3.1.5 Nonthyroidal illness

A few hours after the onset of acute illness, marked changes in serum thyroid hormone levels occur. This is referred to as nonthyroidal illness (NTI). The most characteristic and persistent abnormality is a low level of serum triiodothyronine (T₃). Despite these low levels of serum T₃, patients usually have no clinical signs of thyroid disease. Other terms for this disease state have been used, e.g. the low T₃ syndrome and the euthyroid sick syndrome. In addition to nonthyroidal illness, a low T₃ in euthyroid patients is seen during caloric deprivation and after the use of certain types of medication (see Chapter 3.1.4).

Low levels of thyroid hormone in hypothyroidism are associated with a decreased metabolic rate. Both in nonthyroidal illness and in fasting there is a negative energy balance in the majority of cases. Therefore the low levels of T₃ during nonthyroidal illness and starvation have been interpreted as an attempt to save energy expenditure, and intervention is not required. However, this remains controversial and has been a debate for many years. In this chapter, the changes in thyroid hormone levels, the pathophysiology behind these changes, the diagnosis of intrinsic thyroid disease, and the currently available evidence whether these changes should or should not be corrected will be discussed (Box 3.1.5.1).

Within 2 h of the onset of acute illness (and after 24–36 h of fasting), T₃ levels decrease and reverse T₃ levels rise (1). The magnitude of these changes is related to the severity of the disease. The characteristic pattern of changes in thyroid hormone concentrations in relation to severity of disease is shown in Fig. 3.1.5.1. T₃ levels decrease progressively with increasing severity of disease without reaching a plateau, whereas reverse T₃ increases in relation to severity of disease, but reaches a plateau. It has been reported that reverse T₃ is not invariably elevated in all causes of nonthyroidal illness. It may be normal or even low in acute and endstage renal disease, the nephrotic syndrome, AIDS, and prolonged illness (2). However, recent studies have shown significantly elevated levels of serum reverse T₃ in patients with prolonged critical illness and acute renal failure requiring renal replacement therapy (3).

In mild illness, total and free thyroxine (T₄) levels may rise initially after the onset of disease but in severely ill patients, T₄ levels drop as well. Both low T₄ and T₃ levels, as well as high reverse T₃ levels are associated with a worse prognosis (1, 4). Thyroid-stimulating hormone (TSH) levels may rise briefly for about 2 h after the onset of disease, but despite the drop in serum T₃ (and in severe illness also T₄) levels, circulating TSH usually remains within the low to normal range.

In recent years, evidence has emerged that, in addition to severity of illness, duration of illness is another important determinant of the thyrotropic profile in critical illness (5, 6). Patients requiring intensive care for several days enter a more chronic phase of severe illness, and the low levels of thyroid hormone in prolonged illness have a more neuroendocrine origin (see also next paragraph). Pulsatility and circadian variation in TSH secretion is diminished in prolonged illness, and hypothalamic thyrotropin-releasing hormone (TRH) mRNA expression in patients who died from chronic severe illness is low compared to patients who died from an acute lethal trauma (7). In prolonged illness, both low TSH secretion and TRH expression correlate with the low T₃ levels. This results in even lower levels of T₃ and a low T₄ as well. Reverse T₃ levels remain elevated or may return back to normal with the decrease in serum T₄ (see Fig. 3.1.5.2a). Obviously, severity and duration of illness are both important factors that determine the changes in nonthyroidal illness, and mixed forms of the above-mentioned changes further complicate the interpretation of thyroid function tests in critical illness. In addition to severity and duration of illness, type of illness may also be important with regard to the changes in peripheral thyroid hormone levels (2, 3) (see below).

Both low T₄, low T₃, and high reverse T₃ levels are associated with a more severe illness and a worse prognosis. A study in 451 patients who received intensive care for at least 5 days showed that not only the absolute values but also the time course of serum thyroid parameters between survivors and nonsurvivors is completely different (4). In this study, TSH, T₄, T₃, and the T₃/reverse T₃ ratio increased in patients who survived, whereas there was no such rise in nonsurvivors (see Fig. 3.1.5.3).

