The association between congenital hypothyroidism and profound mental retardation has been recognised for over a century, and the extraordinary influence of thyroid hormones on developing nervous systems has been widely studied (1, 2). By contrast, although the relationship between thyroid disease and disturbances of mood and cognition has been noted clinically (3–5), early functional studies suggested that the mature mammalian brain may not be a target site for thyroid hormone. This may be traced to early reports suggesting that oxygen consumption in the mature human brain did not change with thyroid status (6, 7), and to the lack of suitable technology available in the 1950s and 1960s. With the rapid advances in basic science and methodological techniques over the past 25 years, however, there have been dramatic changes in the concepts of thyroid hormone action in the adult brain (8). Although no direct methods for in vivo measurement of brain thyroid metabolism exist, functional brain imaging techniques to evaluate cerebral blood flow and metabolism have offered some promising insights into the thyroid–brain relationship. It is now widely accepted that thyroid hormone continues to play a critical role in the adult brain, influencing mood and cognition, although the details remain to be elucidated.
Thyroid hormones play a critical role in the metabolic activity of the adult brain, and neuropsychiatric manifestations of thyroid disease have long been recognised. However, it is only recently that methodology such as functional neuroimaging has been available to facilitate investigation of thyroid hormone metabolism. Although the role of thyroid hormones in the adult brain is not yet specified, it is clear that without optimal thyroid function, mood disturbance, cognitive impairment and other psychiatric symptoms can emerge. Additionally, laboratory measurements of peripheral thyroid function may not adequately characterise central thyroid metabolism. Here, we review the relationship between thyroid hormone and neuropsychiatric symptoms in patients with primary thyroid disease and primary mood disorders.
The hierarchy of the hypothalamic-pituitary thyroid (HPT) axis
The follicular cells of the thyroid gland secrete primarily thyroxine (T4), the precursor to the biologically active form of thyroid hormone, triiodthyronine (T3). Synthesis and secretion of thyroid hormone are regulated by a negative feedback system that involves the hypothalamus, pituitary and thyroid gland (the HPT axis) (9). The rate of thyroid hormone synthesis is enhanced by the pituitary thyrotropin-stimulating hormone (TSH), which is stimulated by hypothalamic thyrotropin releasing hormone (TRH) with T3 and T4 as negative feedback regulators. In healthy individuals, about 80% of plasma T3 is produced outside of the thyroid gland with the remaining 20% secreted directly by the thyroid (10). Furthermore, rat experiments have established that the amount of T3 derived from the plasma or from local conversion of T4 varies among tissues. In the cerebral cortex, about 80% of T3 is derived from local tissue conversion (11, 12).
Most T4 is transported into the brain by a carrier-mediated process involving hormone transporters including transthyretin (TTR; 13, 14). However, studies in rodents have shown that although TTR regulates T4 concentrations in the choroid plexus there is no reduction of T3 and T4 in the brain parenchyma in a TTR-null mouse model, thus emphasising the importance of additional transporting systems (15). In contrast to peripheral tissue where T4 concentrations usually far exceed those of T3, in the brain T4 and T3 concentrations are in an equimolar range (16), and the levels of T3 within the brain are tightly controlled within narrow limits even under adverse conditions (17–19). In the brain, most T3 is formed by local tissue conversion of the T4 precursor by deiodination (20). In the brain, deiodination is associated with a differential temporal and spatial expression of types II (D2) and III (D3) deiodinase isoenzymes (21, 22).
D2 is primarily expressed in glial cells (tanycytes and astrocytes) of various regions of the central nervous system (CNS) and plays an important role in mediating thyroid hormone action both during CNS development and in the adult brain. D2-knockout mice show increased serum T4 and TSH levels, decreased brain T3 concentrations but no changes in serum T3 levels. Additionally, the activity of cerebral D3 in D2-knockouts was twice that of wild-type adult mice. However, compared to a hypothyroid mouse model, newborn D2-knockouts displayed normal to only mildly reduced mRNA-levels of T3 responsive genes, a finding that indicates the existence of compensatory mechanisms of the CNS for the lack of D2 (23–25). T4 is converted to T3 and made accessible to thyroid responsive neurones either by a proposed but not yet clearly defined astrocytic paracrine route or by release of T3 into the cerebrospinal fluid (CSF) by means of a yet to be determined pathway. In the adult rodent brain, the highly T3 responsive D3 deiodinase isoenzyme is primarily expressed in neurones throughout the entire CNS with some evidence for region-specific expression in the hippocampus and neocortex. During development, D3 expression is more restricted to brain regions involved in sexual differentiation. Due to its neuronal localisation, it appears to counterbalance excessive neuronal T3 availability by deiodinating T3 (26, 27). In adult human postmortem brain samples, it was shown that D3 appears to be differentially expressed with high levels of activity in the hippocampus and temporal cortex and lower levels of activity in parietal and frontal cortices, diencephalon and mesencephalon. Additionally, an inverse relationship between D3 activity and local T3 content could be demonstrated (28). Conversely, D3 activity appears to be the highest in the cerebellum, diencephalon, mesencephalon and hippocampus in the developing human brain (29).
