4.THYROID TESTING STRATEGIES IN SPECIAL SITUATIONS

4.1 Thyrotoxicosis

4.1.1 Initial diagnosis

The initial diagnosis of thyrotoxicosis is securely established when excess of free T4 and free T3 is associated with an undetectable serum TSH concentration in an assay of appropriate sensitivity. However, thyrotoxicosis can be present without any one of these three criteria. Subclinical thyrotoxicosis is defined by suppression of TSH, without excess of either T3 or T4 (139). T3 thyrotoxicosis, in which serum T4 remains normal, is especially prevalent in iodine deficient regions (160) and can be a premonitory stage of typical thyrotoxicosis (161). When thyrotoxicosis coexists with another severe illness, serum T3 may be transiently normal or even subnormal (162). Detectable or increased serum TSH concentrations in the face of thyrotoxicosis can be due to laboratory artefacts (147), or less frequently to autonomous over-secretion of TSH (135).

In thyrotoxicosis, the increases in free T4 and free T3 are usually more marked than the increases in total hormone concentration. Progressive increases in serum total T4, approach and eventually exceed the limited binding capacity of TBG (114), leading to disproportionate increases in the free serum concentrations of T4 (163) and T3 (164). Serum T3, T4 and TSH values have limited value in distinguishing between the various common causes of thyrotoxicosis. The increase in serum T3 may be less marked in thyrotoxicosis due to thyroiditis (142), or amiodarone (106) and after iodine exposure (141). In general, the pathogenesis of thyrotoxicosis is best established clinically, together with imaging and antibody studies (139).

4.1.2 Treated thyrotoxicosis

In the early drug treatment of thyrotoxicosis, measurements of serum free T4 and T3 are required for dose adjustment because suppression of TSH may persist for months after correction of longstanding thyrotoxicosis ( 99). Hence, failure to decrease thionamide dosage while TSH suppression persists, can result in serious over-treatment during the early phase of therapy. During thionamide therapy, thyrotoxicosis may persist due solely to T3 excess (140); assessment of therapy based on serum T4 alone can therefore result in under-treatment. A daily dose of 15 mg methimazole can result in hypothyroidism in about 10 % of previously thyrotoxic subjects within 4 weeks (165); hence a reassessment of both serum free T3 and free T4 is timely after about 3 weeks to allow appropriate dose-adjustment. In contrast to these discrepancies during early treatment, serum TSH generally gives a reliable index of therapy during the longterm drug treatment of thyrotoxicosis.

Davies et al (166) assessed the prognostic significance of various serum TSH levels in a large cohort of patients treated with radioiodine 2-35 years previously, who were receiving no other treatment. After a further two years follow-up, 83% of those with normal TSH had not changed their diagnostic category, although there was a trend for TSH to increase. An increased TSH was associated with a 14.5 % incidence of hypothyroidism after one year. In contrast, spontaneous normalization of subnormal or undetectable TSH values during the follow-up period was more common than recurrence of overt thyrotoxicosis (166). Hence, while an increased TSH value might be a pointer towards T4 treatment, there appears to be no basis for further radioiodine therapy solely because TSH remains suppressed.

4.2 Hypothyroidism

4.2.1 Initial diagnosis

The initial diagnosis of overt primary hypothyroidism is established by increase in serum TSH with subnormal free T4; serum free T3 may remain normal except in severe cases. Subclinical hypothyroidism or mild thyroid failure is defined by persistent elevation of serum TSH, while free T4 remains normal. The precise upper limit of the reference range for TSH is difficult to define, because of the logarithmic distribution of this parameter (35, 36). The upper "tail" of normal TSH in the range 2-4 mU/l is thin and subjects with TSH values in this range may have evidence of autoimmune thyroid disease, as demonstrated by positive TPOAb (38).

Serum TSH is the key to the distinction between primary and secondary hypothyroidism. A low serum free T4 in the absence of TSH elevation should always raise the possibility of a pituitary or hypothalamic abnormality, although this combination of findings is also frequently seen during critical illness (167) and in a number of other situations (table 7). It should be noted also that immunoreactive serum TSH is often detectable in secondary or central thyroid deficiency (112), a phenomenon that appears to result from dissociation between biological and immunological TSH activity (134). Lack of sensitivity in detecting secondary hypothyroidism is a major deficiency of the "TSH first" strategy of thyroid function testing (see above).

