Keywords

Introduction

The terms non-thyroidal illness syndrome (NTIS), low T3 syndrome, and euthyroid sick syndrome have been interchangeably used to describe a state of low serum total T3 levels associated with various illnesses and starvation. NTIS can be associated with any illness and occurs very rapidly after the onset of acute stress; its prevalence is very high among hospitalized patients and it is considered a predictor of clinical outcome (Alevizaki et al. 2007; Plikat et al. 2007). It is conceivable that NTIS represents a physiological response to serve homeostasis and reduce energy expenditure.

Two major mechanisms contribute to the alterations in thyroid function tests (TFTs) in NTIS: a central component with impaired feedback of the hypothalamus-pituitary-thyroid (HPT) axis and changes in peripheral TH metabolism and action.

In this chapter the most recent literature on NTIS is summarized with an overview of basic facts on thyroid hormone (TH) metabolism, key points in the diagnosis of NTIS, and evidence associating the syndrome with disease severity and prognosis. The plausible pathophysiologic mechanisms underlying NTIS are discussed in detail, and data on the management of patients with NTIS are reviewed.

Thyroid Hormone Metabolism and Action

The secretion of TH is tightly regulated by the HPT axis. The release of TRH from the hypothalamus stimulates the synthesis and release of TSH from the anterior pituitary leading to the secretion of TH from the thyroid gland. TSH is normally under strong genetic control, as shown in studies of healthy twins (Hansen et al. 2004). The individual reference range is much narrower than that in the population (Andersen et al. 2002). The major secretory product is T4, while T3 is released in much smaller amounts. T3 is the most metabolically active TH; 80% of T3 is derived from extrathyroidal tissue by T4 conversion to T3. The classic genomic action of T3 is exerted through binding to the thyroid hormone receptors (TR) TRA1, TRB1, and TRB2, which form heterodimers with retinoid X receptor (RXR) (Brent 2012). This results in conformational changes, dissociation of corepressors, recruitment of coactivators, and recognition of thyroid hormone response elements (THRE) in promoters of target genes to initiate transcription and eventually protein synthesis (Oetting and Yen 2007). T3 exerts an inhibitory action on TRH and TSH synthesis and secretion via THR signaling and completes a negative feedback loop (Fig. 1).

Fig. 1
figure 1

Overview of the hypothalamus-pituitary-thyroid axis and the genomic action of thyroid hormones. RXR retinoid X receptor, TR thyroid hormone receptor, TRE thyroid response element

Full size image

In target organs, thyroid hormone availability and cellular action is regulated by deiodinases, which belong to a selenocysteine containing enzyme family. T4 is converted to T3 by type 1 and type 2 iodothyronine deiodinase (D1 and D2, respectively), whereas the role of type 3 deiodinase (D3) is mainly the conversion of T4 to the metabolically inactive reverse T3 (rT3) (Fig. 2) (Arrojo and Bianco 2011).

Fig. 2
figure 2

Summary of the action of the deiodinases on thyroid hormone metabolism. D1 deiodinase 1, D2 deiodinase 2, D3 deiodinase 3, rT 3 reverse T3, T 2 3,3′-diiodo-I-thyronine

Full size image

Definition of NTIS

The term non-thyroidal illness syndrome is used to describe an ensemble of thyroid function abnormalities in the absence of intrinsic thyroid disease. These abnormalities typically include isolated hypotriiodothyroninemia, whereas in more severe or prolonged cases, low T4 levels may also be found. Serum TSH concentration remains low or normal in most cases, and TSH elevation may be observed during the recovery period.

The decrease of thyroid hormone levels during acute illness and fasting has been long documented in the seminal studies of Burger et al. (Burger et al. 1976) and Harris et al. (Harris et al. 1978). NTIS has a high prevalence, up to 50% among hospitalized patients, and it has been most commonly reported in relation to myocardial infarction, coronary artery bypass grafting (CABG), infectious disease, sepsis, trauma, brain injury, chronic obstructive pulmonary disease, gastrointestinal disease, burns, malignancy, surgery, and hospitalization in the intensive care unit (ICU) (Pappa et al. 2011).

Association of NTIS Severity with Outcome

When the illness is mild or acute, the only manifestation is usually low total T3 levels, whereas when the NTIS is more severe or prolonged, total T4 levels may also drop, while TSH remains normal or paradoxically low for the decreased TH levels (Table 1).

