A number of experimental paradigms have been used to mimic clinical situations that affect the hypothalamic-pituitary thyroid axis in man. However, with the exception of the studies of thyroid status and iodine deficiency, such perturbations have limited application to humans due to differences in the more subtle aspects of TSH regulation between species. For example, starvation is a severe stress and markedly reduces TSH secretion in rats, but only marginally in humans. Cold stress increases TSH release in the adult rat by α-adrenergic stimulation, while this phenomenon is usually not observed in the adult human. Thus, it is more relevant to evaluate the consequences of various pathophysiological influences on TSH concentrations in humans rather than to extrapolate from results in experimental animals. This approach has the disadvantage that, in many cases, the precise mechanism responsible for the alteration in TSH secretion cannot be identified. This deficit is offset by the enhanced relevance of the human studies for understanding clinical pathophysiology.
The concentration of TSH can now be measured with exquisite sensitivity using immunometric techniques (see below). In euthyroid humans, this concentration is 0.4-0.5 to 4.0-5.0 mU/L. This normal range is to some extent method-dependent in that the various assays use reference preparations of slightly varying biological potency. There is no crystalline human TSH preparation, so it is not possible to provide a precise molar equivalent for TSH concentrations.
Free α subunit is also detectable in serum with a normal range of 1 to 5 µg/L, but free TSH β is not detectable210. Both the intact TSH molecule and the a subunit increase in response to TRH. The α subunit is also increased in post-menopausal women; thus the level of gonadal steroid production needs to be taken into account in evaluating α subunit concentrations in women. In most patients with hyperthyroidism due to TSH-producing thyrotroph tumors, there is an elevation in the ratio of α subunit to total TSH. [14, 160, 244-246] In the presence of normal gonadotropin this ratio is calculated by assuming a molecular weight for TSH of 28,000 and for α subunit of 13,600 Da. The approximate specific activity of TSH is 0.2 mU/mg. To calculate the molar ratio of a subunit to TSH, the concentration of α subunit (in μg/L) is divided by the TSH concentration (in mU/L) and this result multiplied by 10. The normal ratio is <1.0 and it is usually elevated in patients with TSH-producing pituitary tumor but is normal in patients with thyroid hormone resistance unless they are post menopausal.
The volume of distribution of TSH in humans is slightly larger than the plasma volume, the half-life is about 1 hour, and the daily TSH turnover between 40 and 150 mU/day. [246] Patients with primary hypothyroidism have serum TSH concentrations greater than 5 and up to several hundred mU/L. [101] In patients with hyperthyroidism due to Graves' disease or autonomous thyroid nodules, TSH is suppressed with levels which are inversely proportional to the severity and duration of the hyperthyroidism, down to as low as <0.004 mU/L. [247-249]
TSH secretion in humans is pulsatile. [250-252] The pulse frequency is slightly less than 2 hours and the amplitude approximately 0.6 mU/L. The TSH pulse is significantly synchronized with PRL pulse: this phenomenon is independent from TRH and suggests the existence of unidentified underlying pulse generator(s) for both hormones. [253] The frequency and amplitude of pulsations increases during the evening reaching a peak at sleep onset, thus accounting for the circadian variation in basal serum TSH levels. [254, 255] The maximal serum TSH is reached between 21:00 and 02:00 hours and the difference between the afternoon nadir and peak TSH concentrations is 1 to 3 mU/L. Sleep prevents the further rise in TSH as reflected in the presence of increases in TSH to 5-10 mU/ml during sleep deprivation. [256, 257] The circadian variation of TSH secretion is probably the consequence of a varying dopaminergic tone modulating the pulsatile TSH stimulation by TRH. [258] There is little, if any, significant seasonal change in basal TSH nor are there any gender-related differences in either the amplitude or frequency of the TSH pulses. [252] The diurnal rythmicity o serum TSH concentration is maintained in mild hyper- and hypothyroidism, but it abolished in severe short-term primary hypothyroidism, suggesting that the complete lack of negative feedback to the hypothalamus or pituitary or both may override the central influences on TSH secretion. [259]