Because both the deiodinases and the T3 target receptors are located intracellularly, the action and metabolism of thyroid hormone are intracellular events that require entry across the cell membrane via plasma membrane carriers (14, 22–30). Recently, two major representatives of such carriers have been characterised in the CNS of both rodents and humans: the monocarboxylate transporter (MCT8) and the organic anion-transporting polypeptide (OATP1C1). The expression, however, is not strictly limited to the CNS. Both transporters were also found in other tissues (31). MCT8 in the brain is expressed predominantly in the choroid plexus of the ventricles as well as in neurones of the neo- and allo-cortex, the hypothalamus and the folliculostellate cells of the pituitary gland. It is a highly specific transporter for T3 across cytoplasmic membranes (32–35). In humans, MCT8 mutations can lead to a severe X-linked psychomotor retardation by inhibiting the entry of T3 into neurones, thus emphasising an important role for MCT8 and thyroid hormones in the development of the CNS (33, 36). OATP1C1, another high affinity thyroid hormone transporter that is widely expressed in the CNS, has been so far less well characterised (31, 37). It belongs to a family of organic anion transporters expressed by brain capillary endothelial cells and cells of the choroid plexus. In rodents, OATP1C1 is responsible for a preferential transport of T4 and rT3 across the blood–brain barrier (38–40).
The complex feedback regulation of thyroid hormone action has also been demonstrated in the human hypothalamus. In post mortem brain samples, D2 enzyme activity coincide with glial D2 expression (as assessed by immunohistochemistry) was detected in the infundibular nucleus/median eminence region and in tanycytes of the third ventricle. TRH neurones in the paraventricular nucleus expressed MCT8, D3 and thyroid hormone receptors (TR). Hence, central feedback regulation is proposed to rely on the local glial uptake of T4 and its conversion to T3 by D2. Subsequently, the biologically active T3 is suggested to be transported to hypothalamic TRH-neurones where T3 either binds to intracellular TR or is inactivated by neuronal D3. The exact mechanisms of this balance are still under investigation (41). Both in patients with major depression and in glucocorticoid-treated patients, a downregulation of TRH-mRNA in hypothalamic paraventricular neurones could be detected, resulting in lower levels of TSH (42, 43). One possible explanation could be that local changes in thyroid hormone metabolism resulting from altered hypothalamic deiodinase or MCT8 expression in patients with major depression or hypercortisolism are a potential cause of the decreased TRH-mRNA expression (34, 41).
In summary, the intracellular T3 concentration in the brain is determined by a complex interplay of factors, including the circulating levels of T4 and T3, the activity of transporters mediating cellular influx and efflux, and the activity of the deiodinases (30–40, 44). Although plasma membrane transport of T4 may have a strong effect on the rate of T3 production (14), the regulatory mechanisms controlling the bioavailability of T3 in the brain are not yet satisfactorily clarified.
The biological actions of thyroid hormones are initiated by the intracellular binding of T3 to nuclear receptors, which are part of the nuclear superfamily of ligand-modulated transcription factors that includes receptors for steroid hormones, vitamin D, and retinoic acid (45). These nuclear thyroid hormone receptors (TRα and TRβ in diverse isoforms) are widely distributed in the adult brain, with higher densities in phylogenetically younger brain regions (e.g. amygdala and hippocampus) and lower densities in the brain stem and cerebellum (46–48). The actions of T3 are mediated through the control of gene expression after interaction with the nuclear receptors, which are associated with regulatory elements in the promoter region of target genes (49). The binding of T3-receptor complex induces the release of co-repressors, resulting in activation of brain transcription machinery and usually increased gene expression (20, 50). Genes that are controlled by thyroid hormones are known to encode proteins of myelin, neurotrophins and their receptors, transcription factors, splicing regulators and proteins involved in intracellular signalling pathways.