Table 1 Course of thyroid function tests (serum hormone levels) in the evolution of the non-thyroidal illness syndrome. NL within normal range
Full size table

Several studies have correlated the degree of decrease in TH levels with the severity and prognosis of the underlying disease and demonstrated that the lower the serum T3 (and, in some cases, serum T4), the higher the mortality risk in these patients (Alevizaki et al. 2007; Plikat et al. 2007). In studies performed in the ICU setting, nonsurvivors had significantly lower T3, T4, and TSH levels and higher rT3 compared to survivors (Peeters et al. 2005; Rothwell and Lawler 1995). The scale of the decline in T3 and T4 levels and the increase in rT3 have been found to reflect the severity of the underlying illness, predict mortality risk, and correlate with several cardiopulmonary and functional parameters, such as the Acute Physiology and Chronic Health Evaluation (APACHE) II score and the degree of disability after acute stroke (Alevizaki et al. 2007) (Fig. 3). Interestingly, an association was found with T3 and rT3 concentrations and the activity of liver D1 and skeletal muscle D3 activities in postmortem tissues. In patients who died of cardiovascular collapse and of brain damage liver D1 activity was lowest (Peeters et al. 2003).

Fig. 3
figure 3

Kaplan–Meier survival curve of patients with “low T3” compared to patients with “normal T3” over the first 12 months after the acute stroke (Reproduced from Alevizaki et al. 2007)

Full size image

Although studies of NTIS in pediatric populations are limited, there are a few studies on infants and children with heart disease or undergoing cardiac surgery, which show a strong correlation between the severity of TFTs alterations and clinical outcome. Children with severe NTIS (lowest serum T3 and T4 concentrations) have prolonged hospitalization and increased requirement of mechanical ventilation and pediatric ICU stay (Marks 2009).

There is also evidence that elevated total rT3 level is associated with mortality even in the independently living elderly population, suggesting that this may serve as a marker of declining health (Forestier et al. 2009).

Diagnosis

It is crucial to correctly and timely differentiate NTIS from other causes of altered TFTs, in which specific treatment is warranted, although this may frequently be challenging. The hallmark of NTIS is low circulating total T3 levels with inappropriately normal TSH, accompanied by elevated total rT3 levels. In severe or protracted diseases, total T4 levels may decline and a decrease in TSH may be noted.

The patients’ clinical history and physical examination are useful in the differential diagnosis between NTIS and intrinsic thyroid disease. An elevated TSH concentration is the key alteration of TFTs in primary hypothyroidism. However, TSH levels may decrease in the acutely ill patients, especially those receiving dopamine or corticosteroids (Haugen 2009). The presence of elevated anti-thyroperoxidase or thyroglobulin antibody levels and/or classical hypoechogenicity at thyroid ultrasound also supports the diagnosis of primary hypothyroidism (Jonklaas et al. 2014). The combination of high TSH with low T4 levels is suggestive of primary hypothyroidism, although this thyroid profile may also be observed during recovery from NTIS. Therefore, in some cases diagnosis may be delayed until after recovery from the acute illness and hence it is recommended to observe the course of TFTs. Regarding the differential diagnosis of NTIS from secondary hypothyroidism, testing the function of other pituitary axes, such as cortisol, prolactin, and gonadotrophin levels, may provide important clues to the diagnosis (Alexopoulou et al. 2004).

Of special note, the effect of several drugs on thyroid function tests should be taken into consideration (Table 2) (Haugen 2009). Dopamine, a frequently used medication in critically ill patients, is known to suppress TSH and significantly decrease T4 and T3 levels to the level of hypothyroidism. Besides dopamine and dopamine agonists (e.g., bromocryptine), other medications, such as glucocorticoids, somatostatin analogues, and bexarotene, an RXR agonist, are associated with substantial TSH suppression (Brabant et al. 1989; Samuels et al. 1992; Ohzeki et al. 1993). The role of certain antiepileptic medication (carbamazepine, oxcarbamazepine, valproic acid) and the biguanide metformin in inhibiting TSH secretion has been reported in some studies (Miller and Carney 2006; Vigersky et al. 2006), but lacks confirmation and, therefore, remains unclarified. In addition, the clinician should be aware of drugs that inhibit the secretion of T4 and T3, such as lithium, iodide, amiodarone, and aminoglutethimide. Further, salicylates and high doses of furosemide inhibit binding of T4 and T3 to thyroxine-binding globulin (TBG) leading to transient increases in free T4, whereas phenytoin increases the hepatic metabolism of thyroid hormones (Haugen 2009; Larsen 1972).

