Evaluation of the Hypothalamic-Pituitary-Thyroid Axis

The development of an RIA for the routine measurement of TSH in serum and the availability of synthetic TRH^390,391 have placed increased reliance on tests assessing the hypothalamic-pituitary control of thyroid function. These tests allow the diagnosis of mild and subclinical forms of thyroid dysfunction, and provide a means to differentiate between primary, pituitary (secondary) or hypothalamic (tertiary), thyroid gland failure.

Thyrotropin (TSH)

The routine measurement of TSH in clinical practice used initially RIA techniques. These first generation assays had a sensitivity level of 1 mU/L which did not allow the separation of normal from reduced values. A major problem with early TSH RIAs was cross-reactivity with gonadotropins (LH, FSH, and hCG) sharing with TSH a common a-subunit.³⁹⁹ Nevertheless, even older RIA methods for measurement of pituitary TSH correlated well with values obtained using bioassay techniques.⁴⁰¹ Another uncommon source of error is the presence in the serum sample of heterophilic antibodies induced by vaccination with materials contaminated with animal serum,⁴⁰² or endogenous TSH antibodies.⁴⁰³ RIA techniques for measurement of TSH in dry blood spots on filter paper are used for the screening of neonatal hypothyroidism.³³

Newer techniques have been developed using multiple antibodies to produce a "sandwich" type assay in which one antibody (usually directed against the a subunit) serves to anchor the TSH molecule and an other (usually monoclonal antibodies directed against the ß subunit) is either radioiodinated (Immunoradiometric assay, IRMA) or is conjugated with an enzyme (Immunoenzymometric, IEMA) or a chemiluminescent compound (Chemiluminescent assay, ICMA).^112,404 In these assays, the signal should be directly related to the amount of the ligand present rather than being inversely related as in RIAs measuring the bound tracer.⁴⁰⁵ This results in decreased background "noise" and a greater sensitivity, decreased interference from related compounds as well as an expanded useful range.^112,404,406 Initial improvements of the TSH assay resulted in assays with sensitivity limit of 0.1 mU/L, a normal range of approximately 0.5 - 4.5 mU/L and the ability to distinguish between low and normal TSH values. Recently, commercial assays have been developed with even higher sensitivity limit of 0.005 - 0.01 mU/L and a similar normal range but an expanded range between the lower limit of normal and the lower limit of sensitivity.^407,408

The nomenclature for differentiating these various assays has not been standardized with manufacturers applying various combinations of "high(ly)", "ultra" and "sensitive". It has been recommended that the sensitivity limit be used in defining the assays with the early radioimmunoassays detecting values ?1 mU/L designated "first generation assays", those with a lower sensitivity limit of 0.1 mU/L designated as "second generation assays" and those with a lower sensitivity limit of ? 0.01 mU/L designated as "third generation assays".¹¹² The determination of the appropriate sensitivity level has also been controversial. Some define it based on the level with a coefficient of variation less than 20% and others as the lowest level which can be reliably differentiated from the zero TSH standard.112,406 At a minimum, for a TSH assay to be considered "sensitive", the overlap of TSH values in sera from clinically hyperthyroid and euthyroid individuals should be less than 5% and preferably less than 1%.¹¹²

In a number of these "third generation" assays, TSH detected in clinically toxic patients and elevated values found in euthyroid subjects were not confirmed when the samples were measured in other assays. In some cases, this has been attributed to the presence of antibodies directed against the animal immunoglobulins used in the assay.^409-411 These act to bind the anchoring and detecting antibodies and lead to an over-estimation of TSH. In some cases, this effect may be blocked by the addition of an excess of non-specific immunoglobulin of the same species.⁴¹¹