Only a few studies have been reported investigating the tissue concentrations of thyroid hormone in critical illness. One study demonstrated that patients who died after critical illness had lower levels of tissue T₃ compared to patients who had died acutely, but the severity seems to vary from one organ to another (8). A different study in patients who had died in the intensive care unit showed that low levels of serum T₄ and T₃ correlated well with local concentrations in liver and skeletal muscle (9). This suggests that the decrease in serum T₃ (and in severe illness also T₄) in nonthyroidal illness also results in decreased tissue levels of thyroid hormone.

In primary hypothyroidism, TSH levels rise sharply in response to low levels of circulating thyroid hormones. After the onset of acute disease, TSH levels may rise briefly for about 2 h, but despite the ongoing decrease in circulating T₃ (and in severe illness also T₄) levels, TSH usually remains within the low to normal range in nonthyroidal illness. This suggests an altered feedback setting at the level of the hypothalamus and/or pituitary. The physiological nocturnal TSH surge is absent in this acute phase of illness (5) (see Fig. 3.1.5.4). These changes in the acute phase cannot be attributed to exogenous glucocorticoids or dopamine, since serum TSH is also in the low to normal range in patients without these drugs.

Different mechanisms have been proposed for this altered feedback setting. Studies in rodents during fasting and after lipopolysaccharide injection, which is a model for acute inflammation, show no compensatory rise in TRH in contrast to hypothyroid animals (10). Local expression of thyroid hormone-activating type 2 deiodinase (D2) is increased and expression of thyroid hormone-inactivating D3 is unaltered, resulting in higher local concentrations of hypothalamic T₃ (11, 12) (see Chapter 3.1.2 for a more detailed description of the function of the different iodothyronine deiodinases). This is in agreement with a down-regulation of TRH in fasting and acute illness. MCT8, a specific thyroid hormone transporter which is important in brain development, and OATP1C1, a high-affinity T₄ transporter, are both expressed in the hypothalamus, suggesting an important role for these transporters in the hypothalamic set point (13). Animal data also show that an altered transmembrane transport of thyroid hormone at the levels of the pituitary and/or hypothalamus may be involved in the altered hypothalamus–pituitary–thyroid axis set point, as well as an enhanced occupancy of nuclear T₃ receptors in the pituitary thyrotrophs.

In the acute phase of critical illness, circulating levels of cytokines are usually high. Injection of cytokines such as interleukin (IL)-1, IL-6, and tumour necrosis factor-α (TNFα) is at least partially able to mimic the thyrotropic alterations of the acute stress response. Studies in IL-12 and IL-18 knockout mice show that these cytokines are also involved. However, cytokine antagonism fails to restore thyroid hormone levels, both in animals (IL-1, IL-6, TNFα, interferon) and in humans (IL-1) (14, 15).

High levels of endogenous cortisol may also contribute to the blunted TSH response in acute illness. In addition, endogenous thyroid hormone analogues, such as thyronamines and thyroacetic acids, may also contribute to the pathogenesis of nonthyroidal illness, by blunting the TSH response to low levels of thyroid hormone and/or by competing with thyroid hormone for binding to transport proteins, transmembrane transporters, deiodinases, and/or nuclear receptors.

Patients with prolonged critical illness have a more severe central dysfunction. In addition to the absent nocturnal TSH surge, TSH pulsatility diminishes dramatically and hypothalamic TRH expression is reduced (5, 6). Both low TSH secretion and low TRH expression correlate with the low T₃ levels in prolonged critical illness (7). Patients who die after severe illness have less than one-half the concentration of hypothalamic and pituitary T₃ compared to patients who die acutely from trauma (8). This combination of a low TRH expression in the hypothalamus, a low TSH secretion from the pituitary, and low levels of T₃ in both tissues implies a major change in the hypothalamus–pituitary feedback regulation, and suggests a more central origin of the low T₃ syndrome in prolonged illness. The positive correlation of TSH secretion and TRH expression with serum T₃ levels also points in this direction. In contrast, an increase in TSH is a marker for recovery, which suggests that recovery from the low T₃ syndrome is also initiated centrally (see Fig. 3.1.5.3 in which the recovery of TSH precedes the recovery of the T₃/reverse T₃ ratio). In addition, continuous infusion with TRH (especially when combined with a growth hormone secretagogue) is able to (partially) restore serum TSH, T₄, and T₃ in prolonged critical illness, both in humans and in animals (16).