Table 2 Summary of drugs interfering with thyroid function tests. TH thyroid hormones, TBG thyroid hormone-binding globulin (Adapted from Haugen 2009)
Full size table

It is important to correctly identify patients with severe hypothyroidism presenting with myxedema coma, a diagnosis associated with high mortality rate (Hampton 2013). Of note, in case of co-occurring NTIS, TSH concentration in these patients may not be as markedly increased. Characteristic features in the clinical presentation that guide the diagnosis include hypothermia and altered mental status. The key points in the management of these patients include supportive care, treatment of the precipitating illness (usually myocardial infarction, infection, or cerebrovascular accident), and administration of stress dose corticosteroids followed by judicious replacement with TH.

Pathogenetic Mechanisms

The traditional view of NTIS was that the decline of TH levels is the result of reduced activity of D1 in the liver, leading to decreased conversion of T4 to T3 in the periphery, and enhanced D3 activity, resulting in high levels of the inactive metabolite rT3. However, in the last decades, there is strong evidence that the pathogenesis of NTIS does not merely involve impaired deiodination of thyroid hormones in the liver, but rather includes two major components: (1) a peripheral one involving changes in TH metabolism and action in target tissues (alterations in deiodinase activity, expression of thyroid hormone receptors, transporters, and binding proteins) (Peeters et al. 2003; Mebis et al. 2009a) and (2) a central one with alterations in the HPT axis and impaired negative feedback in the hypothalamus and the pituitary (decreased TRH secretion and TSH pulsatility) (Fig. 4) (Fliers et al. 1997; de Vries et al. 2015). Also, it appears that these alterations are dependent on the timing, nature, and severity of the underlying illness.

Fig. 4
figure 4

Schematic representation of the main elements underlying the pathophysiology of non-thyroidal illness syndrome. TH thyroid hormones

Full size image

Deiodinases

The majority of literature is in agreement with NTIS being an adaptive, beneficial response to illness in order to reduce the availability of the active T3 and, thus, decrease energy expenditure and limit catabolism (Everts et al. 1996). This is partly mediated by alterations in the activity of deiodinases, in specific suppression of D1, which decreases the conversion of T4 to T3, and induction of D3 resulting in inactivation of TH and rise in total rT3 (Peeters et al. 2003). It appears that the role of deiodinases may be less significant than previously recognized, whereas the changes occurring at the level of the hypothalamus and the pituitary (discussed below) are of greater importance.

Deiodinase 1

The pattern of D1 expression in NTIS has been extensively studied. D1 is localized in the plasma membrane and is expressed mainly in the liver, kidney, thyroid, and pituitary. Several animal models of NTIS and studies in patients with critical illness have demonstrated a significant decrease in liver D1 expression and activity and a correlation with low liver T3 concentration (Peeters et al. 2003).

Two elegant hypotheses have been proposed to explain the suppressed D1 activity in NTIS: Critical illness and stress result in high cytokine levels that activate the nuclear factor kappa-light-chain-enhancer of activated B cell (NF-kB) and the activator protein-1 pathways. These inflammatory signaling pathways are thought to compete with THR for limiting amounts of steroid receptor coactivator-1 (SRC-1), which in turn results in reduced THR activation and, subsequently, downregulation of D1 expression (Yu and Koenig 2000, 2006) (Fig. 5). In the same study, administration of SRC-1 prevented the decrease of TH levels following lipopolysaccharide (LPS) injection and restored D1 expression in the liver. However, another study demonstrated that THRB knockout mice had similar decreases in TH and D1 levels after LPS treatment as wild type mice, suggesting that THRB does not play an important role in regulating D1 expression and activity (Kwakkel et al. 2008).

Fig. 5
figure 5

Possible mechanisms explaining D1 suppression in the non-thyroidal illness syndrome. NF-kB nuclear factor kappa-light-chain-enhancer of activated B cell, AP-1 activator protein-1, SRC-1 steroid receptor coactivator-1, TR thyroid hormone receptor, GSH glutathione

Full size image

Another scenario is that oxidative stress and release of reactive oxygen species lead to depletion of glutathione stores, required for D1 catalytic activity to ensue. This was further supported by the finding that addition of N-acetylcysteine, an antioxidant that restores intracellular glutathione levels, prevented the interleukin-6 (IL-6)-induced D1 suppression (Wajner et al. 2011).

The concept of decreased D1 activity as the causal factor for low T3 levels in NTIS has been challenged by studies in D1-deficient mice in that these have normal T3 levels, suggesting that D1 suppression may in fact be the consequence rather than the cause of low T3 concentration (Schneider et al. 2006). In addition, it was demonstrated that the decrease of T3 levels in wild type mice precedes the decrease in liver D1 expression, further supporting the idea that D1 inhibition is not causally related with the TFTs alterations in NTI (Schneider et al. 2006).