TSH appears abruptly in the pituitary and serum of the fetus at midgestation, and can also be detected in amniotic fluid.^51,412,413 The mean TSH level is higher in cord than in maternal blood. A substantial increase, to levels several fold above the upper range in adults, is observed during the first half-hour of life.⁴¹³ Levels decline to near the normal adult range by the third day of life. Minimal changes reported to occur during adult life and in early adolescence⁴¹⁴ have no significant effect on the overall range of normal. In the absence of pregnancy, no significant sex differences have been observed. Although early studies failed to show diurnal TSH variation,⁴¹⁵ significantly higher values have been recorded during the late evening and early night which are partially inhibited by sleep.⁴¹⁶ This diurnal rhythm of TSH is superimposed upon continuous high-frequency, low-amplitude variations. The nocturnal TSH surge persists in patients with mild primary hypothyroidism,^417,418 and is abolished in hypothalamic hypothyroidism^417,419 and in some patients during fasting⁴²⁰ and with non-thyroidal illness.^421,422 It is enhanced by oral contraceptives,⁴²³ and is abolished by high levels of glucocorticoids.⁴²⁴ The presence of seasonal variation has not been a uniform finding, but it is unlikely to affect the clinical interpretation of serum values.⁴²⁵ Various types of stressful stimuli have no significant effect on the basal serum TSH level, except for a rise during surgical hypothermia in infants but not in adults.⁴²⁶ Various stimuli, such as administration of insulin, vasopressin, glucagon, bacterial pyrogens, arginine, prostaglandins, and chlorpromazine, which elicit in normal humans a secretory response of some pituitary hormones, have no effect on serum TSH. However, administration of any of a growing list of drugs has been found to alter the basal concentration of serum TSH and/or its response to exogenous TRH (see Table 5-4).

In the presence of a normally functioning hypothalamic-pituitary system, there is an inverse correlation between the serum concentration of FT4 and TSH. Changes in the serum concentration of TT4 as a result of TBG abnormalities, or drugs competing with T4 binding to TBG, have no effect on the level of serum TSH. The pituitary is exquisitely sensitive to both minimal decreases and increases in thyroid hormone concentration, with a logarithmic change in TSH levels in response to changes in T4^{404,408,427,428} (Figure 6-7) Thus, serum TSH levels should be elevated in patients with primary hypothyroidism and low or undetectable in thyrotoxicosis. Indeed, in the absence of hypothalamic pituitary disease, illness or drugs, TSH is an accurate indicator of thyroid hormone status and the adequacy of thyroid hormone replacement.^404,429

Figure 6-7. Correlation of the serum TSH concentration and the free thyroxine index (FT4I) in three individuals given increasing doses of L-T4. Note the logarithmic correlation between TSH and FT4I and the variable individual requirement of free T4 to normalize the TSH level. Normal ranges are included in the heavy lined box and those for subjects on L-T4 replacement in the light liquid box. (From D. Sarne and S. Refetoff, Endocrinology, L.J. DeGroot (ed). 1995, Grune & Straton Inc.)

In patients with primary hypothyroidism of whatever cause, levels may reach 1,000 µU/ml or higher. The magnitude of serum TSH elevation grossly correlates with the severity and in part with the duration of thyroid hormone deficiency.^430,431 TSH concentrations above the upper limit of normal have been observed in the absence of clinical symptoms and signs of hypothyroidism and in the presence of serum T4 and T3 levels well within the normal range.^430,432 This condition is most commonly encountered in patients developing hypothyroidism due to Hashimoto's thyroiditis or with limited ability to synthesize thyroid hormone because of prior thyroid surgery, radioiodide treatment, or severe iodine deficiency.^430,433 There is disagreement on whether such patients have subclinical hypothyroidism or a "compensated state" in which euthyroidism is maintained by chronic stimulation of a reduced amount of functioning thyroid tissue through hypersecretion of TSH. Transient hypothyroidism, may occur in some infants during the early neonatal period.⁴³⁴ There are two circumstances in which the usual reverse relationship between the serum level of TSH and T4 is not maintained in patients with proven primary hypothyroidism. Treatment with replacement doses of T4 may normalize or even produce serum levels of thyroid hormone above the normal range before the high TSH levels have reached the normal range.^404,431,435 This is particularly true in patients with severe or long-standing primary hypothyroidism who may require three to six months of hormone replacement before TSH levels are fully suppressed. Conversely, serum TSH concentration may remain low or normal for up to five weeks after withdrawal of thyroid hormone replacement when serum levels of T4 and T3 have already declined to values well below the lower range of normal.^404,436 Causes for discrepancies between TSH and free T4 and T3 levels are listed in Table 6-9.