4.2.2 Treated hypothyroidism

Some have suggested that hormone measurements add little to a clinical assessment made by experts in evaluating patients receiving T4 therapy (168), but there is justification for periodic serum TSH assessment to avoid over-replacement that may be associated with adverse effects on the cardiovascular system or on bones. A serum TSH value in the low-normal range between 0.5 and 1.5 mU/l, close to the geometric mean, is probably the best single indicator of appropriate dosage. In a study of ambulatory patients attending a thyroid clinic, hypothyroid patients taking T4 replacement seldom needed a serum free T4 measurement if the serum TSH was greater than 0.05 mU/L, although at lower TSH values, the magnitude of T4 excess did influence management (169). Numerous studies show that patients taking exogenous thyroxine show higher levels of serum total and free T4 for equivalent levels of serum TSH and T3 when compared with untreated euthyroid control subjects (169, 170). Lack of direct secretion of T3 from the thyroid may be an explanation for this difference. It should be noted that in some situations (eg patients with ischemic heart disease and hypothyroidism, or in the extreme elderly), the appropriate dose of T4 should be based on clinical judgment rather than solely on laboratory findings.

The assessment of optimal dosage in patients with secondary or central hypothyroidism remains a challenge (170a), because the serum TSH fails to show the normal inverse relationship with thyromimetic effect. Since there is no readily available alternative parameter of thyroid hormone action, it is generally appropriate to assess replacement clinically and on the basis of both serum T4 and T3. Measured levels of free and total T4 are influenced by the interval between tablet ingestion and blood sampling. In athyreotic subjects who took 0.15-0.2 mg T4 orally, the serum free and total T4 concentrations were increased by about 20% one to four hours later, with return to baseline about nine hours after T4 ingestion; serum TSH and T3 levels showed no time-dependent variation (171).

4.2.3 Suppressive treatment with T4

During suppressive T4 therapy in the follow-up of differentiated thyroid cancer, the target TSH is generally in the range 0.05-0.2 mU/l, i.e. lower than in standard replacement (143). The benefit of higher T4 dosage with the aim of suppressing TSH to undetectable levels by a highly sensitive assay (e.g. <0.03 mU/l) remains unproven. When the aim of T4 suppressive therapy is regression of benign thyroid tissue, it may be adequate to give sufficient T4 to reduce serum TSH to 0.1 to 0.3 mU/L (143).

4.2.4 Treatment with T3

When T3 is used for replacement, the serum concentration of T3 is not useful in assessing the effectiveness of treatment. Owing to its short plasma half-life, the serum concentration is highly dependent on the interval between dosage and sampling (172). There is also doubt as to whether TSH serves as an accurate index of thyroid hormone action during long-term T3 therapy, with the suggestion that doses of T3 required to normalise TSH could produce tissue hyperthyroidism (173). The difficulty of monitoring T3 replacement by currently available techniques is one of the arguments against its routine use (174).

4.3 Assessment of thyroid function during non-thyroidal illness (NTI)

There are several distinct questions that need to be considered when assessing thyroid function during critical illness. First, there is the possibility that an underlying thyroid abnormality might be missed, and second, prolonged severe illness per se may be associated with an abnormality of thyroid hormone secretion or action that might benefit from treatment (175). Third, some of the observed abnormalities may be methodological artefacts.

It is extremely difficult to rule out thyroid dysfunction by clinical assessment in patients who are critically ill and current laboratory tests often do little to resolve the problem. Clinicians should be aware of the limitations of current diagnostic methodology during critical illness (84 ). During severe illness, one or more of the assumptions that underpin the diagnostic use of the TSH-T4 relationship may not be justified. For example, acute fluctuations from the steady-state can lead to an anomalous T4-TSH relationship because of the marked difference in their respective plasma half lives.

In general, the same sample assayed for TSH by various methods will give similar results during critical illness, while various estimates of free T4 may give widely divergent results (176, see below). When TSH and T4 changes are considered together, the abnormal results rarely correspond to standard criteria for diagnosis of primary hypothyroidism or thyrotoxicosis. In contrast, persistent hypothyroxinemia without the corresponding anticipated rise in serum TSH is a common finding that suggests secondary or central hypothyroidism (167, 176). These changes appear to be part of a wider neuroendocrine response that also involves the pituitary-adrenal axis, the pituitary-gonadal axis and the IGF binding proteins (177). Further study of these responses may eventually lead to therapy that could extend beyond thyroid hormone replacement, for example to substitution of hypothalamic releasing hormones (178).

4.3.1 Effects of medications

Estimation of thyroid function during critical illness can be influenced by multiple medications (table 8), in particular dopaminergics and glucocorticoids, which inhibit TSH secretion, and a wide range of inhibitors of T4 and T3 binding to TBG.