Deiodinase 2

Deiodinase 2 is a major enzyme involved in local TH metabolism and tissue T3 production (Bianco and Kim 2006). D2 is localized in the endoplasmic reticulum and is tightly regulated by TH levels. Specifically, T3 downregulates D2 mRNA expression, and T4 and rT3 (which are substrates for D2) increase its ubiquitination and proteosomal degradation (Bianco and Kim 2006). Skeletal muscle D2 is considered a major source of T3 under basal conditions. Studies using animal models of NTIS as well as studies in ICU patients have demonstrated that D2 expression in skeletal muscle is increased and thought to be the result of increased cAMP response element-binding protein (CREB) signaling, which functions as a transcriptional antagonist of the thyroid hormone receptor (Mebis et al. 2007). However, in studies performed in septic patients, muscle D2 expression was decreased, and this was attributed to the effect of fasting and caloric deprivation (Rodriguez-Perez et al. 2008).

The increasingly recognized role of D2 in alterations of the central component of NTIS will be discussed in detail below in the relevant section of the chapter.

Deiodinase 3

The role of D3, the major TH inactivating enzyme, in the pathogenesis of NTIS has also been studied. D3 is localized in the plasma membrane and is expressed in brain neurons, liver, and the innate immune system. The effect of illness on D3 expression is dependent on the timing and type of illness. It has been shown that fasting and prolonged illness result in marked increases in D3 expression and activity in the liver and also muscle (Fig. 6) (de Vries et al. 2015). These changes are considered to serve thyroid economy and reduce total energy expenditure in the catabolic setting of illness. Low leptin levels have been hypothesized to explain the high D3 activity in the liver during fasting, whereas the upregulation of D3 in muscle may be the result of hypoxia and decreased tissue perfusion and represent an adaptive mechanism to acute stress (Simonides et al. 2008; Boelen et al. 2012). In support of the latter, it was recently shown that D3 induction during myocardial infarction is associated with expression of several microRNAs that promote the proliferative capacity of cardiomyocytes (Janssen et al. 2016).

Fig. 6
figure 6

Pathways underlying D3 induction in the non-thyroidal illness syndrome

Full size image

Interestingly, D3 activity was increased in granulocytes, a component of the innate immune system, during illness. Myeloperoxidase (MPO) is an important enzyme abundantly expressed in granulocytes, critical for their antimicrobial activity. In their study, Boelen et al. suggested that the increased D3 activity observed in NTIS is a means to provide MPO with iodide and thus augment the bactericidal machinery and immune defense (Boelen et al. 2008).

There is convincing data supporting that D3 expression is increased in various tissues in NTIS. However, and contrary to previous considerations, the alterations in D3 activity in NTIS do not seem to be involved in changes in serum TH levels, since D3 knockout and wild type mice have similar serum TH concentrations during inflammation (Boelen et al. 2011).

Thyroid Hormone-Binding Proteins

One of the earliest explanations for the etiology of NTIS involved changes in TH binding (Chopra et al. 1985). This was supported by findings of reduced concentrations of albumin and other TH-binding proteins and of their reduced affinity to thyroid hormones. Also, it was found that serum total TH levels were significantly decreased in NTIS, whereas free TH levels dropped only modestly, further supporting the role of altered TH binding (Chopra 1998).

Prolonged illness and malnutrition are states of high catabolism and associated with lower albumin and transthyretin levels, whereas the acute phase response results in a decrease in TBG and, thus, total TH concentrations (den Brinker et al. 2005). Serine protein inhibitors, such as serpins, are activated at sites of inflammation and inactivate TBG. It has been demonstrated that in acute stress, such as during CABG, TBG is rapidly degraded by protease cleavage, leading to an instant drop in total T3 levels in serum (Afandi et al. 2000). Another hypothesis supports the presence of TBG-binding inhibitors. Reducing the affinity with their main transporter in serum, thyroid hormones are released from the binding proteins and their clearance is increased (Jirasakuldech et al. 2000). Elevated levels of bilirubin and unsaturated nonesterified fatty acids (NEFAs) may impair TH binding, transport, and peripheral conversion of T4 to T3 (Lim et al. 1993). As an example, heparin, which is frequently used in critically ill patients, induces the generation of NEFAs, whereas drugs, such as high dose furosemide, antiepileptics, and salicylates, decrease the binding of T4 to TBG (Bayer 1983). In addition, it has been postulated that alternative pathways, such as sulfation, glucuronidation, and ether link cleavage may accelerate TH degradation and clearance, further reducing total TH levels in serum (Wu et al. 2005). The contribution of these pathways has not been well studied.