At this time, it is uncertain as to what TSH level is appropriate for suppressive thyroid hormone therapy. The frequency with which patients have subnormal, but detectable, TSH values depends on both the population studied and the sensitivity of the assay (Figure 6-8, below). Using an assay with a sensitivity limit of 0.1 mU/L, 3 to 4% of hospitalized patients have been noted to have a subnormal TSH.^432,437 When patients with an undetectable TSH in such an assay were re-evaluated in an assay with a sensitivity limit of 0.005 mU/L, 3 of 77 (4%) with thyrotoxicosis and 32 of 37 (86%) with non-thyroidal illness or on drugs were found to have a subnormal but detectable TSH level.⁴⁰⁷ Thus, the more sensitive the assay, the more likely that patients with clinical thyrotoxicosis will have undetectable serum TSH while those with illness will have a subnormal but detectable level. However, with progressively more sensitive assays, the likelihood of a clinically toxic patient to have a detectable TSH will increase, and if patients on suppressive therapy are treated until the TSH is undetectable, the more likely they will have symptoms of thyrotoxicosis.

Figure 6-8. The effect of serum TSH assay sensitivity on the discrimination of euthyroid subject (Euth) from those with thyrotoxicosis (Toxic). (From C. Spencer, Clinical Diagnostics, Eastman Kodak Co., 1992).

A persistent absence of a reverse correlation between serum thyroid hormone and TSH concentration has a very different connotation. A low serum level of thyroid hormone without clear elevation of the serum TSH concentration is suggestive of trophoprivic hypothyroidism, especially when associated with obvious clinical stigmata of hypothyroidism.⁴³³ An inherited defect of the TSH receptor has been shown to produce marked persistent hyperthyrotropinemia in the presence of normal thyroid hormone levels.⁴³⁸ In some cases, a mild elevation of the serum TSH level measured by RIA is probably due to the presence of immunoreactive TSH with reduced biologic activity.³⁹⁷ Distinction between pituitary and hypothalamic hypothyroidism can be made on the basis of the TSH response to the administration of TRH (see below).

In another group of pathologic conditions, serum TSH levels may not be suppressed despite a clear elevation of serum free thyroid hormone levels. Because such a finding is incompatible with a normal thyroregulatory control mechanism of the pituitary, which is preserved in the more common forms of thyrotoxicosis, it has been termed inappropriate secretion of TSH.⁴³⁹ It implicitly suggests a defective feedback regulation of TSH. When associated with the classical clinical and metabolic changes of thyrotoxicosis, it is usually due to TSH-secreting pituitary adenoma or isolated pituitary resistance to the feedback suppression by thyroid hormone.⁴³⁹ The existence of hypothalamic hyperthyroidism can be questioned.⁴⁴⁰ Precise diagnosis requires further studies, including radiologic examination of the pituitary gland and a TRH test. In addition, the presence of high circulating levels of the a-subunit of pituitary glycoprotein hormones (a-SU), giving rise to a disproportionately high a-SU/TSH molar ratio in serum, is characteristic, if not pathognomonic, of TSH-secreting pituitary tumors.^439,441 Normal, and occasionally high serum TSH levels, associated with a clear elevation in serum FT4 and FT3 but no clear clinical evidence of hypothyroidism or symptoms and signs suggestive of both thyroid hormone deficiency and excess are typical of resistance to thyroid hormone (RTH) ⁴⁴² (see Chapter 16).

Although TSH has been implicated in the pathogenesis of simple, nontoxic goiter, unless hypothyroidism supervenes or iodide deficiency is very severe, TSH levels are characteristically normal. Elevated TSH levels may occur in the presence of normal thyroid hormone levels and apparent euthyroidism in nonthyroidal diseases^437,443 (see also Chapter 5) and with primary adrenal failure.⁴⁴⁴ A more common occurrence in severe acute and chronic illnesses is a normal or low serum TSH concentration despite low levels of T3 and even low T4 levels.^407,429,445 TSH values may be transiently elevated during the recovery phase.446 Various hypotheses to explain these anomalous findings have been proposed, but a satisfactory explanation is not at hand.

