Introduction
The relationship between mood and thyroid function has been recognized since the 19^th century when it became clear that hypothyroidism is associated with a variety of mental disturbances including depression. Nevertheless, it has been only during the past decades that the complex interactions between thyroid hormone and the central nervous system have begun to be elucidated. Direct evidence for the importance of thyroid hormone for the metabolic integrity of the adult human brain was obtained only in the 1990s when studies using magnetic resonance spectroscopy and positron emission tomography in hypothyroid patients showed effects of thyroid hormone supplementation, especially in the frontal lobe, on regional cerebral blood flow, glucose utilization and additional metabolic parameters (1,2). This provided a biological basis for the wide variety of mental symptoms in hypothyroidism including disturbed mood and cognition. In view of these observations it is not surprising that various investigators have focused on the question whether in depressed patients without manifest pathology of the thyroid gland, changes occur in the hypothalamus-pituitary-thyroid axis (HPT axis). Most of these studies have focused on serum thyroid hormone concentrations, but more recently changes in the hypothalamus of patients with depression have been reported as well. Another approach has been the use of thyroid hormones as an adjunct treatment for affective disorders, adding thyroid hormone to antidepressant therapy in an attempt to increase treatment efficacy.

Thyroid hormone serum concentrations in major depression
Major depression has been associated with changes in the HPT axis. Various authors have reported decreased serum TSH with a somewhat blunted response to TRH stimulation, and increased serum thyroxine (T4) or free T4 (FT4). Most of the changes reported were still within the reference range for the particular serum concentration studied. Furthermore, an increased prevalence of subclinical hypothyroidism and thyroid peroxidase (TPO) antibodies has been described. These endocrine changes in major depression may have pathogenetic relevance as they have been proposed to reflect serotonin and/or norepinephrine deficiency which are prominent neurochemical targets for antidepressant therapy (3), or to reflect effects of hypercortisolism within the brain inducing mildly decreased cerebral bioavailability of thyroid hormone (4). However, several other studies did not find any changes in serum TSH and/or FT4 in patients with major depression (e.g., 5). These inconsistencies are probably due to a number of factors. First, studies reported in the literature differ with respect to the in-/outpatient status of the included patients, studies on inpatients being more numerous. This may be somewhat surprising since outpatients comprise the vast majority of patients with major depression. Second, patients were on antidepressant medication in several studies which may influence endocrine parameters even after short-term discontinuation (6). Third, the heterogeneity of depression is a potential bias. Several studies included unipolar as well as bipolar patients, or were designed without stratification for depression subtype such as melancholic depression.
In view of these inconsistencies and the potential clinical importance of HPT axis changes in depression we recently performed a study in 113 outpatients with unipolar major depression who had been free of antidepressants for at least three months and in 113 age- and sex matched controls. To be eligible for the study the patients had to be between 18 and 65 years of age and to fulfill the diagnostic criteria major depressive disorder according to the Structural Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders (DSM), fourth edition (SCID-IV). Patients were required to have a score of at least 16 on the 17-item Hamilton Rating Scale for Depression (HRSD) which is a commonly used questionnaire to rate depression severity. Surprisingly, the serum concentration of TSH was slightly higher in patients with major depression as compared to controls and this difference persisted after the exclusion of patients with TPO antibodies and/or subclinical hypothyroidism (median TSH 1.70 in major depression versus 1.50 mU/L in controls, respectively). There were no signs of activation of the hypothalamus-pituitary-adrenal axis in the depressed patients as evident from unaltered 24h urinary cortisol excretion (7). Dividing the patients according to depression subtype (e.g., melancholic depression versus other subtypes) or depression severity (e.g., HRSD lower versus higher that 23) did not unmask any endocrine changes in the depressed patients. Therefore, the somewhat higher serum TSH found in the present study in outpatients cannot be attributed to less severe depression in these patients as compared with inpatients. An alternative explanation may be altered circadian structure in depressed patients. Serum TSH shows a clear diurnal variation with a nocturnal TSH surge and this rhythm is driven by the biological clock located in the hypothalamic suprachiasmatic nucleus (SCN) (8). Indeed, changes in circadian rhythms as well as alterations in the SCN have been found in depressed patients (9).

Thyroid hormone and hypothalamic changes in major depression
Thyrotropin-releasing hormone (TRH)-containing neurons in the hypothalamic paraventricular nucleus (PVN) represent the major determinant of hypothalamic setpoint regulation of the HPT axis. In animal experimental studies these neurons have been shown to respond to hypothyroidism with increased TRH mRNA and protein content, thereby increasing TSH release from the anterior pituitary, with opposite changes seen in hyperthyroidism. Thyroid hormone status is sensed by thyroid hormone receptors (TRs) expressed by TRH cells in the PVN. An additional feedback pathway is represented by monosynaptic projections from the hypothalamic arcuate nucleus to TRH neurons in the PVN. In the arcuate nucleus, the blood brain barrier is absent and both TRs and iodothyronine deiodinase type 2 (D2) are abundantly expressed. This pathway has been shown to be implicated in mediating the effects of falling serum leptin concentrations on central downregulation of the HPT axis during starvation. Earlier studies from our group revealed both parvo- and magnocellular TRH neurons in the human PVN (10) and decreased TRH mRNA in direct relation with decreased serum T3 and TSH in the framework of non-thyroidal illness (11), suggesting an important role for TRH neurons in human HPT axis setpoint regulation.
In view of several reports of altered serum TSH in depression we performed a study of TRH mRNA expression in the PVN of patients with unipolar depression versus patients without psychiatric disease by quantitative in situ hybridization. Total TRH mRNA signal in the PVN was decreased in depression (figure) (12). Unfortunately, the distinction within the human PVN between neuroendocrine TRH cells projecting to the median eminence and centrally projecting TRH cells cannot be made at present.

