Keywords
proteome - metabolome - transcriptome - hypothyroidism - hyperthyroidism - selenium
- copper
Introduction
Hypothyroidism is a frequent condition with an estimated prevalence of 1.5%
in the northern and 4.1% in the southern parts of Germany [1]. It is conventionally treated by oral
monotherapy with levothyroxine (T4) as supported by all guidelines on hypothyroidism
[2]
[3]
[4]. Monotherapy with T4 as a prohormone
is both possible and safe, because it is converted intracellularly in target tissues
to the biologically active 3,3’,5-triiodothyronine (T3). T3 binds with high
affinity to the nuclear thyroid hormone (TH) receptors TRα and TRβ,
which are encoded by the two separate genes THRA and THRB. TRs
positively or negatively regulate TH dependent gene expression in almost all
tissues, providing the molecular explanation for the plethora of symptoms observed
in thyroid disorders [5]. Tissues and
cells can however to a large extent regulate their TH action independently of the
serum concentrations, as in addition to TRs other control layers exist. The first
prerequisite for TH action is the cell- and hormone-specific uptake of T3
and/or T4 by transmembrane transporters. Secondly, subsequent intracellular
conversion of TH is necessary, facilitated by a family of activating enzymes, the
selenium (Se) dependent deiodinases (DIOs). Type 1 (DIO1) and type 2 (DIO2)
deiodinases are capable of generating bioactive T3 from T4, whereas type 3 (DIO3)
acts as an inactivating enzyme [6].
Therefore, time- and tissue-specific alterations in the expression of any member of
this regulatory network contribute to explain why local TH action can differ between
target organs despite comparable and relatively constant circulating TH levels and
why there are distinct differences between species despite a high evolutionary
conservation of TH activation and TH receptors [5]. Consequently, it is possible that tissues display features of hypo-
or hyperthyroidism, although the serum profile of TH is in the euthyroid range. This
variability of local TH action can occur physiologically, e. g., by an
upregulation of DIO activity in certain periods of development [7], as part of a pathological condition,
for instance a downregulation of THRB expression in human non-alcoholic
steatohepatitis [8] or during
non-thyroidal illness syndrome [9], or as
a consequence of inactivating mutations in TH transporters or receptors [10]
[11]. As these conditions cannot be
assessed by serum TH concentrations, additional circulating biomarkers are urgently
required that provide information about the local TH action in a specific tissue,
e. g., the liver.
Diagnostic Approach and Current Limitations
Diagnostic Approach and Current Limitations
Currently, the accepted gold standard to diagnose primary hypothyroidism and to
monitor thyroid status under treatment is to measure serum thyrotropin (TSH), a
hormone produced in a TH- and TRβ-dependent manner in the anterior pituitary
[6]. This parameter is often superior
to the measurement of serum free or total TH, which may be subject to a number of
interfering factors such as the duration and severity of the underlying thyroid
disease [12]
[13]. In addition, measurement of TH,
particularly of free serum T3 or T4 levels, is methodologically demanding and
vulnerable to various interferences e. g., by drugs or structurally similar
endocrine disrupting compounds due to the very low serum concentrations of T3 [14]. Guidelines therefore advice against
drawing firm conclusions from minor alterations in circulating TH levels, but accept
TSH as the best single, robust, sensitive, and reproducible test for determining
thyroid function [15]. Yet, several
conditions are known, where relying on TSH serum concentrations are misleading. For
instance, in central hypothyroidism the normal glycosylation of TSH is reduced
leading to a partial or complete loss of bioactivity, while clinical assays
measuring TSH would not detect any differences due to a preserved immunoreactivity
of the less glycosylated protein [16].
TSH may thus still be inadequately maintained within the normal range despite low
circulating TH. As another example, a TSH-secreting pituitary tumor (TSHoma)
presents with high levels of circulating TSH and consequently highly elevated
circulating TH levels, which fail to suppress TSH expression. Using solely the serum
profile, this condition is impossible to distinguish from inactivating mutations in
TRβ (resistance to thyroid hormone β, RTHβ, [17]), which impair the
TRβ-controlled suppression of pituitary TSH expression by TH and also
presents with inappropriately high TSH serum concentrations for the corresponding
TH
levels [18]. Consequently, the current
standard of TSH measurement is clearly insufficient under certain pathological
conditions to faciliate a correct diagnosis.