Today, several thyroid hormone receptor mutant mice exist, that elucidate the effects of TRα and TRβ on brain function expressed in behavioural disturbances (51). TRα1-knockout mice show abnormalities in open field and fear conditioning tests with reduced exploratory behaviour and a higher freezing response respectively. In these mice, fewer GABAergic terminals on CA1 pyramidal neurones were demonstrated. These findings imply a role of TRα1 in hippocampal structure and function (52). Additionally, heterozygous mice harbouring a TRα1-mutation, leading to a ten-fold reduction of thyroid hormone affinity, display extreme anxiety-like behaviour and a reduced recognition memory, again implying a dysfunctional hippocampal circuitry (53). They also show locomotor dysfunction as a result of cerebellar disturbances. Both the hippocampal and cerebellar behavioural alterations could be treated by high-dose thyroid hormone treatment. However, the cerebellar dysfunction responded to T3 only, when it was applied postnatally, whereas the hippocampal dysfunction could be relieved by an adult age T3 treatment alone (53). Whereas no human mutations in the TRα are known so far, a human syndrome called ‘resistance to thyroid hormone’ (RTH) is caused by mutations in the ligand-binding domain of the TRβ gene. Interestingly, RTH is often associated with attention-deficit hyperactivity disorder (ADHD) (54). In line with this observation, transgenic mice bearing human mutant TRβ1 genes displayed enhanced locomotor activity in an open field and impaired learning of an autoshaping task, thus mimicking at least some of the behavioural abnormalities encountered in patients with RTH/ADHD (55, 56).
Recently, nongenomic actions of thyroid hormones have also been described in the brain and peripheral tissues. After binding to cytoplasmic thyroid hormone receptors, T3 appears to be able to rapidly activate the PI3K-Akt signal transduction cascade in conditions such as experimental stroke or during neuronal development (57, 58). The exact role of this alternative faster signalling mechanism, its regulation and its relationship to the slower genomic action pathways of thyroid hormones, remains to be determined in future studies.
Thyroid hormones and mood: potential mechanisms of action
Thyroid hormone receptors are widely distributed in the brain. Many of the limbic system structures where thyroid hormone receptors are prevalent have been implicated in the pathogenesis of mood disorders. However, the cellular and molecular mechanisms underlying these metabolic effects, and the specific neuropharmacological basis and functional pathways for the modulatory effects of thyroid hormones on mood, are yet to be understood. Interactions of the thyroid and neurotransmitter systems, primarily norepinephrine and serotonin, which are generally believed to play a major role in the regulation of mood and behaviour, may contribute to the mechanism of action in the developing and mature brain (59–63). There is ample evidence, particularly from animal studies, that the modulatory effects of thyroid hormones on the serotonin system may be due to an increase in serotonergic neurotransmission, by a reduction of the sensitivity of 5-HT1A autoreceptors in the raphe nuclei and increase in 5-HT2 receptor sensitivity (62). Thyroid hormones also interact with other neurotransmitter systems involved in mood regulation, including dopamine post-receptor and signal transducing processes, as well as gene regulatory mechanisms (45, 64–66). Furthermore, within the CNS, the regulatory cascade through which the thyroid hormones, particularly T3, exert their effects is not well understood: deiodinase activity, nuclear binding to genetic loci and, ultimately, protein synthesis may all be involved. Other proposed mechanisms for thyroid involvement in the aetiology of mood disorders include disturbances or reactive hyperactivity in the HPT axis, as manifested in the blunted TSH response to TRH found in some patients with depression (67–70).
Neuropsychiatric changes in thyroid disorders
Disturbances of thyroid metabolism in the mature brain may profoundly alter mental function, influencing cognition and emotion. The most frequently occurring thyroid diseases of adult life are autoimmune disorders, with autoimmune (Hashimoto’s) thyroiditis being the most frequent cause of hypothyroidism (inadequate hormone production), and Graves disease being the most frequent cause of hyperthyroidism (excess hormone production). There are three important antibodies that are involved in thyroid autoimmunity: thyroglobulin antibodies, thyroid peroxidase (TPO) antibodies and thyroid-stimulating hormone-receptor antibodies. Clinically, testing for TPO antibodies may be required to confirm an autoimmune cause for thyroid disease (71). Both hyperthyroidism and hypothyroidism are associated with changes in mood and intellectual performance; and severe hypothyroidism can mimic melancholic depression and dementia (72–74). The neurocognitive impairments accompanying dysfunction of the thyroid gland are usually reversed rapidly following return to euthyroid hormone status, although severe hypothyroidism, if left untreated, may rarely result in irreversible dementia (75, 76).