The pathophysiology behind this suppression of the hypothalamus–pituitary–thyroid axis is not fully understood. Circulating cytokines are usually low in the chronic phase of severe illness, so other mechanisms must be involved. An up-regulation of hypothalamic D2, which is seen in animal models of acute illness, and/or a down-regulation of D3 could suppress TRH expression via relatively high concentrations of hypothalamic T₃. However, hypothalamic and pituitary T₃ levels are low in patients who die after prolonged illness (8). This makes an important contribution of hypothalamic and pituitary deiodinases to the central suppression in prolonged critical illness less likely. Similarly, the low levels of hypothalamic T₃ in prolonged illness make an important contribution of thyroid hormone transporters to the altered set point in chronic illness less likely.

Other pathways, such as the melanocortin signalling pathway and neuropeptide Y (NPY), seem to be involved in regulating hypothalamic TRH secretion in chronic critical illness as well (6, 17, 18). However, the exact role of these neuropeptides in nonthyroidal illness is not yet elucidated. Different experimental and clinical conditions show different results. For example, patients with chronic illness showed weak immunocytochemical staining of NPY cells in the infundibular nucleus compared to patients who died acutely, and low NPY expression was associated with decreased TRH mRNA expression in the paraventricular nucleus (18). However, an inverse relationship is observed during fasting (19), again illustrating that changes in fasting and acute illness should not be extrapolated to the chronic phase of severe illness.

Exogenous glucocorticoids and dopamine are known to suppress the hypothalamus–pituitary–thyroid axis, and perhaps prolonged hypercortisolism and/or endogenous dopamine in these patients may also play a role.

In the acute phase of critical illness and after starvation, changes in thyroid hormone levels are mainly caused by changes in the peripheral metabolism of thyroid hormones and by alterations in the capacity of serum binding proteins (1). In the more chronic phase of critical illness, these changes persist but a decreased T₄ production by the thyroid is superimposed on the altered peripheral metabolism. The serum T₃/reverse T₃ ratio is the most accurate reflection of the peripheral metabolism, since this ratio is independent of variations in binding proteins and independent of a decreased T₄ production by the thyroid (see also Chapter 3.1.2). The decrease in serum T₃ and increase in serum reverse T₃ that occurs within a few hours of the onset of disease suggests major changes in the peripheral metabolism of thyroid hormone.

Approximately 20% of serum T₃ is produced by the thyroid, whereas the rest is derived from conversion of T₄ in peripheral tissues such as liver and skeletal muscle. The availability of T₃ for nuclear thyroid hormone receptors is largely regulated by different transmembrane transporters and by three deiodinases that catalyse deiodination of the different iodothyronines (20) (Fig. 3.1.5.2b, see also Chapter 3.1.2). D1 is present in liver, kidney, and thyroid, and plays a key role in the production of serum T₃ from T₄ and in the breakdown of the inactive metabolite reverse T₃. D2 is present in brain, anterior pituitary, thyroid, and skeletal muscle. D2 also converts T₄ to the active hormone T₃. D2 is important for local T₃ production in tissues such as the brain, but the enzyme in skeletal muscle may also contribute to plasma T₃ production. D3 is present in brain, skin, placenta, pregnant uterus, and various fetal tissues, and is the major T₃ and T₄ inactivating enzyme by converting T₄ and T₃ to reverse T₃ and T₂, respectively. D3 protects tissues from excess thyroid hormone. All three deiodinases are selenoproteins and use reductive compounds, such as reduced glutathione, as cofactors.

The fall in serum T₃ levels and increase in serum reverse T₃ levels in the acute phase of critical illness and in fasting are largely due to a decreased conversion of T₄ to T₃ and of reverse T₃ to T₂, as demonstrated by multiple kinetic studies (see reference (1) for a review of the literature). Different factors contributing to these decreased conversions have been proposed, such as a decreased tissue uptake due to a negative energy balance and increased levels of bilirubin and nonesterified fatty acids, decreased availability of selenium and/or cofactors, and drugs inhibiting deiodinase activity.

Liver and skeletal muscle biopsies obtained minutes after death of intensive care unit patients demonstrated that liver D1 activity in these patients is low compared to values observed in healthy individuals, except for patients who died acutely from severe brain damage (Fig. 3.1.5.2b) (3). Deiodinase activities were measured in tissue homogenates in the presence of excess cofactor, which suggests a down-regulation of D1 independent of the above- mentioned mechanisms. Low levels of D1 activity were clearly correlated with high levels of reverse T₃ and a low T₃/reverse T₃ ratio, independent of duration of illness (see Fig. 3.1.5.5). D1 activities also showed a clear correlation with local T₃ concentrations and T₃/reverse T₃ ratios in liver (9).