On the other hand, the notion of impaired TH binding as a causal factor in NTIS has been disputed (Brent and Hershman 1986). In this study, patients with severe NTIS were treated with LT4, which quickly restored serum T4 levels. Finding that the TH pool could easily be replenished with LT4 argues against a significant role of TBG deficiency or TH-binding inhibitors in the pathophysiology of NTIS.

Cytokines

Cytokines, such as tumor necrosis factor alpha (TNF-alpha), interleukin 1 and 6 (IL-1 and IL-6, respectively), are major mediators of the acute phase response and regulate the release of acute phase proteins and stress hormones (Moshage 1997). Their downstream pathway involves activation of nuclear factor kappa-light-chain-enhancer of activated B cell (NF-kB), a major transcription factor in stress and inflammation. There is a plethora of data from studies in animal models of NTIS and in man, on the role of cytokines in the illness induced decrease of TH levels, yet these data are often contradictory and the role of cytokines in the pathogenesis of NTIS is not straightforward (Fig. 7). IL-6 levels were negatively correlated with serum T3 in hospitalized patients. Furthermore, infusion of TNF-alpha, IL-1, and IL-6 resulted in a decrease of TH levels mimicking NTIS, although administration of neutralizing antibodies did not reverse the alterations in thyroid function (van der Poll et al. 1995).

Fig. 7
figure 7

Summary of the role of pro-inflammatory cytokines in non-thyroidal illness syndrome. TPO thyroid peroxidase, NIS sodium iodide symporter, Tg thyroglobulin, TH thyroid hormones

Full size image

Several studies have shown that pro-inflammatory cytokines result in changes in the expression of genes involved in TH production and metabolism. In specific, IL-1 infusion in human and rat cell culture impaired thyroperoxidase (TPO) mRNA expression and protein content, as well as the sodium iodide symporter (NIS) mediated iodide uptake under basal conditions and upon TSH stimulation (Gerard et al. 2006). Interferon-gamma (IFN-gamma) impaired TSH-induced TH and thyroglobulin (Tg) secretion and mRNA expression, as well as TSH-induced TPO and NIS expression and iodide uptake (de Vries et al. 2015). Moreover, administration of TNF-alpha inhibited the TSH-induced cAMP response, Tg secretion, and NIS expression (Tang et al. 1995). Cytokines were also found to inhibit D1 expression and activity and play a major role in central D2 upregulation in the hypothalamus (mechanism described in detail below) (de Vries et al. 2015; Fekete et al. 2004, 2005).

In aggregate, the literature is in agreement on cytokines being key partners in the pathogenesis of NTIS. This concerns interference with various parts of the synthesis pathway of TH in the thyroid, from iodide uptake to TH secretion, as well as the central component of the HPT axis. It seems that it is not a single cytokine, but rather a network of pro-inflammatory molecules, components of the acute phase response, that are implicated (de Vries et al. 2015).

Thyroid Hormone Transporters and Receptors

The main feature of NTIS is the low T3 levels, observed not only in serum but also at the tissue level (liver, kidneys, brain, lungs). This is considered an adaptive mechanism in response to catabolic states, in order to reduce energy requirements.

The transport of thyroid hormones in the cells is subject to the energy state intracellularly. When T4 transport in the liver is inhibited, this is rate limiting for the total plasma T3 production, because there is decreased substrate available for conversion to T3. In the study of Kaptein et al., T4 tissue transport in NTIS was decreased by 50% and T3 production by 70%. The authors suggested hepatic ATP depletion as an underlying mechanism, since NTIS patients are in a negative energy balance, whereas the acidic cellular environment in acute illness and the presence of high concentrations of NEFAs might also explain the decreased T4 uptake in the liver (Kaptein et al. 1982).

Regarding TR signaling, there is data from mouse models of NTIS after LPS administration, which evokes a systemic inflammatory response, showing decreased liver TR expression. Along the same line, patients with severe NTIS suffering from septic shock had very low mRNA levels of TRs (both TRA1 and TRB1), RXR, and the thyroid hormone specific transporter MCT8 in their muscle and adipose tissue (Mebis et al. 2009a).

Taken together, these data indicate that NTIS reflects a systemic stress response and most likely serves as a mechanism to combat critical illness by suppressing energy demands.

Alterations in the Central Component of the HPT Axis

The hypothesis that has increasingly gained ground over the last decades is that the etiology of NTIS is mainly a disturbed negative feedback regulation at the level of the hypothalamus and the pituitary, and this is particularly relevant in prolonged illness. It is now believed that in prolonged illness reduced hypothalamic stimulation of the pituitary thyrotrophs and, subsequently, the thyroid gland leads to reduced TH secretion (de Vries et al. 2015; Warner and Beckett 2010; Van den Berghe 2014). Another supporting argument is that recovery from NTIS is usually heralded by a rise in TSH levels (Table 1).