Figure: (a.b) macroscopic photographs of film autoradiograms of representative sections through the PVN of a control subject and weaker hybridization signal in a subject with major depression, respectively. (c,d) TRH mRNA containing cells in the PVN of the same subjects. Bars represent 2 mm (a,b) and 50µm (c,d). OT= optic tract. (e,f) Total TRH mRNA hybridization signals in depressed subjects and controls. Subject pairs matched for severity of fatal illness are interconnected with lines. Each dot represents one subject. Note strongly decreased TRH hybridization signal in subjects with depression. Asterisk represents patient with primary hyperthyroidism and without depression. Reproduced with permission from (12)

CRH mRNA was shown earlier to be increased in the PVN of depressed patients (13) which suggests an increased hypothalamic drive of the HPA axis in depression. Since glucocorticoids have been shown to decrease TRH gene expression in the PVN of rats (14), we hypothesized that decreased TRH mRNA in depression might result, at least in part, from HPA axis activation. To obtain circumstantial evidence for this we assessed total TRH mRNA hybridization signal in the PVN of patients treated with pharmacological doses of glucocorticoids just before death and found a clear decrease as compared with untreated patients (15). Thus, our findings in postmortem studies are congruent with central, i.e., hypothalamic, activation of the HPA axis which may lead to decreased hypothalamic TRH stimulation of TSH from the anterior pituitary. Notably, these observations were made in deceased patients who had been admitted to a hospital before death which may explain the discrepancy with the paucity of endocrine changes found in depressed outpatients mentioned above.

Thyroid hormone and treatment of major depression
Selective serotonin reuptake inhibitors (SSRI) are currently favored in therapeutic practice as a first step in the treatment of major depression. SSRI increase the bioavailability of serotonin at the synapse, which is assumed to elicit the therapeutic response. There is evidence from animal experimental studies that triiodothyronine (T3) is also capable of increasing serotonergic neurotransmission by desensitization of inhibitory 5-HT1a autoreceptors in the raphe nucleus, thus disinhibiting cortical and hippocampal serotonin release, and by increasing cortical 5-HT2 receptor sensitivity, further increasing 5-HT neurotransmission (16). In addition, various antidepressants (including SSRI) stimulate D2 activity, theoretically leading to increased cerebral T3 concentrations, thus enhancing serotonergic neurotransmission (3). These considerations have led several investigators to perform clinical studies in patients with major depression using thyroid hormone in combination with antidepressants. A number of rather small clinical trials have been performed aimed at increasing efficacy of treatment with tricyclic antidepressants (TCA) by adding T3, showing inconsistent albeit encouraging results. A meta-analysis suggested that larger placebo-controlled studies should be performed including patients with the currently favored SSRI (17). Another approach has been to try and accelerate response to TCA by adding T3 which was shown to be effective (18).
We recently investigated the efficacy of T3 addition to an SSRI (paroxetine) in patients with major depression in a double-blind, placebo controlled clinical trial in 113 patients with major depression. Patients were randomly assigned to 8 weeks of treatment with low dose T3 (25 µg), higher dose T3 (25 µg twice daily), or placebo in addition to paroxetine 30 mg daily. Response rate after 8 weeks (defined as a reduction in HRSD score by =50%) was 46% in all three treatment arms (p=0.99). T3 addition did not accelerate clinical response to paroxetine, while patients on T3 addition reported more adverse events than patients on placebo co-medication (19). These results did not support a role for T3 addition to SSRI in the treatment of major depressive disorder. However, there still is a possibility that T3 co-medication is effective in patients with refractory depression.

Conclusion
HPT axis changes in major depression have been studied for many decades and results have been somewhat controversial. Inpatients have been reported to frequently exhibit hypercortisolism which may be explained in part by an increased hypothalamic drive of the HPA axis as suggested by postmortem studies. Hypercortisolism may explain recently reported decreased hypothalamic TRH mRNA expression in major depression, in turn giving rise to lower serum TSH. By contrast, outpatients with major depression exhibit only marginal endocrine alterations, including slightly higher serum TSH, which is not explained by less severe depression. To test the hypothesis that T3 can be used to further enhance serotonergic neurotransmission in patients with major depression treated with an SSRI, we performed the first randomized clinical trial to investigate the efficacy of T3 addition to paroxetine and found no effect on efficacy or on time interval to response. Whether T3 addition to antidepressants may be effective in refractory depression is subject of our current studies.

REFERENCES