Likewise, even in more common conditions such as substitution with T4 in
hypothyroidism, monitoring of thyroid function with TSH and TH concentrations fails,
because serum total or free T3 and T4 will vary with the timing and dose of T4
replacement [4]. In addition, serum TSH
may take several weeks to normalize in severe hypo- or hyperthyroidism. Furthermore,
in a cohort of almost 2000 athyroid patients studied under T4 replacement therapy
more than 20% did not maintain their free T4 and T3 levels within the
reference range and almost 30% showed an abnormally low fT3/fT4
ratio. Interestingly, the gradient between TSH and fT3 increased with the dose of
L-T4, indicating that TSH may not be sufficient for assessing euthyroidism [19]
[20]. Moreover, it is still
controversially discussed whether it is required to titrate TSH back to the optimal
range, as TSH might also be involved in the regulation of extrathyroidal tissues
expressing the TSH receptor, e. g., kidney or bone [21]. The unreliability of TSH as treatment
progression marker is unfortunate, as impaired psychological well-being, depression
or anxiety can still be present in 5–10% of patients despite normal
TSH levels under standard T4 substitution [22], and the resolution of hypothyroid symptoms is one of the most
important targets of TH replacement therapy in hypothyroidism [4]. Likewise, biological markers
indicative of hypothyroidism such as heart rate, pulse wave arrival time,
echocardiographic parameters of left ventricular function, Achilles reflex time,
voice fundamental frequency, or basal metabolic rate are not sensitive enough to be
used in the diagnosis or follow-up of therapy in hypothyroidism due to their high
variability and their lack of sensitivity and/or specificity [4].
Alternative Biomarkers
To overcome these problems with the key diagnostic markers to assess thyroid status,
other biochemical markers have been tested. The liver-derived sex hormone-binding
globulin (SHBG), another TRβ-dependent protein, is of some clinical
importance under certain conditions, but its specificity is reduced due to the
responsiveness to other non-thyroidal factors such as sex hormones [23]. A similar modulation by non-thyroidal
regulators has also been described for other potential biochemical markers of TH
status such as lipid parameters (total cholesterol, LDL cholesterol,
lipoprotein(a)), bone markers (osteocalcin, urinary n-telopeptides (NTX)) or
creatine kinase, ferritin, myoglobin, tissue plasminogen activator, ACE, and glucose
6-phosphate dehydrogenase [24]. These
limitations resulted in guideline recommendations that “tissue biomarkers of
thyroid hormone action are not recommended for routine clinical use, outside
of the research setting, since these parameters are not sensitive, specific, readily
available, or standardized” or “would require invasive procedures
such as tissue biopsy” [4].
Therefore, they are not part of the routine work-up of patients, although SHBG is
occasionally used clinically as an additional biomarker, for instance in the
aforementioned differential diagnosis of an inappropriately elevated TSH. Here,
elevated SHBG levels indicate liver thyrotoxicosis and point to TSHoma, whereas in
RTHβ the liver is partially protected from the effects of the high
circulating TH due to the impaired TRβ signaling, and SHBG levels would not
be increased.
Selenium and Copper as Novel Biomarkers
Selenium and Copper as Novel Biomarkers
More recent studies in humans and animal models with RTHβ suggested that the
trace elements Se and copper (Cu) could be helpful in the differential diagnosis
between TSHoma and RTHβ [25]
[26], as the T3-induced increase in serum
Cu depends on hepatic TRβ action. Consequently, in mice lacking TRβ
and individuals with RTHβ, the serum Cu levels were reduced, as despite the
elevated serum TH, the liver is in a hypothyroid state due to the lack of
TRβ action. In contrast, serum Cu was similarly elevated by T3 treatment in
control mice as well as animals expressing a mutant TRα1, demonstrating the
TRβ- but not TRα1-dependency of this regulation [26]. Most interestingly, serum Se was
elevated in RTHβ and T3-treated animals, but lower in mice expressing a
mutant TRα1, suggesting a TRα1-dependent regulation of this trace
element [25]
[26]. Since data from hyperthyroid humans
have not been available, we analyzed the serum Se and Cu levels now in our
experimental thyrotoxicosis model in humans [27]. The results showed an induction of serum Cu after 4 but not 8 weeks
of T4 treatment ([Fig 1a]), suggesting a
quick response to TH but compensation after a longer period of hyperthyroidism.