Cognitive changes relating to thyroid disease
Although patients with hyperthyroidism frequently report cognitive symptoms that even persist beyond the acute phase (77), impairments have only been found inconsistently (78, 79). By contrast, cognitive changes have frequently been detected in patients with hypothyroidism, including defects ranging from minimal to severe in general intelligence, psychomotor speed, visual–spatial skills and memory (4, 72, 80–83). Several recent studies have suggested that hypothyroid-related memory defects are not attributable to an attentional deficit but rather to specific retrieval deficits (83–86). Motor skills, language, inhibitory efficiency, and sustained attention appear to be less impacted by hypothyroidism (81–83). Thus, the memory deficit found in hypothyroidism appears to be distinct from that associated with major depression, in which patients typically experience broad executive difficulties (86). However, older adults may be more vulnerable to cognitive changes due to thyroid failure (81, 87).
There is an ongoing controversy as to whether subclinical hypothyroidism is associated with cognitive impairment. Subclinical hypothyroidism is characterised by a serum TSH level elevated above the statistically defined upper limit of the reference range, in association with a serum free T4 level within the reference range (88).
No association was found between subclinical hypothyroidism and measures of cognition in some studies (81, 89–92), whereas, in other studies, patients with subclinical hypothyroidism performed worse than normal controls on neuropsychological tests including the Wechsler Adult Intelligence Scale, the Wechsler Memory scale and in verbal fluency (93–95). Most cognitive deficits measured in patients with subclinical hypothyroidism are minimal in severity. A functional magnetic resonace imaging study found that working memory, but not other memory functions, was impaired in patients with subclinical hypothyroidism, and these impairments were reversible with L-thyroxine (L-T4) treatment (96). A systematic review of the literature regarding the risks associated with subclinical hypothyroidism, including cognitive deficit, was inconclusive (97).
There is also controversy as to whether the cognitive and mood symptoms of thyroid disease, particularly hypothyroidism, are completely reversible with normalisation of thyroid levels. A population based study comprising 37 000 individuals did not find any relation between depression and anxiety and prior thyroid disease (98). In another population based study comprising 30 000 individuals, depression and anxiety were not associated with current thyroid dysfunction, but were associated with prior thyroid disease (99). In a register-based study of 165 000 patients who have been hospitalised for hypothyroidism, there was an increased risk for hospitalisation with depression or bipolar disorder especially within the first year (100). Some recent community and controlled studies also suggest that a subset of patients experience some level of persistent psychological impairment (101–103). Although this may be due to inadequate replacement therapy, no relationship was found between current TSH concentration and symptoms (103, 104), although this issue remains controversial (105). Furthermore, the pathogenesis of the thyroid disease may influence the persistence of impairment (106, 107). Evidence from both postnatal women and those with Hashimoto’s encephalopathy suggests that thyroid autoimmunity may be an independent risk factor for depression or cognitive impairment (106, 108–110). Hashimoto’s encephalopathy is a rare syndrome of encephalopathy and high serum TPO antibodies that is not due to current hypothyroidism (110).
The potential association between Alzheimer’s disease and thyroid disease has also been investigated and found to be ambiguous. Although a relationship between hypothyroidism or thyroid hormone and Alzheimer’s disease has been reported (111, 112), many recent studies of the elderly have not found any relation between TSH levels and the risk of Alzheimer’s disease (91, 113, 114). Indeed, in a study of 599 patients aged > 85 years, patients with abnormally high levels of TSH were found to have a prolonged life span (91). Additionally, there have been several reports that subclinical hyperthyroidism increases the risk for Alzheimer’s disease (115, 116). Furthermore, patients with Alzheimer’s disease were found to have increased levels of T3 in cerebral spinal fluid (117). However, a prior study found a negative association between Graves disease and Alzheimer’s disease (118), and no link was found between the presence of thyroid antibodies and Alzheimer’s disease (119).
Neuropsychiatric symptomatology in thyroid disease
Hyperthyroidism or thyrotoxicosis is accompanied by psychiatric symptoms, including dysphoria, anxiety, restlessness, emotional lability, and impaired concentration. In elderly patients, depressive symptoms such as apathy, lethargy, pseudodementia and depressed mood can also occur (120). Approximately 60% of thyrotoxic patients have an anxiety disorder and between 31% and 69% have a depressive disorder (121, 122). However, overt psychiatric illness only occurs in approximately 10% of thyrotoxic patients (123). Patients developing mania when in a thyrotoxic state commonly have an underlying mood disorder or positive family history (124–126).In hypothyroid patients, depression-like symptoms including psychomotor retardation, decreased appetite, fatigue, and lethargy often occur (127). Neurocognitive dysfunction and depression as well as impaired perception with paranoia and visual hallucinations may develop, and severe hypothyroidism mimics melancholic depression and dementia (72, 73).