Age does not have a major effect on serum TSH with the exception of the extremes. [260] There is a marked increase in serum TSH in neonates which peaks within the first few hours of delivery returning towards normal over the next few days (see Chapter 15). It is thought to be a consequence of the marked reduction in environmental temperature at birth. Serum TSH concentrations in apparently euthyroid patients over the age of 70 may be somewhat reduced, [261, 262] although usually this is a pathological finding indicating either exogenous or endogenous thyrotoxicosis. [263] However, slightly decreased serum TSH has been also documented in elderly patients without clinical or subclinical hyperthyroidism, as assessed by 4 year follow-up, [262] and in healthy centenarians. [264] An age-dependent reduction of daily TSH secretion rate has been reported in humans. [265] The physiological nyctohemeral rhythm of TSH is maintained in the elderly, but the nocturnal peak is blunted. [266]
In the rat, starvation causes a marked decrease in serum TSH and thyroid hormones. While there is an impairment of T4 to T3 conversion in the rat liver due to a decrease in both thiol co-factor and later in the Type 1 deiodinase, [267-269] the decrease in serum T3 in the fasted rat is primarily due to the decrease in T4 secretion consequent to TSH deficiency. [270, 271] In humans, starvation and moderate to severe illness are also associated with a decrease in basal serum TSH, pulse amplitude and nocturnal peak. [272-276] In the acutely-fasted man, serum TSH falls only slightly and TRH responsiveness is maintained, although blunted. [277, 278] This suggests that the thyrotroph remains responsive during short-term fasting and that the decrease in TSH is likely due to changes due to decreased TRH release. There is evidence to support this in animal studies, showing reduced TRH gene expression in fasted rats. [279, 280] Administration of anti-somatostatin antibodies prevents the starvation induced serum TSH falls in rats, suggesting a role for hypothalamic somatostatinergic pathways. [281] However, fasting-induced changes in dopaminergic tone do not seem to be sufficient to explain the TSH changes. [275, 281]
Recent studies provides compelling evidence that starvation-induced fall in leptin levels (Fig. 4-11) plays a major role in the decreased TSH and TSH secretion of fasted animals and, possibly, humans. [223, 282, 283] This concept stems from the observation that administration of leptin prevents the starvation-induced fall of hypothalamic TRH. [284] The mechanisms involved in this phenomenon include decreased direct stimulation by leptin of TRH production by neurons of the PVN, [223, 285] as well as indirect effects on distinct leptin-responsive neuroendocrine circuits communicating with TRH neurons. [284, 286] The direct stimulatory effects of leptin on TRH production are mediated by binding to leptin receptors, followed by STAT3 activation and subsequent binding to the TRH promoter. [287, 288] One of the latter circuits has been identified in the melanocortin pathway, a major target of leptin action. This pathway involves 2 ligands expressed in distinct populations of arcuate nucleus neurons in the hypothalamus (the α-MSH and the Agouti receptor protein [AgRP]) and the melanocortin 4 receptor (MC4R) on which these ligands converge, but exert antagonistic effects (stimulation by α-MSH; inhibition by AgRP). Leptin activates MC4R by increasing the agonist α-MSH and by decreasing the antagonist AgRP and this activation is crucial for the anorexic effect of leptin. The specific involvement of melanocortin pathway in TRH secretion is suggested by the presence of α-MSH in nerve terminals innervating hypothalamic TRH neurons in rat[289] and human[290] brains and by the ability of α-MSH to stimulate and of AgRP to inhibit hypothalamus-pituitary thyroid axis both in vitro and in vivo. [285]The activities of α-MSH and AgRP on thyroid axis are fully mediated by MCR4, as shown by experiment carried out on MCR4 knock out mice. [291] Fasting may inhibit the hypothalamic-pituitary-thyroid axis also via the orexogenic peptide NPY, which inhibits TRH synthesis194.1 by activation of Y1 and Y5 receptors in hypophysiotropic neurons of the hypothalamic paraventricular nucleus. [292] At least two distinct population of NPY neurons innervate hypophysiotropic TRH neurons, [293] suggesting that NPY is indeed an important regulator of hypothalamic-pituitary-thyroid axis.