A specific RIA for the ß subunits of human TSH is also available but has not found clinical application.⁴⁴⁷

Thyrotropin-Releasing Hormone (TRH)

TRH. The hypothalamic tripeptide TRH (protirelin) plays a central role in the regulation of pituitary TSH secretion.^391,419 It is thus not surprising that attempts have been made to measure its concentration in a variety of body fluids, with the purpose of deriving information relevant to the function of the thyroid gland in health and in disease. Several methods have been used for quantitation of TRH,^448-451 but for many reasons, measurement in humans has failed to provide information of diagnostic value. These include, high dilution of TRH by the time it reaches the systemic circulation, rapid enzymatic degradation and ubiquitous tissue distribution.^448,450,451 Mean serum TSH levels of 5 and 6 pg/ml have been reported. It is uncertain whether measurements carried out in urine truly represent TRH.⁴⁴⁹

TRH Test.

The TRH test measures the increase of pituitary TSH in serum in response to the administration of synthetic TRH. The magnitude of the TSH response to TRH is modulated by the thyrotroph response to active thyroid hormone and is thus almost always proportional to the concentration of free thyroid hormone in serum. The response is exquisitely sensitive to minor changes in the level of circulating thyroid hormones, which may not be detected by direct measurement.^427,428 A direct correlation between basal serum TSH values and the maximal response to TRH has been observed even in the absence of thyroid hormone abnormalities, suggesting that there may be a fine modulation of pituitary sensitivity to TRH in the euthyroid state.⁴⁵²

TRH normally stimulates pituitary prolactin secretion and, under certain pathologic conditions, the release of GH and ACTH.³⁹¹ Accordingly, the test has been used for the assessment of a variety of endocrine functions, some unrelated to the thyroid. In clinical practice, the TRH test is used mainly (1) to assess the functional integrity of the pituitary thyrotrophs and thus to aid in differentiating hypothyroidism due to intrinsic pituitary disease from hypothalamic dysfunction and (2) in the diagnosis of mild thyrotoxicosis when results of other tests are equivocal, and (3) in the differential diagnosis of inappropriate TSH secretion, in particular when a TSH-secreting adenoma is suspected.

TRH is effective when given intravenously as a bolus or by infusion,^414,453 intramuscularly,⁴⁵⁴ or orally⁴⁵⁵ in single or repeated doses. Doses as small as 6 µg can elicit a significant TSH response, and there is a linear correlation between the incremental changes in serum TSH concentrations and the logarithm of the administered TRH dose.⁴¹⁴ The standard test uses a single TRH dose of 400 µg/1.73 m2 body surface area, given by rapid intravenous injection. Serum is collected before and at 15 minutes and then at 30 minute intervals over 120-180 minutes although many clinicians chose to obtain a single post-injection sample at 15, 20 or 30 minutes. In normal persons there is a prompt increase in serum TSH, with a peak level at 15-40 minutes, which is, on the average, 16 µU/ml, or fivefold the basal level. The decline is more gradual, with a return of serum TSH to the preinjection level by three to four hours.^414,453 Results can be expressed in terms of the peak level of TSH achieved, the maximal increment above the basal level (?TSH), the peak TSH value expressed as a percentage of the basal value, or the integrated area of the TSH response curve. Determination of TSH before and 30 minutes after the injection of TRH provides information concerning the presence or absence of TSH responsiveness but cannot detect delayed or prolonged responses.