Brain perfusion abnormalities in thyroid and mood disorders
There are only a limited number of recent functional imaging studies of patients with thyroid disorder. These studies include patients with hypothyroidism of varying levels of severity from autoimmunity or thyroid cancer, and generally employed single photon emission computed tomography or positron emission tomography. The most consistent finding from studies of patients with hypothyroidism is global, diffuse hypoperfusion (128–135). Several studies found the perfusion deficits were most pronounced in posterior brain regions (131, 132, 134) or in the parietal lobe (130). The degree to which these perfusion abnormalities reverse with treatment remains unclear. In several studies, some degree of normalisation of perfusion was reported when patients became euthyroid (128, 129, 133, 135). One study of patients with previously untreated mild hypothyroidism found reversible hypoperfusion in the subgenual and perigenual anterior cingulate cortex, posterior cingulate cortex, amygdala and hippocampus (135). In other studies (131, 132, 134), the hypoperfusion remained evident after initiation of the thyroxine replacement therapy, although this finding did not predict the outcome of long-term treatment (134).
Some investigators have suggested that the perfusion patterns in hypothyroid patients may resemble those found in patients with early-stage Alzheimer’s disease, with selective regional hypoperfusion in posterior brain areas (136). The posterior cingulate cortex is the brain region most significantly decreased in the earliest stages of Alzheimer’s disease (137), a condition marked by irreversible decline in short-term memory abilities. In a few studies of patients with dementia from Hashimoto’s encephalopathy, there was widespread and diffuse cerebral perfusion with localisation to tempoparietal regions, but without specific sparing of the medial temporal lobe (128, 129, 138, 139). Diffuse perfusion deficits have also been demonstrated in patients with hyperthyroid encephalopathy (140, 141). Moreover, in the limited data from patients with autoimmune encephalopathy, normalisation of diffusion deficits after clinical recovery has been reported (128, 141). The role of autoimmunity in the development of cerebral perfusion abnormalities in patients with thyroid disease is unclear. In two studies of asymptomatic, euthyroid patients with autoimmune thyroiditis and high titers of TPO antibodies, there was a high prevalence of mild brain perfusion abnormalities (142, 143), whereas no perfusion abnormalities were found in patients with nontoxic nodular goitre (143). This lack of relation between thyroid hormone levels and perfusion abnormalities suggests that there may be a continuing CNS involvement related to autoimmunity (142, 143).
There are many more imaging studies of patients with primary mood disorders, especially major depressive disorder. The most consistent finding from studies of patients with depression is frontal lobe hypoperfusion (144–146). More specifically, ventromedial frontal regions may display increased perfusion whereas more caudal regions such as anterior cingulate and dorsolateral prefrontal cortex may show decreased perfusion, but the results are variable (144, 146, 147). However, involvement of cortical, paralimbic and subcortical regions is generally seen across studies. Differences in the diagnosis and symptoms of patients with depression may contribute to the diversity of findings. In patients with major depressive disorder, activity in subcortical ventral, ventral prefrontal and limbic structures correlates positively with the severity of symptoms, whereas dorsal cortical structures correlate negatively (148). Normalisation of perfusion abnormalities in patients with depression has been reported after a response to a variety of treatments including antidepressants, electroconvulsive therapy, repetitive transcranial magnetic stimulation and psychotherapy, but not in all patients (144, 147). Cortical deficits are most likely to normalise with treatment, while paralimbic and subcortical regions show a more complex state-trait pattern (147). In a study of depressed patients with bipolar disorder who received supraphysiological dosages (300 μg/day) of T4 significant, mood improvement was accompanied by normalisation of cerebral metabolic perfusion in frontal, limbic and subcortical regions (149). An inverse correlation was found between levels of TSH and regional cerebral blood flow in patients with mood disorders and normal or mildly elevated TSH levels (150).
In a direct comparison of the perfusion changes found in depression and hypothyroidism, patients with hypothyroidism showed diffuse hypoperfusion with posterior localisation, whereas depressed patients showed hypoperfusion in the anterior parts of the brain (134). The results reported for the hypothyroid and depressed patients were consistent with those described previously. Although there was overlapping areas of hypoperfusion in pre- and post-central gyri and the inferior occipital, gyrus, different circuits may be involved in the behavioural symptoms expressed in depression and primary hypothyroidism (134).