Serum Se levels did unexpectedly not change significantly (not shown).
Interestingly, the induction of Cu was not observed after 2 weeks treatment with a
comparable dose of T4 in mice ([Fig
1b]); however, levels of Se and Cu in the livers of these animals increased
markedly ([Fig 1c+d]). This
contrasts with previous studies showing elevated serum Cu concentrations, but
reduced hepatic Cu content after 2 weeks treatment with a high dose of T3 [26], which could indicate that TH
initially and at lower doses elevates hepatic Cu storage, while higher doses also
trigger the release from the liver e. g., as ceruloplasmin as demonstrated
previously [26]. Together with recent
studies showing a strong correlation between serum TH and Cu concentrations in
children with congenital hypothyroidism [28] or even in larger epidemiological studies in the United States [29], these observations support a positive
regulation of serum Cu by TH. However, before these findings can be translated to
clinical use to assess hepatic (TRβ dependent) TH action in humans,
additional clinical studies are required to substantiate these findings, their
possible responsiveness to non-thyroidal factors, their time- and dose dependency
as
well as the species-specific differences.
Fig. 1 a Serum Cu concentrations in human volunteers exposed for 8
weeks to a mild T4-induced thyrotoxicosis (grey box) and subsequent recovery
(31). b Serum Cu concentrations in mice treated with a comparable
dose of T4. c+d Hepatic Cu and Se content in the
animals from b. Values depict mean±SEM, p<0.001:
***; p<0.0001: ***
(repeated measure ANOVA with paired a or non-paired samples
b-d), corrected for multiple comparison with Holm-Sidak
test). Se and Cu were measured using a picofox instrument as published
previously [25]
[26].
Perspectives for New Biomarkers With Better Sensitivity to Monitor Local TH
Action
Perspectives for New Biomarkers With Better Sensitivity to Monitor Local TH
Action
Given that measurement of TSH serum levels fail in a number of important
pathophysiological conditions to adequately represent thyroid function and the
direct measurement of TH and/or of the available tissue biomarkers is
associated with other major limitations in routine use [15], additional sensitive means to
diagnostically capture the local action of TH in selected tissues are urgently
needed to improve the assessment of euthyroidism and guide pharmacological treatment
in common (T4 replacement therapy in hypothyroidism) as well as rare (inactivating
mutations in TRα, TRβ, TH transporters) conditions. Given the
sensitivity of several biomarkers to non-thyroidal regulators, it is unlikely that
a
single biomarker will be sufficient for routine clinical use. Therefore, recent
strategies employing an unbiased multifactorial approach, summarized as
“OMICs” techniques for the common ending in targets like genomics,
proteomics and metabolomics, may be suited to capture the pleiotropic effects of TH.
The development of high throughput analytical platforms with better sensitivity and
precision increasingly allows applying these techniques in many diagnostic fields,
and software packages may be able to extract a TH-dependent fingerprint of
tissue-selective biomarkers. To date, only few studies on OMIC analyses in blood in
relation to thyroid function have been performed, and within-subject studies before
and following T4 withdrawal revealed almost 500 differentially regulated targets
[30]. Another study investigated
patients with overt hypothyroidism at the time of diagnosis and after serum TSH
levels had been normalized for at least 6 weeks under T4 substitution therapy and
identified 20 differentially regulated protein spots [31]. Unfortunately, the heterogeneity of
the patient group is a major factor interfering with the reliability of the results,
as especially gender strongly modulates the effects of TH treatment. In a
within-subject study we aimed to avoid these problems by inducing hyperthyroidism
experimentally in male volunteers only [27]. Interestingly, despite clear biochemical changes fitting to
thyrotoxicosis 4 and 8 weeks under T4 treatment, none of the subjects developed any
clinical sign or any significant change in thyroid specific symptoms (with the
exception of a nightly increase in heart rate by an average of 6 bpm), neither as
compared to baseline before treatment nor as compared to the final state following
8
weeks after the treatment. These results support the previously described
dissociation between biochemical tests for thyroid function and symptoms as seen in
cross-sectional studies [27]
[32]. Subsequent proteome analysis,
conducted in collaboration with our partners in Greifswald, revealed a significant
TH status associated change in several protein signatures, namely in the abundance
of apolipoproteins (decreased APOD, APOB and APOC-III levels in the thyrotoxic
state) and of proteins related to the coagulation cascade and the complement system
(increased levels). Especially several coagulation factors, namely II, V, XI and
XIII were increased shortly after the induction of thyrotoxicosis [33], likely reflecting a
TRβ-dependent regulation as direct testing in RTHβ patients
confirmed an unchanged balance in the coagulation factors [34]. However, another study using a 2 week
experimental treatment period with T4 in healthy subjects did not observe
alterations in the complement system or blood clotting [35], suggesting that the timing of thyroid
dysfunction appears to be important and the response curve is most likely U-shaped,
as the proteome indicated an increased activity of these factors in hypothyroidism
as well [31].