A further contributing cause of the decreased TSH release in fasting may be an abrupt increase in the free fraction of T4 due to the inhibition of hormone binding by free fatty acids. [294] This would cause an increase in pituitary T4 and, hence, in pituitary nuclear T3. Fasting causes a decrease in the amplitude of TSH pulses, not in their frequency. [295]
Ingestion of food results in an acute decline of serum TSH concentration: this is the consequence of meal composition, rather than stomach distension. [296] Long-term overfeeding is associated to a transient increase of serum T3 concentration and a sustained increased response of TSH to TRH. [297]
Taken together, the above data provide compelling evidence that hypothalamic-pituitary-thyroid axis is strictly related to the mechanisms involved in weight control. In keeping with this concept, preliminary epidemiological studies suggest that small differences in thyroid function may be important for the body mass index and the occurrence of obesity in the general population. [298]
The changes in circulating TSH which occur during fasting are more exaggerated during illness. In moderately ill patients, serum TSH may be slightly reduced but the serum free T4 does not fall and is often mildly increased. [294, 299-301] However, if the illness is severe and/or prolonged, serum TSH will decrease and both serum T4 (and of course T3) decrease during the course of the illness (see Chapter 5). This may be due to decreased pulse amplitude and nocturnal TSH secretion. [255, 302-304] Since such changes are short-lived, they do not usually cause symptomatic hypothyroidism. They are often associated with an impaired TSH release after TRH. [272] However, the illness-induced reductions in serum T4 and T3 will often be followed by a rebound increase in serum TSH as the patient improves. This may lead to a transient serum TSH elevation in association with the still subnormal levels of circulating thyroid hormones and thus be mistaken for primary hypothyroidism. [305] On occasion, transient TSH elevation occurs while the patient is still ill. The pathophysiology of this apparent thyroid gland resistance to TSH is not clear, [306] although this phenomenon could be the consequence of reduced TSH bioactivity, possibly consequent to abnormal syalilation. [307] The transient nature of these changes is reflected in normalization of the pituitary-thyroid axis after complete recovery. It is currently not clearly established whether the above abnormalities on hypothalamic-pituitary-thyroid axis during critical illness reflect an adaptation of the organism to illness or instead a potentially harmful condition leading to hypothyroidism at tissue level. [308, 309]
Certain neuropsychiatric disorders may also be associated with alterations in TSH secretion. In patients with anorexia nervosa or depressive illness, serum TSH may be reduced and/or TRH-induced TSH release blunted. [310] Such patients often have decreases in the evening enhancement of TSH secretion. [311] The etiology of these changes is not known though it has been speculated that they are a consequence of abnormal TRH secretion. [312, 313] The latter is supported by observations that TRH concentrations in cerebrospinal fluid of some depressed patients are elevated. [314, 315] There may be a parallel in such patients between increases in TRH and those of ACTH secretion. [316] In agreement with this are the increased serum T4 and TSH levels sometimes found at the time of admission to psychiatric units. [313, 317]