Thyroid status in patients with mood disorders
Several lines of evidence suggest that there may be abnormalities in thyroid hormone metabolism in patients with mood disorders, and that these may not be readily apparent with the standard tests to screen for thyroid disease. The vast majority of patients with depression do not have overt thyroid disease, although subclinical hypothyroidism has been detected in approximately 15% of patients (151, 152). By contrast, the prevalence of subclinical hypothyroidism in the US adult population without known thyroid disease is approximately 4–8% (97), although the prevalence increases with age and in women, such that subclinical hypothyroidism is present in up to 20% of women aged over 60 years (97). In patients with depression, the time to recurrence of episodes was found to be inversely correlated to serum T3 but not T4 levels (153). Investigators have found an abnormal TSH response to TRH stimulation in approximately 25–30% of depressed patients, with a peak response that was blunted (154). However, the TRH stimulation test seems to have low sensitivity, specificity and usefulness in a clinical psychiatric setting mainly due to the heterogeneity of the depressed patients under investigation (155–157). The most common finding in patients with depression is an elevated T4 level that falls with treatment (158). This decrease in peripheral T4 serum concentrations during treatment has been reported after a variety of agents, including antidepressants, carbamazepine and lithium (159–161), and after response to nonpharmacological treatments, including sleep deprivation, light therapy (162), electroconvulsive therapy and psychotherapy (163). In general, little is known about the exact interactions between antidepressants and the thyroid hormone system so far. However, in view of an increasingly better understanding of thyroid hormone action regulation on a molecular level (31, 42), this question can be addressed more appropriately in the future. In the rat, treatment with various antidepressants (fluoxetine, tranylcypromine, mianserine) results primarily in changes of local D2-activities and to a lesser extent of D3-activities. However, no clear correlation between D2- or D3-activity changes and changes of local thyroid hormone concentrations that were also altered after treatment with antidepressants could be found, indicating additional pathways of iodothyronine metabolism in the brain (164). Additionally, treatment with antidepressants (desipramine, paroxetine, venlafaxine, tianeptine) in the rat results in a rather specific increase of T3 in the myelin fraction of homogenates of the amygdala, an essential structure implicated in emotion and fear regulation (165). Another new line of research might also be found in the role antidepressants are assumed to play in the hippocampal adult neurogenesis based model of depression (166) because thyroid hormones have been shown to influence adult hippocampal neurogenesis in the rat (167–169).
A clinical study reported lower levels of CSF transthyretin in depressed patients than in healthy controls despite normal peripheral blood thyroid hormone measures (170), suggesting a limitation in the uptake of T4 into the brain via transthyretin, which is synthesised in the choroid plexus. However, in a transthyretin null mice strain, the complete absence of transthyretin has no impact on thyroid hormone levels, development, or fertility (15, 171). These findings emphasise the importance of other membrane carrier systems such as MCT8 or OATPs to maintain thyroid hormone homeostasis in the brain (14, 172). Nevertheless, reduced transthyretin CSF levels might still reflect a state or trait marker of patients suffering from depression and its exact relationship to other transporter systems both in healthy and depressed populations remains to be addressed in further studies (170, 173).
There is also growing evidence of thyroid abnormalities in patients with bipolar disorder. Patients receiving lithium prophylaxis who have free T4 levels in the low–normal range may experience more affective episodes (174). Although within the normal range, a lower free thyroxine index and higher TSH were significantly associated with a poorer treatment response during the acute depressed phase of bipolar disorder (175). Furthermore, the free T4 index was inversely related to the hospital length of stay in males with affective disorders (176).
TPO antibodies were reported to be elevated in bipolar disorder with a prevalence of 28% (177), whereas results from other studies were inconsistent with reported rates in the range 0–43% (178–181). In community studies, the rates of prevalence of TPO antibodies generally range from approximately 12–18% (182–185). The estimate of TPO antibody prevalence will vary with the sensitivity and specificity of the testing methodology (186), is increased in females, in old age (182, 184, 185), when TSH levels are abnormally high or low (187, 188), and when individuals with known thyroid disease are included in the population. The impact of increasing age and female gender on the detection of TPO antibodies has also been noted in patients with affective disorders (181). Elevated thyroid antibodies were also reported in some studies of patients with unipolar depression (106, 152, 189), but not in others (180, 190). It was also hypothesised that autoimmune thyroiditis, with TPO antibody as marker, may be a potential endophenotype for bipolar disorder, and is related to the genetic vulnerability to develop bipolar disorder rather than to the disease process itself (191). Although the offspring of parents with bipolar disorder were found to have increased vulnerability to develop thyroid autoimmunity compared to high-school aged controls, this was independent of any psychiatric disorder or symptoms (192). Thyroid antibody status was also associated with an increased risk for lithium-induced hypothyroidism, but not with current or former lithium treatment (177).