No D2 activity could be detected in skeletal muscle samples of these patients, although there is evidence that D2 activity is expressed in normal skeletal muscle. A reduced activity of D2 may therefore also contribute to the low levels of T₃ in the acute phase of nonthyroidal illness. However, other studies show some D2 expression in muscle, especially after prolonged illness (21). This suggests that an altered D2 activity may not play a role in the pathogenesis of the low T₃ syndrome in prolonged critical illness.

A clear induction of D3 activity was demonstrated in both liver and muscle samples of these patients, whereas these tissues normally do not express D3 in adult subjects (Fig. 3.1.5.2b). High liver and high muscle D3 activity was associated with high serum and local tissue reverse T₃ levels (see Fig. 3.1.5.5). D3 induction was independent of duration of illness. From these data it can be concluded that a down-regulation of thyroid hormone-activating D1 (and in the acute phase of illness also D2) and an induction of thyroid hormone-inactivating D3 are important factors contributing to the low levels of T₃ and high levels of reverse T₃ in nonthyroidal illness.

Tissue deiodinase activities and serum thyroid hormone levels are significantly associated with cause of death (3). A postmortem study in over 60 patients demonstrated that liver D1 activity and serum T₃/reverse T₃ were highest in patients who died from severe brain damage, intermediate in those who died from sepsis or excessive inflammation, and lowest in patients who died from cardiovascular collapse (see Fig. 3.1.5.6). Liver D3 showed an opposite relationship. There was no relation between deiodinase activities and a marker of inflammation (C-reactive protein), but patients who needed inotropes and/or those requiring dialysis because of acute renal failure had a lower liver D1 activity and higher liver and muscle D3 activity. This suggests that poor tissue perfusion and cellular hypoxia may be an important determinant regulating deiodinase activities in vivo. Recently it has been shown that D3 activity and D3 mRNA are increased by hypoxia and by hypoxia mimetics that increase hypoxia-inducible factor 1 (22). This supports the hypothesis that up-regulation of D3 by cellular hypoxia may be a way to alter thyroid hormone bioactivity during limited oxygen supply (see also Chapter 3.1.2).

Thyroid hormone mediates its effects by binding to nuclear T₃ receptors, resulting in initiation or repression of transcription. Depending on the target tissue, nuclear T₃ is derived from plasma and/or from intracellular generation from T₄. This means that both T₃ and T₄ have to cross the plasma membrane of target cells for biological action (see Chapter 3.1.2). The process of uptake of thyroid hormones by cells is rate limiting for subsequent intracellular metabolism and nuclear T₃ binding.

Uptake of T₄ by human hepatocytes is temperature, Na⁺, and energy dependent, and kinetic analyses indicate that T₄ and T₃ cross the plasma membrane by different transporters (23). Kinetic studies have shown that fasting and nonthyroidal illness result in attenuation of uptake of liver T₄ and reverse T₃, probably via decreased concentrations of intracellular ATP. Liver T₃ uptake is less sensitive to intracellular ATP concentrations. Inhibition of thyroid hormone uptake has also been shown with nonesterified fatty acids and bilirubin, both elevated in critical illness, and certain drugs such as amiodarone. It has not been studied whether these alterations in transport persist during prolonged illness, but there is no evidence to assume otherwise.

In recent years, different thyroid hormone transporters have been identified, exhibiting a different tissue distribution, substrate specificity, and selectivity. Human MCT8, with a preference for T₃ over T₄, is probably the best studied transporter. Mutations in MCT8 lead to a phenotype of severe psychomotor retardation (24). Transport activity by MCT8 is not Na⁺ and/or energy dependent, but MCT8 is expressed in, among other tissues, liver and skeletal muscle. In critically ill patients, neither liver nor skeletal muscle MCT8 expression were related to the ratio of the serum over tissue concentration of T₄, T₃, or reverse T₃ (9). This suggests that MCT8 is not crucial in the transport of these iodothyronines over the plasma membrane in liver and skeletal muscle. However, this does not exclude an important regulatory function of other known (i.e. MCT10) and as yet unknown thyroid hormone transporters in nonthyroidal illness, and needs to be addressed in future studies.

