictors of antidepressant response, serum TSH con-
centration should be included as a putative factor which
may contribute to treatment response.
Screening for mild thyroid failure and subsequent
treatment with levothyroxine sodium is cost-effective
because it lowers both hypercholesterolaemia and the risk
of cardiovascular disease, and progression toward overt
hypothyroidism (Danese et al., 1996). Similarly, screening
for mild thyroid failure including ‘ high-normal ’ serum
TSH concentrations may be a cue to the clinician to
consider whether such patients have a more severe form
of depression. Our findings also suggest that it would be
useful to study whether adjunctive treatment with
levothyroxine sodium may be beneficial for such patients
when they fail to respond to antidepressants.
Acknowledgements
This study was supported in part by GERMED, Assistance
Publique-Ho
#
pitaux de Paris.
References
American Psychiatric Association (1987). Diagnostic and
Statistical Manual of Mental Disorders (3rd ed., revised).
Washington, DC: American Psychiatric Association.
Bagchi N, Brown TR, Parish RF (1990). Thyroid dysfunction
in adults over age 55 years. A study in an urban US
Community. Archives of Internal Medicine 150, 785–787.
Bauer M, Hellweg R, Gra
$
f K-J, Baumgartner A (1998).
Treatment of refractory depression with high-dose
thyroxine. Neuropsychopharmacology 18, 444–455.
Cleare AJ, McGregor A, Keane V (1995). Neuroendocrine
evidence for an association between hypothyroidism,
reduced central 5-HT activity and depression. Clinical
Endocrinology 43, 713–719.
Danese MD, Powe NR, Sawin CT, Ladenson PW (1996).
Screening for mild thyroid failure at the periodic health
examination. Journal of the American Medical Association 276,
285–292.
Fava M, Labbate LA, Abraham ME, Rosenbaum JF (1995).
Hypothyroidism and hyperthyroidism in major depression
revisited. Journal of Clinical Psychiatry 56, 186–192.
Gold MS, Pottash ALC, Extein I (1981). Hypothyroidism and
depression. Evidence from complete thyroid function
evaluation. Journal of the American Medical Association 245,
1919–1922.
Haggerty JJ, Stern RA, Mason GA, Beckwith J, Morey CE,
Prange AJ (1993). Subclinical hypothyroidism: a modifiable
risk factor for depression? American Journal of Psychiatry
150, 508–510.
Hickie I, Bennett B, Mitchell P, Wilhelm K, Orlay W (1996).
Clinical and subclinical hypothyroidism in patients with
chronic and treatment-resistant depression. Australian and
New Zealand Journal of Psychiatry 30, 246–252.
Howland RH (1993). Thyroid dysfunction in refractory
depression: implication for pathophysiology and treatment.
Journal of Clinical Psychiatry 54, 47–54.
Jackson I (1996). Does thyroid hormone have a role as
adjunctive therapy in depression? Thyroid 6, 63–67.
Joffe RT, Levitt AJ (1992). Major depression and subclinical
(grade 2) hypothyroidism. Psychoneuroendocrinology 17,
215–221.
Kraus RP, Phoenix E, Edmonds MW, Nicholson IR, Chandara
PC, Tokmakejian S (1997). Exaggerated TSH responses to
TRH in depressed patients with ‘normal’ baseline TSH.
Journal of Clinical Psychiatry 58, 266–270.
Lecrubier Y, Sheehan D, Weiller E, Amorim P, Bonora I,
Sheehan K, Janas J, Dunbar GC (1997). The MINI
International Neuropsychiatric Interview (MINI): a short
structured diagnostic interview: reliability and validity
according to the CIDI. European Psychiatry 12, 224–232.
Michalopoulou G, Alevizaki M, Piperingos G, Mitsibounas D,
Mantzos E, Adamopoulos P, Koutras DA (1998). High
serum cholesterol levels in persons with ‘high-normal’ TSH
levels: should one extend the definition of subclinical
hypothyroidism? European Journal of Endocrinology 138,
141–145.
Nemeroff CB (1989). Clinical significance of
psychoneuroendocrinology in psychiatry: focus on the
thyroid and adrenal. Journal of Clinical Psychiatry 50 (Suppl.),
13–20.
Ordas DM, Labbate LA (1995). Routine screening of thyroid
function in patients hospitalized for major depression or
dysthymia? Annals of Clinical Psychiatry 7, 161–165.
Parle JV, Franklyn JA, Cross KW, Jones SC, Sheppard MC
(1991). Prevalence and follow-up of abnormal thyrotrophin
(TSH) concentrations in the elderly in the United Kingdom.
Clinical Endocrinology (Oxford) 34, 77–83.
Prange AJ (1996). Novel uses of thyroid hormones in patients
with affective disorders. Thyroid 5, 537–543.
Sawin CT, Castelli WP, Heshman JM, McNamara P,
Bacharach P (1985). The aging thyroid. Thyroid deficiency
in the Framingham study. Archives of Internal Medicine 145,
1386–1388.
Spencer CA, LoPresti JS, Patel A, Guttler RB, Eigen A, Shen
D, Gray D, Nicoloff JT (1990). Applications of a new
chemiluminometric thyrotropin assay to subnormal
measurement. Journal of Clinical Endocrinology and
Metabolism 70, 453–460.
Sternbach HA, Gold MS, Pottash AC, Extein I (1983).
Thyroid failure and protirelin (thyrotropin-releasing
hormone) test abnormalities in depressed outpatients.
Journal of the American Medical Association 249, 1618–1620.
Sullivan PF, Wilson DA, Mulder RT, Joyce PR (1997). The
hypothalamic–pituitary–thyroid axis in major depression.
Acta Psychiatrica Scandinavica 95, 370–378.
Szabadi E (1991). Thyroid dysfunction and affective illness.
British Medical Journal 302, 923–924.
Whybrow PC, Prange AJ (1981). A hypothesis of
thyroid–catecholamine receptor interaction. Archives of
General Psychiatry 38, 106–113.