Treatment of psychological symptoms in thyroid disorders
In the majority patients with thyroid disease, psychological symptoms are reversed with the restoration of euthyroid status. According to The National Academy of Clinical Biochemistry guidelines (NACB), the currently recommended treatment for hypothyroidism is L-T4 monotherapy, titrated optimally such that the serum TSH level is in the range 0.5–2.0 mIU/l (186). However, although serum TSH levels reflect the feedback effect of thyroid hormones on the hypothalamic–pituitary level, it is now recognised that no single measure is likely to accurately reflect the thyroid hormone concentration in all tissues (193). It is also acknowledged that L-T4 replacement therapy in hypothyroid patients cannot precisely duplicate the physiological function of thyroid hormones. In thyroidectomised rats, only a combined infusion of the T3 and T4 could restore normal levels of circulating levels of T3, T4 and TSH, and tissue levels of T3 and T4 (193). However, when combination T3 and T4 replacement therapy was investigated in patients with hypothyroidism, only one of nine controlled studies found any improvement in neurocognitive function, although patients in several studies preferred the combination (193). In another open label, nonrandomised clinical study, patients with hypothyroidism only reported psychological well-being when taking approximately 50 μg of T4 greater than that required to normalise TSH (194), at a dosage associated with an increased risk of osteoporosis or atrial fibrillation (97). However, these results could not be reproduced in a double-blind randomised clinical trial with a crossover design performed in 56 subjects with primary hypothyroidism (104). The authors concluded that small changes in T4 dosage do not produce measurable changes in hypothyroid symptoms, well-being or quality of life, thus indicating a possible placebo effect in the open-label study.
The severe neurocognitive impairments found in Hashimoto’s encephalopathy may be reversed in association with but not necessarily because of glucocorticoid treatment (110).
Thyroid hormones in the treatment of mood disorders
Because of the relationship between thyroid disease states and psychiatric symptoms, there has long been an interest in using thyroid hormones to treat mood disorders. In the 1930s, Norwegian physicians used desiccated sheep thyroid gland to treat patients with cyclic mood disorders (195). Although thyroid hormone monotherapy is not an adequate treatment for patients with primary mood disorders, since the late 1960s (196), a series of open and controlled clinical trials have confirmed the therapeutic value of adjunctive treatment with thyroid hormones in mood disorders. Specifically, there is good evidence that T3 can accelerate the therapeutic response to tricyclic antidepressants (197) and, in treatment-resistant depression, T3 may augment the response to tricyclic antidepressants, although the results have been inconsistent (198, 199). T3 has also been shown to augment the response to sertraline (200) but not to paroxetine (201).
In a series of open-label studies, adjunctive treatment with supraphysiological doses of L-T4 was found to be effective in the maintenance treatment of patients with severe rapid cycling or resistant bipolar disorder who did not respond to standard measures (67, 68, 149, 201–203). Supraphysiological L-T4 may also have immediate therapeutic value in antidepressant-resistant bipolar and unipolar depressed patients during a phase of refractory depression (204, 205). In these patients with malignant affective disorder, doses of 250–600 μg/day L-T4 are required to achieve therapeutic effect, which is much higher than those used in the treatment of primary thyroid disorders. Although treatment with supraphysiological T4 requires close monitoring, the hyperthyroxinemia is tolerated surprisingly well. No serious effects, including loss of bone mineral density, were observed even in patients treated for extended periods (199, 202, 206–208). The low incidence of adverse effects and high tolerability reported by patients with affective disorders who are receiving high-dose thyroid hormone therapy contrasts with that typically seen in patients with primary thyroid disease. For example, patients with thyroid carcinoma treated with high doses of L-T4 to achieve suppression of TSH commonly complain of the symptoms of thyrotoxicosis. Furthermore, total thyroxine, free thyroxine, and total triiodthyronine levels in depressed patients were less elevated in response to supraphysiological doses of L-T4 than in healthy controls (209, 210). This could be explained by the hypothesis that, in unipolar depression, T4 is to a greater extent metabolised into inactive compounds such as rT3 compared to in healthy subjects. Support for this hypothesis stems from older studies that describe elevated rT3 serum and CSF concentrations in depressed patients (211–213). Clearly, this issue merits further investigation, as well as with respect to a possible causal involvement of the deiodinase D3 in these findings (28).