Consequently, to unravel the underlying molecular mechanisms additional studies will
be needed, including animal models with impaired local TH action, i. e.
animal models for RTHα or RTHβ [25]
[26] as well as pharmacological studies
[36]. The majority of existing data
from animal models are mainly derived from the comparison of hypothyroidism to the
euthyroid state in rodents [37]
[38], which may not be of biological
relevance for other conditions. Moreover, there are some discrepancies to the human
situation especially with regard to the size of biological effects such as heart
rate or body temperature [27]
[36]. Furthermore, the direct comparison
of OMICs based biomarkers between mice and man is hampered by technical problems,
as
e. g., many of the most abundant proteins are different, thus impacting on
the rate of recovery of other rare peptides and proteins. Despite these possible
problems, parallel and discrepant regulation can be distinguished when comparing the
two species. For the coagulation system, hyperthyroidism in humans is associated
with an increased clot formation and a higher risk of venous thromboembolism whereas
coagulability in mice appears to be reduced [39]. Other systems like the apolipoproteins were reduced in parallel
between mice and humans. In our interspecies comparison study between human and mice
[27]
[36], we have already identified 16 serum
proteins that are concordantly altered by T4 in both species, including a
macrophage-derived protein [40]. This
demonstrates that such an approach has great potential to not only reveal a robust
set of TH dependent biomarkers, but also to understand the underlying regulatory
mechanisms and identify the involved components of the TH regulatory network or
non-thyroidal regulators.
Despite of discrepancies in proteome analysis, which may largely depend on the
methodological uncertainties, most of the pleotropic effects of TH would be expected
in the regulation of endogenous metabolic pathways. Therefore, techniques capturing
the small molecules comprising the metabolome have been used to identify changes in
more abundant metabolites such as carbohydrates, lipids, amino acids and alterations
in some low concentration molecules like organic acids [24], showing a different signature in
association to serum TSH or T4 [41],
which was partially confirmed in an independent study [42]. Most recently, a study in 4 different
animal models of RTHα demonstrated that using NMR, a clear metabolic
fingerprint of the disease in biofluids could be identified, which was stable over
time and dinstinctly different from hypothyroidism [43]. These initial findings in humans and
rodents suggests that despite the fact that many metabolites are variably regulated
by several different conditions, TH may still leave a detectable shift in the overal
metabolomic pattern, which could be used as a putative diagnostic tool. However, as
almost all of these metabolites are highly variable and by nature even more affected
by non-thyroidal regulators or external factors such as food, circadian rhythms or
drugs than protein biomarkers, it remains to be seen whether a metabolomic
fingerprint can be clinically useful. Moreover, the subtle differences in the
induction of hyper- or hypothryoidism in rodent studies regarding route of
administration and dose also need to be considered as variable, although recently
defined guidelines for these experimental protocols have been suggested [44].
Perspectives
The results of our experimental studies in mice and humans provided novel molecular
insight, which due to the tightly controlled conditions, are likely to represent
pure T4-dependent effects. This combined approach in animal models and humans also
allowed assessing organ- and TR-specific TH effects, for instance the TRβ
mediated regulation of hepatic Cu metabolism. If these initial findings can be
validated successfully, direct, specific and highly sensitive assays with high
discriminatory power may be developed in the future to establish new and more
targeted biomarkers providing valuable clinical information about the local TH
action.
Quick Summary
This minireview summarizes the status of the search for new tissue- and
pathway-specific biomarkers reflecting local thyroid hormone action.