Different thyroid hormone receptor (TR) isoforms are generated from the THRA and THRB genes by alternative splicing and different promoter usage. THRA encodes five proteins, but only TRα1 has intact DNA- and T₃-binding domains. There is evidence that TRα2 acts as a dominant negative isoform. THRB encodes three proteins that can bind T₃ and DNA. The T₃-binding thyroid hormone receptors are highly homologous, except in the N-terminal α- and β-domains (25) (see Chapter 3.1.2). Both in the liganded and unliganded state, thyroid hormone receptors bind to T₃-response elements (TREs) in the promoter region of target genes. Unliganded thyroid hormone receptors repress basal transcription. Binding of T₃ releases corepressors and allows recruitment of coactivators required for gene expression above basal levels. TRα1 and TRβ1 are ubiquitously expressed; TRα1 is expressed preferentially in brain, heart, and bone, and TRβ1 preferentially in liver, kidney, and thyroid.

Little is known about the regulation of thyroid hormone receptors in nonthyroidal illness. In rats, it has been shown that starvation results in a decreased expression and occupancy of hepatic thyroid hormone receptors (26). In peripheral mononuclear cells in humans, an increased expression of both TRα and TRβ has been demonstrated in patients with chronic liver and renal disease, whereas in patients in the intensive care unit only TRβ mRNA was increased (27). In patients with liver disease, both liver TRα (20-fold) and liver TRβ (fivefold) were increased compared to healthy liver controls. A postmortem study in 58 subjects who had died in the intensive care unit showed an increased expression of the TRα1/TRα2 ratio (active isoform/dominant negative isoform), which was positively related to severity of disease and age (28). In this study, no relation between severity of disease and TRβ1 expression was observed.

The clinical relevance of these changes is not yet clear. One might argue that an increase in the expression of the active receptor isoforms is an adaptive response by the body to decreasing levels of thyroid hormone. On the other hand, a higher thyroid hormone receptor expression with low levels of T₃ will lead to an increase in the percentage of unliganded receptors, which would have an opposite effect. Interestingly, no relation was demonstrated between liver TRβ1 mRNA levels and serum thyroid hormone parameters in critically ill patients, although D1 expression is, among other things, regulated by T₃ via TRβ1 (28).

Alternate metabolic pathways of thyroid hormone metabolism include sulfation and glucuronidation. Sulfated iodothyronines do not bind to thyroid hormone receptors, and sulfation mediates the rapid and irreversible degradation of iodothyronines by D1. Inner ring deiodination (see Chapter 3.1.2) of T₄ and T₃ by D1 are markedly facilitated after sulfation, whereas outer ring deiodination of T₄ is completely blocked after sulfation. As a consequence, serum concentrations of sulfated iodothyronines are usually low. D2 and D3 are incapable of deiodinating sulfated iodothyronines.

Elevated levels of T₄ sulfate and T₃ sulfate/T₃ ratios have been reported in patients with nonthyroidal illness, and postmortem serum T₄ sulfate levels in critically ill patients were positively correlated with the length of stay in the intensive care unit (29, 30). Low hepatic D1 activity in these patients plays an important role in the increased levels of T₄ sulfate.

Glucuronidation in nonthyroidal illness may be important with regard to the use of several drugs. In particular, the anticonvulsant drugs carbamazepine and phenytoin and the antituberculous drug rifampicin have been shown to induce hepatic glucuronidation. This may lower T₄ levels, but T₃ and TSH levels are usually unaffected.

More than 99% of iodothyronines are bound by serum thyroid hormone-binding proteins (see Chapter 3.1.2) leaving only a small proportion in the free form, about 0.02% of T₄ and 0.3% of T₃ and reverse T₃. As thyroxine-binding globulin (TBG) binds the bulk of circulating thyroid hormones, any change in its binding capacity will markedly affect total hormone levels. TBG, transthyretin, and albumin are decreased in nonthyroidal illness as a reflection of the catabolic state of the patient. However, increased TBG levels can be present in liver disease. Different drugs that are used in severe illness may cause alterations in serum binding of thyroid hormone, either by decreasing TBG (e.g. glucocorticoids) or by displacing thyroid hormones from binding proteins (e.g. acetylsalicylic acid, furosemide) (see also Chapter 3.1.4). However, the altered ratios of thyroid hormones that occur in critical illness must be independent of any variation in serum binding capacity.