4.3.2 Methodological discrepancies in non-thyroidal illness

4.3.2.1 The heparin artefact and free T4

The effect of heparin to increase serum free T4 is an important in vitro phenomenon that can lead to spuriously high estimates of circulating free T4 (117). In the presence of a normal serum albumin concentration, non-esterified fatty acid (NEFA) concentrations >3 mmol/l will increase free T4 by displacement from TBG (138), but these concentrations are uncommon in vivo. However, in samples from heparin-treated patients, serum NEFA may increase to these levels during in vitro sample storage or incubation as a result of heparin-induced lipase activity (117) (figure 6). This effect is accentuated by incubation of serum at 37 C and by increased serum triglyceride or low serum albumin concentrations. Under these conditions doses of heparin as low as 10 units may result in NEFA-induced increases in the apparent concentration of serum free T4 (118). The assay result is analytically correct, but does not reflect the in vivo concentration of free T4. Low molecular weight heparin preparations have a similar effect (179).

Figure 6. Heparin-induced release of lipase in vivo can lead to in vitro generation of non-esterified fatty acids during sample incubation or storage. An increase in serum NEFA concentrations to > 3 mmol/l is sufficient to displace T4 from TBG, but such values are uncommon in vivo. This artefact is accentuated by high triglyceride or by low albumin concentrations.

4.3.2.2 Competitors for plasma protein binding

The accuracy of virtually all methods of free T4 estimation is compromised by medications that displace T4 and T3 from TBG. Current free T4 methods underestimate these effects with because of dilution-related artefacts. Binding competitors are usually less protein-bound than T4 itself so that progressive sample dilution leads to a fall in the free concentration of competitor before the free T4 concentration alters (116, 138). (For a hormone such as T4, with a free fraction in serum of about 1:4000, progressive dissociation will sustain the free T4 concentration up to at least 1:100 dilution. In contrast, 1:10 dilution of serum will result in a marked decrease in the free concentration of a drug that is 98% bound, i.e. has a free fraction in serum of 1:50). Because displacement depends on the relative free concentrations of primary ligand and competitor, the underestimate of free T4 will be greatest in assays with the highest sample dilution. This important dilution-dependent difference between various free T4 methods was shown by the relative ability of three commercial free T4 assays to detect the T4-displacing effect of therapeutic concentrations of furosemide (180) (figure 7 ).
Similarly, therapeutic concentrations of phenytoin and carbamazepine increased the free concentration of T4 by 40-50% using ultrafiltration of serum that had not been diluted, while the free hormone estimate was spuriously low using a commercial single-step free T4 assay after 1:5 serum dilution (128).

Figure 7. Influence of increasing serum concentrations of added furosemide on estimates of serum free T4 using three commercial free T4 methods that involve varying degrees of sample dilution. The effect of the competitor is progressively obscured with increasing sample dilution. (Redrawn from 180).

It is possible that methodologic artefacts have influenced previous descriptions of free T4 changes during critical illness. On the one hand, an apparent increase in free T4 may arise from heparin-induced in vitro generation of free fatty acids during sample incubation (117). On the other, estimates of free T4 may be spuriously low in assays that use diluted serum (128, 180).

4.3.2.3 Divergent estimates of free T4

That estimates of free T4 by different methods may show opposite discrepancies in identical samples, was shown by Sapin et al (176) in a prospective study of bone marrow transplant recipients. Twenty previously euthyroid subjects were studied on the seventh day after bone marrow transplantation using six commercial free T4 kits, during multiple drug therapy, including heparin and glucocorticoids (figure 8). Free T4 methods that involved sample incubation at 37 C showed supranormal free T4 values in 20-40% of these subjects (see heparin effect above), while analog tracer methods that are influenced by tracer binding to albumin (127), gave subnormal estimates of free T4 in 20-30%. By contrast, total T4 was normal in 19 of these 20 subjects. Serum TSH was <0.1 mU/l in half the subjects, independent of the method that was used. Thus, there was the possibility that an erroneous diagnosis of either thyrotoxicosis or secondary or central hypothyroidism could be considered, solely as a result of variations in free T4 methodology.