Hypothalamus

Hypothalamic TRH neurons play an important role in the set point of thyroid hormone homeostasis. The main feature of the altered HPT axis in NTIS is the suppression of TRH expression. TRH expression was much lower in the hypothalamic paraventricular (PVN) nuclei in nonsurvivors with chronic illness compared to those with acute illness, and, additionally, a positive correlation was observed between TRH mRNA in the PVN and antemortem tissue T3 and TSH (Fliers et al. 1997). Also, in a pioneer study of Van den Berghe et al., infusion of TRH resulted in increased T4 and T3 concentrations. This intervention led to an increase in rT3 as well, which was overcome by combining TRH with a GH-secretagogue, which prevented the rise of rT3 and induced an anabolic response (Van den Berghe et al. 2002).

Several factors have been proposed leading to TRH suppression, including the neurohormonal agents endogenous dopamine and cortisol (Haugen 2009; Alkemade et al. 2005). However, there is accumulating evidence for local induction of D2 inhibiting TRH secretion in the PVN (Fig. 8). The inflammatory response in critical illness can increase D2 and decrease D3 expression in the hypothalamus, leading to a local elevation of T3 levels altering the negative feedback (Mebis et al. 2009b).

Fig. 8
figure 8

Proposed central mechanisms of the non-thyroidal illness syndrome. Alpha-MSH alpha melanocyte stimulating hormone, AGP agouti-related protein, NPY neuropeptide Y

Full size image

T3 may be taken up by TRH neurons by either diffusion from the cerebrospinal fluid or by axonal terminals of the TRH neurons present in the median eminence. Another proposed mechanism is that T3 is released in the arcuate nucleus influencing the neurons projecting in the PVN (Fekete and Lechan 2007). Fekete et al. elegantly demonstrated a marked increase in D2 mRNA expression in tanycytes, specialized cells lining the wall of the third ventricle, in a rat model of NTIS after LPS administration (Fekete et al. 2004). The induction of local D2 leads to increased T4 to T3 conversion, which can influence adjacent neurons in a paracrine fashion and lower TRH mRNA expression.

Possible triggers for the increase of local D2 expression are: inflammation- induced NF-kB and its associated downstream pathway and a decrease in leptin concentration (Kwakkel et al. 2009). Regarding the latter, it is known that during fasting leptin levels decrease accompanied by a decrease in the production of alpha-melanocyte stimulating hormone (alpha-MSH). Moreover, agouti-related protein (AGP) and neuropeptide Y (NPY) attenuate CREB phosphorylation in TRH neurons (Vella et al. 2011). As a result, the set point for feedback inhibition of the TRH gene by thyroid hormones is lowered. Interestingly, the upregulation of D2 in the hypothalamus appears to be independent of the decrease in TH levels, contrary to D2 expression in the pituitary or other areas in the brain, such as the cortex (Escobar-Morreale et al. 1997). An alternative hypothesis could involve D2 induction in the pituitary, which suppresses local TSH mRNA, but this has not been validated in animal models of prolonged NTIS (Beck-Peccoz and Mariotti 2000).

Pituitary

Patients with NTIS have inappropriately normal or low levels of TSH in the presence of low T3 (Table 3). The inappropriately low TSH levels in the presence of low T3 and the absence of TSH pulsatility are prominent in prolonged NTIS.

Table 3 Causal factors of low TSH levels in non-thyroidal illness syndrome
Full size table

This could be the result of suppressed leptin, mediating changes similar to those induced by fasting. Leptin acts via the hypothalamic arcuate nucleus, induces pro-opiomelanocortin and alpha-MSH production, and activates melanocortin 4 receptor (MC4R) (Walley et al. 2009). Studies in mice have shown that fasting results in decreased leptin, suppressed MC4R activation and, thus, decreased TRH and TSH secretion, whereas leptin administration prevents the fasting-associated TSH decrease (Boelen et al. 2006). This mechanism is considered to be energy saving, although, paradoxically, in studies of humans with NTIS leptin levels are either normal or elevated (Bornstein et al. 1997).

As discussed previously, critical illness stimulates D2 expression in the mediobasal hypothalamus, thus increasing local T3 availability and suppressing TRH, which subsequently inhibits TSH expression. Alternatively, the locally produced T3 can be transported from the hypothalamus to the pituitary via the portal capillaries and directly inhibit TSH expression.