The precise mechanism(s) underlying the suppression of the hypothalamic-pituitary-thyroid axis are only partially known. Evidence for a direct involvement of TRH-producing neurones in humans has been recently provided by the demonstration of low levels of TRH mRNA in PVN of patients died for nonthyroidal disease. [318] Alteration in neuroendocrine pathways including opioidergic, dopaminergic and somatostatinergic activity have been advocated, but in acutely ill patients the major role appears to be played by glucocorticoids. [319] (See below for a more detailed discussion). Activation of pro-inflammatory cytokine pathways is an other mechanism potentially involved in the suppression of TSH secretion in nonthyroidal illness. As discussed before, IL-2β, TNFα and IL-6 exert in vivo and in vitro a marked inhibitory activity on TRH-TSH synthesis/secretion. High levels of pro-inflammatory cytokines (particularly IL-6 and TNF-α) have been described in sera of patients with nonthyroidal illnesses. [235, 320-323] Serum cytokine concentration is directly correlated with the severity of the underlying disease and to the extent of TSH and thyroid hormone abnormalities observed in these patients. Furthermore, cytokines also affect thyroid hormone secretion, transport and metabolism providing all the characteristics to be considered important mediators of thyroid hormone abnormalities observed in nonthyroidal illness. [324-326]
Dopamine and dopamine agonists inhibit TSH release by mechanisms discussed earlier. Dopamine infusion can overcome the effects of thyroid hormone deficiency in the severely ill patient, suppressing the normally elevated TSH of the patient with primary hypothyroidism nearly into the normal range. [208, 327] Dopamine causes a reduction of the amplitude of TSH pulsatile release, but not in its frequency. [302] However, chronic administration of dopamine agonists, for example in the treatment of prolactinoma, does not lead to central hypothyroidism despite the fact that there is marked decrease in the size of the pituitary tumor and inhibition of prolactin secretion.
The acute administration of pharmacological quantities of glucocorticoids will transiently suppress TSH. [187, 328, 329] The mechanisms responsible for this effect may act both at hypothalamic and pituitary level, as discussed before. Recently, direct evidence of suppressed TRH synthesis was provided by an autoptic study showing reduced hypothalamic TRH mRNA expression in subjects treated with corticosteroids before death. [330] TSH secretion recovers and T4 production rates are generally not impaired. In Cushing's syndrome, TSH may be normal or suppressed and, in general, there is a decrease in serum T3 concentrations relative to those of T4. [328] High levels of glucocorticoid inhibit basal TSH secretion slightly and may influence the circadian variation in serum TSH.222 Perhaps as a reflection of this, a modest serum TSH elevation may be present in patients with Addison's disease. [331, 332] TSH normalizes with glucocorticoid therapy alone if primary hypothyroidism is not also present. Similarly to patients treated with long-acting somatostatin analogs, patients receiving long-term glucocorticoid therapy do not have sustained reduction of serum TSH nor does hypothyroidism develop, because of the predominant effect of reduced thyroid hormone secretion in stimulating TSH secretion. [333]
Aside from the well described effects of estrogen on the concentration of thyroxine binding globulin (TBG), estrogen and testosterone have only minor influences on thyroid economy (see Chapters 5 and 14). In contrast with the mild inhibitory activity on α and TSH β gene subunity expression described in rats, [191] in humans TSH release after TRH is enhanced by estradiol treatment perhaps because estrogens increase TRH receptor number. [334, 335] Treatment with the testosterone analog, fluoxymesterone, causes a significant decrease in the TSH response to TRH in hypogonadal men, [336] possibly due to an increase in T4 to T3 conversion by androgen. [337] This and the small estrogen effect may account for the lower TSH response to TRH in men than in women although there is no difference in basal TSH levels between the sexes. This is one of the few instances where there is not a close correlation between basal TSH levels and the response to TRH (see below).