The stimulatory effect of TRH is specific for pituitary TSH, its free a- and ß- subunits,⁴⁴⁷ and prolactin. Under normal circumstances, no significant changes are observed in the serum levels of other pituitary hormones⁴⁵⁶ or potential thyroid stimulators.⁴⁵⁷ Responsiveness is present at birth,⁴⁵⁸ is greater in women than in men, particularly in the follicular phase of the menstrual cycle,⁴⁵⁹ and may be blunted in older men,^414,454,455 but this is not a consistent finding.⁴⁶⁰ On the average, the magnitude of the response is greater at 11 P.M. than at 11 A.M.,⁴⁵² in accordance with the diurnal pattern of the basal TSH level which correlates to its response to TRH. Repetitive administration of TRH to the same subject at daily intervals causes a gradual obtundation of the TSH response,⁴⁵³ presumably due to the increase in thyroid hormone concentration⁴⁶¹ and also in part due to TSH "exhaustion".⁴⁶² However, more than one hour must elapse between the increase in thyroid hormone concentration and TRH administration for inhibition of the TSH response to occur. A number of drugs (see Table 5-4) and nonendocrine diseases (see Chapter 5) may affect to various extents the magnitude of the response.

TRH-induced secretion of TSH is followed by a release of thyroid hormone that can be detected by direct measurement of serum TT4 and TT3 concentrations.¹⁶⁰ Peak levels are normally reached approximately four hours after the administration of TRH and are accompanied by an increase in serum Tg concentration. The incremental rise in serum TT3 is relatively greater, and the peak is, on the average, 50% above the basal level. Measurement of changes in serum thyroid hormone concentration after the administration of TRH has been proposed as an adjunctive test and is useful in the evaluation of the integrity of the thyroid gland or bioactivity of endogenous TSH.⁴⁶³ Increase in RAIU is minimal and occurs only with high doses of TRH given orally.⁴⁵⁵

Side effects from the intravenous administration of TRH, in decreasing order of frequency, include nausea, flushing or a sensation of warmth, desire to micturate, peculiar taste, light-headedness or headache, dry mouth, urge to defecate, and chest tightness. They are usually mild, begin within a minute after the injection of TRH, and last for a few seconds to several minutes. A transient rise in blood pressure has been observed on occasion, but there are no other changes in vital signs, urine analysis, blood count, or routine blood chemistry tests.^456,464 The occurrence of circulatory collapse is exceedingly rare.⁴⁶⁵

The test provides a means to distinguish between secondary (pituitary) and tertiary (hypothalamic) hypothyroidism (Fig. 6-9). Although the diagnosis of primary hypothyroidism can be easily confirmed by the presence of elevated basal serum TSH levels, secondary and tertiary hypothyroidism are typically associated with TSH levels that are low or normal. On occasion the serum TSH concentration may be slightly elevated due to the secretion of biologically less potent molecules,³⁹⁷ but it remains inappropriately low for the degree of thyroid hormone deficiency. Differentiation between secondary and tertiary hypothyroidism cannot be made with certainty without the TRH test. A TSH response is suggestive of a hypothalamic disorder, and a failure to respond is compatible with intrinsic pituitary dysfunction.⁴⁶⁶ Furthermore, the typical TSH response curve in hypothalamic hypothyroidism shows a delayed peak with a prolonged elevation of serum TSH before return to the basal value (Figure 6-9). The lack of a TSH response in association with normal prolactin stimulation may be due to isolated pituitary TSH deficiency.⁴⁶⁷ Caution should be exercised in the interpretation of test results after withdrawal of thyroid hormone replacement or after treatment of thyrotoxicosis when, despite a low serum thyroid hormone concentration, TSH may remain low and not respond to TRH for several weeks.^{404,433,436,468}

Figure 6-9. Typical serum TSH responses to the administration of a single intravenous bolus of TRH at time 0 in various conditions. The normal response is represented by the shaded area. Data used for this figure are the average of several studies. (From S. Refetoff, Endocrinology, L.J. DeGroot (ed). 1979, Grune & Straton Inc.)