Changing perspectives of normal thyroid laboratory values
There have been remarkable changes in the measurement of thyroid hormone status over the last several decades. The most diagnostically sensitive test to detect thyroid disease, and the primary screening test for both subclinical and overt hypothyroidism and hyperthyroidism, is serum TSH (186). During the past 20 years, the upper limit for the reference range for TSH has decreased from approximately 10–4.0 mIU/l, reflecting primarily improvements in the sensitivity and precision of the TSH assays. Reference limits will be reduced even further in the future because, with vigorous screening for personal or family history of thyroid disease, 95% of individuals are found to have a serum TSH value in the range 0.4–2.5 mIU/l (186). Furthermore, exclusion of individuals who are positive for TPO will further tighten the reference range (186, 214). In a study (NHANES) of 16 000 individuals aged > 12 years ethnically representing the US population, the mean normal TSH values were only in the range 1.18–1.4 mIU/l (215). Currently, the NACB suggests that the optimal therapeutic target for TSH is in the range 0.5–2.0 mIU/l (186). The recent changes in the statistical reference range for TSH can have a dramatic impact on the number of individuals considered potentially ‘at risk’ for neuropsychiatric symptoms. In a retrospective analysis of data from 75 000 patients without thyroid disease, a decrease in the upper limit of normal for TSH from 5.0 to 3.0 mIU/l would result in a four-fold increase in the number of patients with an elevated TSH level, increasing from 4.6% to 20.0% (216).
In comparison to the statistical reference range, within any one individual, the levels of TSH only fluctuate within a very narrow range in response to changing free T4 (217). This is consistent with findings from twin studies showing that each person has a genetically determined free T4-TSH set point (218, 219). Regarding this within-person variability, the NACB guidelines state that a change in serum TSH of 0.75 mIU/l would be required for clinical significance when monitoring a patient’s response to replacement thyroxine therapy (186). However, for an individual patient, the clinical significance of variation in the results of a thyroid function that is still within the reference range is not known (220).
There are also ongoing efforts to redetermine the cut-off ranges for measuring thyroid antibodies (221) and to standardise reference methods used to determine free thyroxine (222). Continued improvement in laboratory measurement techniques, and changing reference ranges, will help increase knowledge of normal thyroid hormone function, and may have significant clinical implications. Additionally, there is a need to better understand how current laboratory measures directly relate to thyroid hormone metabolism within the brain.
Thyroid hormones have a multitude of effects on the central nervous system, and it is now widely recognised that disturbances of mood and cognition often emerge in association with putative disturbance of thyroid metabolism in the brain. As knowledge in basic science and appropriate technology evolves, our understanding of the role of thyroid hormone function in the adult brain will continue to be refined. In patients with primary thyroid disorders, both excess and inadequate thyroid hormones can induce behavioural abnormalities that mimic depression, mania, and dementia. These neuropsychiatric impairments are generally reversible following return to euthyroid status, although some defects may persist in a subset of patients. In patients with primary mood disorders, thyroid hormones appear to be capable of modulating the phenotypic expression of their illness. Even though most patients with primary mood disorders do not have overt thyroid disease, relative abnormalities in thyroid function are associated with a worse outcome. Furthermore, the adjunctive use of supraphysiological doses of L-T4 in malignant affective disorders frequently provides remission without adverse physiological effects where all other treatments have failed.
Although the behavioural disturbances reported in studies of patients with primary thyroid disorders and those of patients with primary mood disorders may appear to be similar, any direct comparison, especially in postulating a common aetiology, is fraught with danger. Both groups comprise patients with disparate diagnoses, and with disease states of different levels of intensity and duration. Furthermore, a comparison of thyroid laboratory values obtained from studies performed in different decades can be problematical because of the use of assays of different sensitivity and specificity, a lack of standardisation, and the changing reference ranges. Hence, although patients with both primary mood disorders and primary thyroid disorders display similar neuropsychiatric symptoms and have involvement of central thyroid metabolism, the psychopathology may be due to diverse primary aetiologies. Different characteristic patterns for depression and hypothyroidism have been observed in functional imaging studies. There is great disparity between patients with primary thyroid disease and primary mood disorders with respect to the response to treatment with thyroid hormone, the long-term outcome in relation to neuropsychiatric symptoms, and the potential role of autoimmunity in the aetiology of the disease.
In conclusion, studies of the biology of thyroid hormone action show that these hormones play an important role in normal brain function and that current laboratory tests of thyroid status may not provide a sufficiently accurate measure of thyroid hormone function within the CNS. However, studies of patients with primary thyroid disease and primary mood disorders are less conclusive and the relationship of these conditions to central thyroid hormone function needs further research.