The possibility that hypothyroidism could be induced by GH replacement in GH-deficient children was raised in early studies. [338, 339] However, these patients received human pituitary GH which in some cases was contaminated with TSH, perhaps inducing TSH antibodies. More recent studies employing recombinant GH have shown no significant changes in TSH concentrations during therapy of adults with GH deficiency. [340] Growth hormone does cause an increase in serum free T3, a decrease in free T4, and an increase in the T3 to T4 ratio in both T4-treated and T4 untreated patients. This suggests that the GH-induced increase in IGF-1 stimulates T4 to T3 conversion. In keeping with this concept, IGF-1 administration in healthy subjects is followed by a fall in serum TSH concentration. [341]
At difference with rat, there is scanty evidence of an adrenergic control of TSH secretion in humans. Acute infusions of α or β adrenergic blocking agents or agonists for short periods of time do not affect basal TSH, [342, 343] although a small stimulatory activity for endogenous adrenergic pathways is suggested by other studies. [344, 345] Furthermore, there is no effect of chronic propranolol administration on TSH secretion even though there may be modest inhibition of peripheral T4 to T3 conversion if amounts in excess of 160 mg/day are given. [346] Evidence of a tonic inhibition of TSH secretion mediated by endogenous catecholamines has been obtained in women during the early follicular phase of the menstrual cycle. [347]
In the last decade, the application of TSH measurements to the evaluation of patients with thyroid disease has undergone a revolutionary change. This is due to the widespread application of the immunometric TSH assay. This assay uses monoclonal antibodies which bind one epitope of TSH and do not interfere with the binding of a second monoclonal or polyclonal antibody to a second epitope. The principle of the test is that TSH serves as the link between an immobilized antibody binding TSH at one epitope and a labelled (radioactive, chemiluminescent or other tag) monoclonal directed against a second portion of the molecule. This approach has improved both sensitivity and specificity by several orders of magnitude. Technical modifications have led to successive "generations" of TSH assays with progressively greater sensitivities.218,316 The first generation TSH assay is considered to be the standard radioimmunoassay which generally has minimal detection limits of 1-2 mU/L. The "second" generation (first generation immunometric) assay improved the sensitivity to 0.1-0.2 mU/L and the "third" reduced the sensitivity to approximately 0.005 mU/L. Third generation assays are currently being introduced into many clinical laboratories. From a technical point-of-view, the American Thyroid Association recommendations are that third generation assays should be able to quantitative TSH in the 0.010 to 0.020 mU/L range on an interassay basis with a coefficient of variation of 20% or less. [348] The most recent development is an assay with a minimal usable sensitivity of 0.0004 mU/L. Such assays are currently available only in specialized laboratories. It would appear that the third generation assays will provide sufficient sensitivity for even the most rigorous clinical applications. As assay sensitivity has improved, the normal range has not changed, remaining between approximately 0.5 and 5.0 mU/L in most laboratories. However, the TSH concentrations in the sera of patients with severe thyrotoxicosis secondary to Graves' disease have been lower with each successive improvement in the TSH assays: using a fourth generation assay, the serum TSH is <0.004 mU/L in patients with severe hyperthyroidism. [249] While the potential for such high sensitivity is inherent to the technology, the clinician should always ascertain that the performance in his/her clinical laboratory meets the appropriate sensitivity criteria before assuming that an assay stipulated to be "second" or "third" generation is achieving that sensitivity on site. [97, 349]
The primary consequence of the availability of the sensitive TSH assays is to allow the substitution of a basal TSH measurement for the TRH test in patients suspected of thyrotoxicosis. [247-249, 350, 351] Nonetheless, it is appropriate to review the results of TRH tests from the point-of-view of understanding thyroid pathophysiology, particularly in patients with hyperthyroidism or autonomous thyroid function. In healthy individuals bolus i.v. injection of TRH is promptly followed by a rise of serum TSH concentration peaking after 20 to 30 minutes. The magnitude of the TSH peak is proportional to the logarithm of TRH doses between 6.25 up to ³400 μg, is significantly higher in women than in men and declines with age. [352, 353] The individual TSH response to TRH is very variable and declines after repeated TRH administrations at short time intervals. [353] In the presence of normal TSH bioactivity and adequate thyroid functional reserve, serum T3 and T4 also increase 120 – 180 minutes after TRH injection. [353] There is a tight correlation between the basal TSH and the magnitude of the TRH-induced peak TSH. (Fig. 4-12) Using a normal basal TSH range of 0.5 to 5 mU/L, the TRH response 15 to 20 minutes after 500 mg TRH (intravenously) ranges between 2 and 30 mU/L. The lower responses are found in patients with lower (but still normal) basal TSH levels. [249] These results are quite consistent with older studies using radioimmunoassays. [354] When the TSH response to TRH of all patients (hypo-, hyper- and euthyroid) is analyzed in terms of a "fold" response, the highest response (approximately 20 fold) occurs at a basal TSH of 0.5 mU/L and falls to less than 5 at either markedly subnormal or markedly elevated basal serum TSH concentrations (Fig. 4-12). [249] Thus a low response can have two explanations. The low response in patients with hyperthyroidism and a reduced basal TSH is due to refractoriness to TRH or depletion of pituitary TSH as a consequence of chronic thyroid hormone excess. In patients with primary hypothyroidism, the low fold-response reflects only the lack of sufficient pituitary TSH to achieve the necessary increment over the elevated basal TSH.