In the most common forms of thyrotoxicosis, the mechanism of feedback regulation of TSH secretion is intact but is appropriately suppressed by the excessive amounts of thyroid hormone. Thus, both the basal TSH level and its response to TRH are suppressed unless thyrotoxicosis is TSH induced.^404,407,417 With the development of more sensitive TSH assays, the TRH test is generally not needed in the evaluation of a thyrotoxic patient with an undetectable TSH.407 Differential diagnosis of conditions leading to inappropriate secretion of TSH may be aided by the TRH test result. Elevated basal TSH values that do not respond by a further increase to TRH are typical of TSH-secreting pituitary adenomas.^439,441 Patients with inappropriate secretion of TSH due to resistance to thyroid hormone have a normal or exaggerated TSH response to TRH that, in most instances, is suppressed with supraphysiologic doses of thyroid hormone.⁴⁴²

Because of the high sensitivity of the pituitary gland to the feedback regulation by thyroid hormone, small changes in the latter profoundly affects the response of TSH to TRH. Thus, patients with non-TSH-induced thyrotoxicosis of the mildest degree have a reduced TSH response to TRH whereas those with primary hypothyroidism exhibit an accentuated response that is prolonged (Figure 6-9, see above). These changes may occur in the absence of clinical or other laboratory evidence of thyroid dysfunction.

The TSH response to TRH, is subnormal or absent in one-third of apparently euthyroid patients with autoimmune thyroid disease, and even members of their family, may not respond to TRH.^469,470 Most, but not all patients with reduced TSH response to TRH, will also show thyroid activity that is nonsuppressible by thyroid hormone. A common dissociation between these two tests is typified by a normal TRH response in a nonsuppressible patient. This finding is not surprising since patients with nonsuppressible thyroid glands often have limited capacity to synthesize and secrete thyroid hormone, due to prior therapy or partial destruction of their glands by the disease process. Clinically, euthyroid patients, who do not respond to TRH, admittedly have a slight excess of thyroid hormone. It is less easy to reconcile the rare occurrence of TRH unresponsiveness in a patient who is suppressible by exogenous thyroid hormone. It should be remembered, however, that a suppressed pituitary may take a variable amount of time to recover, a phenomenon that may be the basis of such discrepancies.^404,436,468 Despite discrepancies between the results of the TRH and T3 suppression tests,^469,470 the use of the former is much preferred particularly in elderly patients in whom administration of T3 can produce untoward effects.

Thyroid Suppression Test

The maintenance of thyroid gland activity that is independent of TSH can be demonstrated by the thyroid suppression test. Under normal conditions, administration of thyroid hormone in quantities sufficient to satisfy the body requirement suppresses endogenous TSH resulting in reduction of thyroid hormone synthesis and secretion. Since thyrotoxicosis due to excessive secretion of hormone by the thyroid gland implies that the feedback control mechanism is not operative or has been perturbed, it is easy to understand why under such circumstances the supply of exogenous hormone would also be ineffective in suppressing thyroid gland activity. The test is of particular value in patients who are euthyroid or only mildly thyrotoxic but suspected of having abnormal thyroid gland stimulation or autonomy.

Usually the test is carried out with 100 µg of L-T3 (liothyronine) given daily in two divided doses over a period of 7-10 days. 24 hour RAIU is obtained before and during the last two days of T3 administration.476 Normal persons show a suppression of the RAIU by at least 50% compared to the pre-L-T3 treatment value. No change or lesser reduction is not only typical of Graves' disease but also other form of endogenous thyrotoxicosis, including toxic adenoma, functioning carcinoma, and thyrotoxicosis due to trophoblastic diseases. The presence of nonsuppressibility indicates thyroid gland activity independent of TSH but not necessarily thyrotoxicosis. Euthyroid patients with autonomous thyroid function have a normal TSH response to TRH before the administration of L-T3. However, inhibition of TSH secretion by the exogenous T3 does not suppress the autonomous activity of the thyroid gland. This is the most commonly encountered discrepancy between the results of the two related tests. When the T3 suppression test is used in conjunction with the scintiscan, localized areas of autonomous function can be identified. The test can be carried out without the administration of radioisotopes by measuring serum T4 before and two weeks following the ingestion of L-T3. Although total suppression of T4 secretion never occurs, even after prolonged treatment with L-T3, a reduction by at least 50% is normal.⁴⁷⁷

Variants of the test have been proposed to reduce the potential risks of L-T3 administration in elderly patients and in those with angina pectoris or congestive heart failure. With the availability of sensitive TSH determinations and the TRH test, which are less dangerous, thyroid suppression tests are no longer indicated.