In addition, patients in the intensive care unit are frequently treated with heparin. In these patients, the measured concentration of serum free T₄ can be higher than the true in vivo concentration. This is the result of heparin-induced lipase activity during sample storage and incubation, resulting in in vitro generation of nonesterified fatty acids which displace T₄ and T₃ from TBG (see also Chapter 3.1.4). Low-molecular-weight heparin preparations have a similar effect.

Low binding of thyroid hormone in nonthyroidal illness due to the presence of a circulating binding inhibitor has been proposed in older studies (1). However, exogenous T₄ administration can easily replenish the T₄ pool in patients with prolonged illness, making it unlikely that such a binding inhibitor is an important cause of the low levels of T₄ in prolonged nonthyroidal illness (31).

Evaluation of thyroid status in nonthyroidal illness can be very difficult, especially in patients in the intensive care unit, not only regarding interpretation of laboratory results, but also on clinical grounds as signs and symptoms of the illness may imitate or mask any accompanying thyroid disease. Despite these difficulties the value of clinical examination in this respect should not be underestimated. Thus, the presence of eye signs (ophthalmic Graves’ disease), goitre, and a family history of thyroid disease or autoimmune disease in general, are important points that may be supportive for the diagnosis of autoimmune thyroid disease.

Because of the changes that occur in serum thyroid parameters in critical illness and because there is currently no evidence that these changes should be treated, thyroid function should not be tested in critically ill patients unless there is strong suspicion of thyroid disease. In unselected hospitalized patients in the late 1980s, TSH was undetectable (at that time TSH assays were less sensitive, defining undetectable as <0.1 mU/l) in only 3.1% of cases, whereas TSH concentration was above 20 mU/l in only 1.6% of patients (32). When thyroid function is tested, measurement of TSH alone is often not sufficient. Most free T₄ assays are unreliable in critical illness, due to alterations in binding proteins, increased use of heparin in an intensive care unit setting, and the possible presence of circulating binding inhibitors. Therefore, the total thyroid hormone should be measured as well (Box 3.1.5.2).

In a patient with nonthyroidal illness suspected of having hyperthyroidism, serum TSH is the most helpful test. If serum TSH is still within the normal range, the presence of thyrotoxicosis is virtually excluded. But when serum TSH is low, this could be a consequence of nonthyroidal illness or it could be caused by hyperthyroidism. However, nonthyroidal illness almost never results in TSH levels less than 0.01 mU/l (33, 34). Nearly all patients with a low but detectable TSH level will have normal thyroid function tests after recovery from illness. On the other hand, approximately 75% of patients with the low T₃ syndrome and a TSH level of less than 0.01 mU/l have hyperthyroidism (35). Interpretation of serum TSH becomes more difficult in patients treated with TSH-suppressing agents such as dopamine and corticosteroids, which are often used in intensive care units. Additional measurement of T₄ and T₃ levels is mandatory, but should be interpreted with care. T₄ levels are low in approximately 50% of critically ill patients, and T₃ levels are low in the majority of patients, which could mask active hyperthyroidism. However, serum T₄ and T₃ levels should be high (or high to normal) in hyperthyroidism, and low (or low to normal) in nonthyroidal illness.

The diagnosis seems straightforward for critically ill patients with suspected hypothyroidism and elevated serum TSH. However, in patients recovering from nonthyroidal illness, TSH levels may become temporarily elevated. As well as for hyperthyroidism, the magnitude of the change in TSH is important. Nevertheless, even in patients with TSH levels of more than 20 mU/l, hypothyroidism is permanent in only about 50% of cases (32). T₄ and T₃ levels may help to differentiate these patients, since patients with permanent hypothyroidism had significantly lower levels of T₄ and T₃. Most patients with an elevated TSH level (<20 mU/l) will have normal thyroid function tests after recovery from illness, especially if thyroid peroxidase and thyroglobulin antibodies are negative.

In central hypothyroidism, serum TSH is usually low and differentiation from nonthyroidal illness on the basis of these data becomes very difficult. Other pituitary deficiencies and related clinical signs are commonly present in these patients, but prolonged critical illness often leads to suppression of other neuroendocrine axes as well (5). Serum reverse T₃ may be helpful in some cases, since reverse T₃ levels are high in patients with nonthyroidal illness. However, reverse T₃ assays are not available in all centres and reverse T₃ levels may be slightly high in mild hypothyroidism as well. In general, a high T₃/T₄ ratio and a low reverse T₃ favour the presence of hypothyroidism over nonthyroidal illness and vice versa. If no definite diagnosis can be established, thyroid function tests should be repeated after recovery from illness.