Figure 12. Relationship between basal and absolute (TRH stimulated-basal TSH) TRH-stimulated TSH response in 1061 ambulatory patients with an intact hypothalamic-pituitary (H-P) axis compared with that in untreated and T4-treated patients with central hypothyroidism. (From C.A. Spencer et al, [249] with permission)
![Relationship between basal and absolute (TRH stimulated-basal TSH) TRH-stimulated TSH response in 1061 ambulatory patients with an intact hypothalamic-pituitary (H-P) axis compared with that in untreated and T4-treated patients with central hypothyroidism. (From C.A. Spencer et al, [249] with permission)](./figures4/figure12.jpg)
Although, as stated before, the clinical relevance of the TRH test is presently limited, there are still some conditions in which the test may still be useful. These include subclinical primary hypothyroidism, central hypothyroidism, the syndromes of inappropriate TSH secretion and non-thyroidal illnesses.
In patients with normal serum thyroid hormone concentrations and borderline TSH, an exaggerated TSH response to TRH not followed by an adequate increase in serum thyroid hormone levels may confirm the presence of subtle primary hypothyroidism. [353]
An abnormal relationship between the basal TSH and the TRH-response is found in patients with central hypothyroidism. Here the fold TSH response to TRH is lower than normal. [20, 249, 355] Again, however, TRH testing does not add substantially to the evaluation of such patients in that the diagnosis of central hypothyroidism is established by finding a normal or slightly elevated basal TSH in the presence of a significantly reduced free T4 concentration. While statistically lower and sometimes delayed increments in TSH release after TRH infusion are found in patients with pituitary as opposed to hypothalamic hypothyroidism, the overlap in the TSH increments found in patients with these two conditions is sufficiently large[20, 21, 355, 356] that other diagnostic technologies, such as MRI, must be used to provide definitive localization of the lesion in patients with central hypothyroidism. It should be recalled that the TRH test may be useful in the diagnosis and follow-up of several pituitary disorders, but the discussion of this point is beyond the purpose of this chapter.
TRH test still provides fundamental information in the differential diagnosis of hyperthyroidism due to TSH-secreting adenomas from syndromes with non-neoplastic TSH hypersecretion due to pituitary selective or generalized thyroid hormone resistance. In all the above conditions increased or “inappropriately normal” serum TSH concentrations are observed the presence of elevated circulating thyroid hormone levels. However in most (>92%) of TSH-secreting adenomas serum TSH does not increase after TRH, while TRH responsiveness is observed in >95% of patients with nontumoral inappropriate TSH secretion. [188, 246, 353]
Perhaps of most interest pathophysiologically is the response to TRH in patients with non-thyroidal illness and either normal or low free T4 indices (Fig. 4-12). Results from these patients fit within the normal distribution in terms of the relationship between basal TSH (whether suppressed or elevated) and the fold-response to TRH. Thus the information provided by a TRH infusion test adds little to that obtained from an accurate basal TSH measurement. [357] With respect to the evaluation of sick patients, while basal TSH values are on average higher than in patients with thyrotoxicosis, there is still some overlap between these groups. [249, 299, 358, 359] This indicates that even with second or third generation TSH assays, it may not be possible to establish that thyrotoxicosis is present based on a serum TSH measurement in a population which includes severely ill patients.