Furthermore, illness and prolonged restriction of macronutrients result in a reduction of peripheral TH uptake and of THR activation, and the feedback loop leads to a decrease in TSH expression. The TSH decrease following LPS administration is blunted in THRB knockout compared to the wild type mice, suggesting that the NTIS-associated decrease in TSH expression is dependent on THRB signaling (Fekete and Lechan 2014).

There is much evidence that in NTIS the diurnal rhythm and nocturnal surge of TSH may be lost and TSH response to TRH blunted. There is also data supporting that TSH biological activity is reduced due to impaired glycosylation, possibly mediated by cytokines or glucocorticoids (Rothwell and Lawler 1995). In a mouse model of acute NTIS after LPS injection, inflammation induced by cytokines resulted in increased D1 and D2 activities in the pituitary along with decreased TSH expression (Boelen et al. 2004).

Overall, in NTIS, it appears that cytokine activation elicited by inflammation and fasting are the two core elements affecting various pathways in the hypothalamus and the pituitary. Especially with regard to decreased nutritional intake, the aim is clearly to decrease T3 in favor of thyroid economy, reduce the metabolic rate, and prevent muscle breakdown in order to promote survival. This way NTIS becomes a protective response towards critical illness.

Acute and Prolonged NTIS

A major difference between acute and protracted illness is that in prolonged NTIS peripheral tissues respond to increase rather than to decrease the availability of thyroid hormones, in an attempt to limit catabolism. Compensation for low tissue T3 is achieved through upregulation of D2 expression and activity (in skeletal muscle) (Mebis et al. 2007), increased expression of thyroid hormone receptors (contrary to the acute illness, in which their expression is decreased), and increased expression of TH transporters. In a rabbit model with prolonged NTIS, expression of transporters MCT8 and MCT10 increased in liver and muscle, respectively, which was reversed after treatment with T4 and T3 (Mebis et al. 2009a).

Clinical Management

The role of treatment of NTIS patients with thyroid hormone has been fiercely debated over the last decades, but currently most experts seem to agree that the majority of critically ill patients do not clearly benefit from TH treatment. Surprisingly, only few randomized controlled trials (RCT) have been performed to properly address this issue. In most studies, pharmacologic doses of either T4 or T3 were used and none showed that treatment results in improved patient outcome (Brent and Hershman 1986). A RCT in patients with acute renal failure even showed increased mortality in the group receiving thyroxine therapy (Acker et al. 2000). There are several studies evaluating the effect of T3 supplementation in heart disease patients, the majority of which showed improvement in cardiac parameters but no significant benefit as regards survival and prognosis (Spratt et al. 2007; Bennett-Guerrero et al. 1996; Pingitore et al. 2008). The limited number of studies with small patient numbers and poor statistical power are limitations that do not allow rejecting, confidently, a beneficial role of treatment of NTIS with T4 and/or T3. In addition, only few studies have used mortality and morbidity rates as primary endpoints and most have evaluated the effect of TH therapy on indirect indices, such as cardiopulmonary and functional parameters. There has also been significant variability in the age composition of the study populations and the severity of the underlying illnesses. Lastly, even in patients receiving TH treatment, the intervention did not normalize the tissue levels of TH (Peeters et al. 2005).

Overwhelming evidence indicates that NTIS is an adaptive response to critical illness, and at least part of the TH alterations accompanying critical illness can be explained by fasting (Van den Berghe 2014). Yet, there is a subset of patients with protracted disease, who are well nourished and have TFTs compatible with NTIS. Signs and symptoms of hypothyroidism may be found in these patients, which could be central given the background of suppressed TRH expression. In the presence of true hypothyroidism, recovery from the underlying illness can be delayed and the evolution of the hospital stay complicated (Schulman and Mechanick 2012). Additionally, an important consideration for patients who have known hypothyroidism and are admitted to the ICU is that treatment with levothyroxine should be considered during their stay in the ICU, although continuation of chronic care is not a primary focus in the ICU.

Although there is currently no compelling evidence to advocate use of TH in critically ill patients, individuals with very low TH levels (T4 below 4 ug/dl- or 51 nmol/L) who have high mortality rates represent a small subgroup that might benefit from treatment with TH (De Groot 2006). If therapy was to be given in these selected cases, and in the absence of contraindications such as cardiac decompensation and arrhythmias, one could consider initially giving higher LT3 doses to rapidly restore the TH pool, followed by lower replacement doses and coadministration of LT4. Adjustment of the dosing schedule should follow serum T4 and T3 values targeting low normal TH levels. As deiodination increases, LT3 doses can gradually be tapered and LT4 increased (De Groot 2006). It has been argued that patients with NTIS have selenium deficiency, which may contribute to the decreased deiodinase activity. In line with that seen in elderly without serious disease (Winther et al. 2015), selenium supplementation in ICU patients had only a modest effect on TH levels, and therefore selenium is currently not considered to have a role in the treatment of NTIS (Berger et al. 2001).