Figure 8. Free T4 estimated by six different kit methods in 20 previously euthyroid patients on the seventh day after bone marrow transplantation. There was a high proportion of abnormal values, either increased or decreased, depending on the type of free T4 method used (see text). Therapy included heparin and glucocorticoids at the time of sampling. The mean for each method has been normalized to 100%, with the limits of the range shown by the box. Serum total T4 remained normal in 19 of the 20 study subjects, while serum TSH was subnormal in 11, independent of assay method. (Redrawn from 176)

4.3.2 Serum TSH in non-thyroidal illness

Serum TSH assessment in severe NTI depends on the sensitivity of the particular method. The "third generation" assays with a functional sensitivity below 0.01 mU/L are generally sufficiently sensitive to distinguish the very low values in most thyrotoxic patients from the subnormal but somewhat higher TSH levels of NTI (181, 182). Among a group of patients with low serum TSH values (<0.1 mU/L), almost all thyrotoxic patients had values less than 0.01 mU/L, when assessed with a highly sensitive assay, whereas most critically ill euthyroid patients had values between 0.01 and 0.1 mU/L (181). However, in another study, about 4% of patients with NTI had values below the functional sensitivity of a "third generation" assay, indicating that an absolute distinction cannot be made on the basis of TSH alone (182).

4.4 Differentiated Thyroid Cancer

The key laboratory measurements in the follow-up management of differentiated thyroid cancer are TSH and thyroglobulin; their interpretation is inter-dependent. Knowledge of the serum TSH concentration is important to confirm a level sufficiently high to maximally stimulate radioiodine uptake, whether for diagnostic or therapeutic purposes. For studies done after initial near total thyroidectomy, following withdrawal or temporary reduction of suppressive therapy (183), or with the use of recombinant human TSH (184), serum TSH levels in excess of 30 mU/l appear to achieve this objective (185). The failure of TSH to increase into this range suggests too short a period of T4 withdrawal, a compliance problem, or the presence of a substantial amount of active thyroid tissue.

Serum thyroglobulin measurement and whole body radioiodine scanning have generally been used in a complementary fashion, but there is now good evidence that current assays for thyroglobulin have greater sensitivity than follow-up whole body scanning with 2mCi 131I (150,150a). There is a clear trend to place greater emphasis on measurements of thyroglobulin and to move away from repeated low dose diagnostic whole body radioiodine scanning, a procedure that has limited sensitivity (150, 186). It may become standard practice to regard abnormal thyroglobulin levels as an indication for therapeutic radioiodine dosage of 100 mCi or more, followed by a whole body scan after the therapy dose to define the extent of iodine-avid tissue (186). Management of low risk patients now tends to be based on assessment of serum thyroglobulin after recombinant TSH stimulation, without the need for diagnostic scanning (186a)

During long-term follow-up, it is crucial that clinicians are fully informed by the reporting laboratory of methodological changes that may give a false impression of remission or recurrence. It should be noted that standardization of thyroglobulin assays remains arbitrary (77) and that wide discrepancies can occur between various methods. Some reports document lower values by radioimmunoassay than by immunoradiometric assay, without a clear relationship to detectable thyroglobulin antibodies (186b).

Serum thyroglobulin after stimulation with recombinant TSH, in conjunction with ultrasonography has been reported to have the highest sensitivity in monitoring differentiated thyroid carcinoma (186c). While ultrasonography has high sensitivity in detecting metastatic tissue in the neck (186d), its specificity remains uncertain. Assay of thyroglobulin on the needle washes from ultrasound-guided biopsy of extrathyroidal masses may be useful in determining whether or not a lesion is of thyroidal origin (150a).

4.5 Psychiatric Illness

An unusual variety of euthyroid hyperthyroxinemia occurs in some patients hospitalized with acute psychiatric illness ( 85 ). Serum T4 is increased, but serum T3 is less frequently elevated; serum TSH is generally normal or slightly high (187). These abnormalities, presumed to be due to central activation of the hypothalamic-pituitary-thyroid axis, often resolve in several weeks (85 ).

4.6 Free T4 estimates during pregnancy (see Chapter 14)

Interpretation of measured serum free T4 values during pregnancy is complicated by methodological differences (188, 189), but there is now consensus that serum free T4 and free T3 decrease in the second and third trimesters, with mean levels reduced about 20-40% below the normal mean (188,188a). Subnormal levels are uncommon. It is notable that Roti et al (189) found wide method-dependent variations in free T4 estimates between seven different commercial methods in euthyroid women at term. Albumin-dependent methods of free T4 estimation, for example analog tracer techniques, gave up to 50% of subnormal values (189), perhaps attributable to the decreased serum albumin binding of the assay tracer. These methods can be misinterpreted during pregnancy because of their marked negative bias. Conversely, because of an increase in the pool of protein-bound T4 during pregnancy, methods that involve a high degree of sample dilution would be expected to show positive bias in relation to standards that contain a normal concentration of TBG. For example, corrections of total T4 based on tracer dialysis in highly diluted serum to determine free fraction, tend to overestimate free T4 in the presence of TBG excess (190), thus obscuring the normal decline in free T4 as pregnancy progresses.