Both in acute and in prolonged critical illness, low levels of thyroid hormone are associated with a higher mortality rate, but it remains controversial whether nonthyroidal illness is an adaptation protecting against catabolism or a maladaptation. It is important to re-emphasize the teleological differences between the acute and chronic phase of severe illness (Fig. 3.1.5.2). Acute changes within the thyroid axis after the onset of critical illness (low T₃ and elevated reverse T₃) are similar to the changes observed in starvation. These changes have been interpreted as an attempt to save energy expenditure and protein wasting and do not need intervention. Thyroid hormone replacement in fasting subjects results in an increased nitrogen excretion and negative nitrogen balance, suggesting catabolism. Whether this also applies to the changes in the acute, and especially in the more chronic phase of critical illness, is controversial. Thyroid hormone treatment in critically ill rats shows no beneficial or even negative effects.

In humans, only a few studies have been performed, and studies were carried out in few patients. So far, no clear beneficial effect on clinical outcome has been demonstrated. Intravenous T₄ (150 μg) administration every 12 h for 48 consecutive hours in 28 patients with acute renal failure was even associated with an increased mortality compared to the control group (36). This might have been due to the suppression of TSH in the treatment group, although free T₄ and free T₃ levels were similar in both groups. Intravenous T₄ (1.5 μg/kg per day) administration to 11 patients with nonthyroidal illness for 14 days did not alter the outcome compared to 12 control patients (31). T₄ levels returned to the normal range in the treated patients, but serum T₃ concentrations remained low and did not differ between the two treatment groups. This is probably due to the decreased T₄ to T₃ conversion, which is seen in both the acute and chronic phase of critical illness, and by the accelerated breakdown of T₄ and T₃ by D3.

Because of the decreased T₄ to T₃ conversion in nonthyroidal illness, T₃ treatment may be a better choice. However, T₃ will also be degraded by D3, and T₃ treatment may be harmful as well. T₃ administration to 14 patients with burn injuries did not improve outcome compared to placebo-treated patients (37). An improved cardiac function has been observed in different studies in adult patients treated with pharmacological doses of T₃ after coronary artery bypass grafting, and in dopamine-treated children who received T₃ substitution after cardiopulmonary bypass surgery (38, 39). However, no effect on (perioperative) survival has been demonstrated.

In the chronic phase of critical illness, altered thyroid hormone levels appear to have a more central origin, although the peripheral metabolism is also altered (see Fig. 3.1.5.2). Studies performed in fasting subjects and in patients with acute critical illness should therefore not be extrapolated to the chronic phase of severe illness. Serum thyroid hormone levels in prolonged illness are negatively correlated with markers of increased protein degradation and bone resorption, suggestive of catabolism (16).

In a recent study, tissue thyroid hormone levels were measured in patients who stayed on the intensive care unit for more than 5 days (9). Some of these patients were treated with a combination of T₄ and T₃, but not in a randomized controlled study. Patients were treated if they had a serum T₄ concentration below 50 nmol/l, a normal TBG, and clinical signs of hypothyroidism. Higher serum T₃ levels in treated patients were accompanied by higher levels of tissue T₃. However, the increase in liver T₃ concentrations in patients who received thyroid hormone was disproportional compared to the increase in serum and muscle T₃ concentrations (c.4 times higher in liver compared to c.2 times higher in serum and skeletal muscle). In addition, TSH levels were suppressed in patients who were treated with thyroid hormone, suggesting overtreatment although their serum T₃ levels were still in the low or low to normal range. So, if patients are given thyroid hormone therapy, should we aim for thyroid hormone levels within or still below the normal range?

Intervention with hypothalamic releasing factors has the advantage that the negative feedback inhibition of thyroid hormone on the pituitary is maintained, thereby providing a safer therapy option. It has been shown that in patients with prolonged critical illness, and in an animal model, continuous infusion of TRH in combination with a growth hormone secretagogue is able to restore thyroid hormone levels. In these patients, this therapy resulted in a reduction of catabolic markers. Whether this also results in a beneficial effect on mortality remains to be addressed in future studies.