Alternative Therapies

An alternative approach, which focuses on the pathophysiology of reduced TRH expression in prolonged illness, would be infusion of hypothalamic releasing factors. The aim is to reactivate the HPT axis and induce an anabolic response, which is required to augment the recovery from prolonged illness. Van den Berghe et al. introduced the approach of combination therapy with TRH and GH-releasing peptide-2, intravenously, and showed encouraging results in terms of restoring TH concentrations, TSH pulsatility, as well as overall metabolic indices (bone markers and anabolic variables) (Van den Berghe et al. 1999). This pioneer study highlighted that patients with NTIS have a multifaceted illness that requires management of multiple hormonal deficits and correction of their catabolic status. However, the study included only a limited number of patients. Large, well-designed and adequately powered RCTs are needed to properly investigate the effect of treatment of NTIS with pituitary secretagogues.

Specific Populations

The management of NTIS in premature infants deserves specific attention. Such infants, without exception, present with some degree of hypothyroxinemia. They also often have concurrent illness, such as respiratory distress and infection, which further aggravates NTIS. It is well established that untreated congenital hypothyroxinemia in neonates has a deleterious impact on brain development. Taken together, it seems reasonable to consider lowering the threshold of TFTs alterations when treating premature infants. This is also supported by a study showing a lower mortality rate in premature infants receiving prophylactic combined T4 and T3 treatment compared to untreated ones (Schonberger et al. 1979). Complicating the interpretation, a meta-analysis demonstrated failure of TH treatment in reducing mortality, improving severity of underlying disease or improving neurodevelopmental outcome (Osborn and Hunt 2007). A recent study also failed to show a significant benefit of T4 supplementation on the growth pattern and neurodevelopmental outcome in very low birth weight infants (Uchiyama et al. 2015).

There is limited data on NTIS in pediatric populations, and most studies have been performed in children undergoing cardiac surgery. Only few prospective studies have assessed the effect of intravenous T3 treatment in infants and children. These either found no significant difference in outcome measures of illness severity or showed only modest improvement in ventilation requirements and hospital stay, yet with minimal adverse events (Marks 2009).

Regarding treatment of NTIS in subjects undergoing CABG, there is abundant data from studies using animal models demonstrating improvement in cardiac contractility and left ventricular function with T3 replacement, as well as a decrease in systemic vascular resistance. Patients were found to require less inotropic support and improve their hemodynamic parameters. This was also demonstrated in the pediatric population, but with no clear benefit in terms of survival (Portman et al. 2010).

A limited number of small studies in patients with congestive heart failure have found that LT3 therapy decreased the systemic vascular resistance and improved the cardiac output and the neurohumoral profile. A decrease in serum norepinephrine, N-terminal pro-B type natriuretic peptide, and aldosterone was noted along with an increase in the left ventricle end-diastolic volume. However, compelling evidence of improved survival is currently lacking (Sacca 2009).

Another specific population to be mentioned is that of brain dead heart donors. Studies in animals as well as man have shown that intravenous T3 therapy restored hemodynamic and biochemical abnormalities, decreased inotropic support, and preserved the cardiac function prior to transplantation (Novitzky et al. 2014). Again, confirmation by RCTs is awaited. Because T3 therapy was not deleterious and might even be beneficial, treatment of brain dead heart donors with T3 has been advocated by many (McKeown et al. 2012).

Summary

The prevalence of the non-thyroidal illness syndrome (NTIS) is high among hospitalized and critically ill patients. NTIS is independently associated with the severity of the underlying disease and serves as a marker of prognosis. Over the last decades, our knowledge concerning the pathogenetic mechanisms behind NTIS has expanded. The syndrome has two major components: one that involves altered feedback regulation at the level of the hypothalamus and the pituitary, and a peripheral one interfering with the production, transport, and action of thyroid hormones. The bulk of the literature is in agreement with NTIS being a physiologic, adaptive response to environmental factors (nutrient availability and inflammatory and stressful stimuli) and an energy conserving mechanism. Most evidence suggests that replacement therapy with TH is not required in NTIS. It could be considered as an option in carefully selected individuals with severe NTIS, in whom some degree of tissue hypothyroidism may be present.

NTIS reflects the severity of the underlying disease and is not causally linked to the primary illness. Therefore, efforts should focus on improving the management of the disease itself, rather than supplementing the patients with thyroid hormone to improve their thyroid function tests.