Is pituitary TSH an adequate measure of thyroid hormone-controlled homoeostasis during thyroxine treatment?

in European Journal of Endocrinology
(Correspondence should be addressed to R Hoermann; Email: rudolf.hoermann@gmail.com)

Objective

In recognition of its primary role in pituitary–thyroid feedback, TSH determination has become a key parameter for clinical decision–making. This study examines the value of TSH as a measure of thyroid hormone homoeostasis under thyroxine (T4) therapy.

Design and methods

We have examined the interrelationships between free triiodothyronine (FT3), free T4 (FT4) and pituitary TSH by means of i) a retrospective analysis of a large clinical sample comprising 1994 patients either untreated or on varying doses of l-T4 and ii) independent mathematical simulation applying a model of thyroid homoeostasis, together with a sensitivity analysis.

Results

Over a euthyroid to mildly hyperthyroid functional range, we found markedly different correlation slopes of log TSH vs FT3 and FT4 between untreated patients and l-T4 groups. Total deiodinase activity (GD) was positively correlated with TSH in untreated subjects. However, GD was significantly altered and the correlation was lost under increasing l-T4 doses. Ninety-five per cent confidence intervals for FT3 and FT4, when assessed in defined TSH concentration bands, differed significantly for l-T4-treated compared with untreated patients. Higher doses were often needed to restore FT3 levels within its reference range. Sensitivity analysis revealed the influence of various structural parameters on pituitary TSH secretion including an important role of pituitary deiodinase type 2.

Conclusion

The data reveal disjoints between FT4–TSH feedback and T3 production that persist even when sufficient T4 apparently restores euthyroidism. T4 treatment displays a compensatory adaptation but does not completely re-enact normal euthyroid physiology. This invites a study of the clinical consequences of this disparity.

Abstract

Objective

In recognition of its primary role in pituitary–thyroid feedback, TSH determination has become a key parameter for clinical decision–making. This study examines the value of TSH as a measure of thyroid hormone homoeostasis under thyroxine (T4) therapy.

Design and methods

We have examined the interrelationships between free triiodothyronine (FT3), free T4 (FT4) and pituitary TSH by means of i) a retrospective analysis of a large clinical sample comprising 1994 patients either untreated or on varying doses of l-T4 and ii) independent mathematical simulation applying a model of thyroid homoeostasis, together with a sensitivity analysis.

Results

Over a euthyroid to mildly hyperthyroid functional range, we found markedly different correlation slopes of log TSH vs FT3 and FT4 between untreated patients and l-T4 groups. Total deiodinase activity (GD) was positively correlated with TSH in untreated subjects. However, GD was significantly altered and the correlation was lost under increasing l-T4 doses. Ninety-five per cent confidence intervals for FT3 and FT4, when assessed in defined TSH concentration bands, differed significantly for l-T4-treated compared with untreated patients. Higher doses were often needed to restore FT3 levels within its reference range. Sensitivity analysis revealed the influence of various structural parameters on pituitary TSH secretion including an important role of pituitary deiodinase type 2.

Conclusion

The data reveal disjoints between FT4–TSH feedback and T3 production that persist even when sufficient T4 apparently restores euthyroidism. T4 treatment displays a compensatory adaptation but does not completely re-enact normal euthyroid physiology. This invites a study of the clinical consequences of this disparity.

Introduction

Thyrotoxicosis and hypothyroidism have been traditionally defined as states where elevated or decreased concentrations of circulating free thyroid hormones exert their own appropriate influences on the numerous metabolic outputs and organ functions of the body (1). Cellular access to triiodothyronine (T3) is tightly controlled by two mechanisms, an active transport into the cell and an enzymatic conversion of thyroxine (T4) to T3 by deiodinases, which exist in different subtypes (2, 3). Conversion is also the main source of T3 in humans, making up about 80% of its production, the rest being directly supplied by the thyroid gland (4). Failure of the thyroid gland or any change in hormone production is sensitively reflected at the pituitary level via a negative feedback within the hypothalamus–pituitary–thyroid regulatory loop. As a result, statistically derived TSH reference ranges have shaped definitions of thyroid dysfunctions, introducing subclinical disease entities and thus have been given the various roles of screening tool, therapeutic target and prognostic marker (5, 6, 7, 8, 9).

The TSH-centred definition of thyroid function has both merits and disadvantages. While TSH has become accurately and readily measurable, the exact reference values remain elusive (5, 10, 11, 12). Moreover, in an individual, normal distribution of thyroid hormone concentrations encompasses only a fraction of the span of the population reference range and the TSH set point is particularly tightly controlled (13).

On the other hand, appreciation of the interplay of TSH, free T3 (FT3) and free T4 (FT4) should facilitate a more individualised approach than relating analytic decisions to the broad inter-individual variation of each parameter. This concept requires a clear understanding of the interaction of these parameters under various physiological and pathophysiological conditions, in particular the implications of their homoeostatically controlled equilibria.

We have therefore sought to assess the interrelation of FT3, FT4 and pituitary TSH, querying reliance on TSH as a reliable measure for thyroid hormone homoeostasis under T4 therapy. To this end, we i) retrospectively analysed a broad clinical sample and ii) tested our conclusions by a mathematical simulation.

Materials and methods

Patients and laboratory samples

This study extends an earlier publication where full details of methods were described (14). Data were obtained from a large number of unselected patients referred for thyroid testing to the Departments of Endocrinology or Nuclear Medicine of the Klinikum Ludenscheid, Germany, between October 2006 and January 2007. Excluded from analysis were patients with pituitary or hypothalamic disorders, critically ill patients, patients with renal failure and pregnant women, as those conditions can interfere unpredictably with thyroid testing.

The focus of this study was on treatment-related aspects of the thyroid–pituitary relationship. To this end, 1994 patients and as many samples (including only the first sample per patient) were included for analysis. Of them, 1159 patients were untreated, receiving neither thyroid medication nor any other drugs at the time of sampling. Others were on levo T4 (l-T4) or iodine in varying doses of 50–200 μg/day (n=785) or 100 or 200 μg/day (n=50) respectively. The small group of iodine-treated patients was deliberately separated from untreated subjects to determine any effects of supplementation, as iodine status is of known influence on the FT3:FT4 ratio. Median patient age was 61 years (range 18–90 years). The majority of patients were women (78%) and seen in an ambulatory setting (80%). Thyroid peroxidase antibodies (TPO Abs) or TSH receptor antibodies (TSH R Ab) were measured only when clinically relevant to confirm or exclude suspected thyroid autoimmune disease. In the majority of patients with hypothyroidism in its overt or subclinical form (n=190), the condition was due to thyroid ablative therapy, surgery and/or radioiodine treatment; 53/190 patients (28%) suffered from thyroid autoimmunity, as evidenced by elevated serum levels of TPO Abs above the reference level. For classification purpose, subclinical thyroid disorders were defined by a TSH value outside reference limits, plus corresponding FT3 and FT4 values within the reference interval. The study was approved by the local ethics committee.

Laboratory methods

TSH was measured using an enzyme immunoassay (Abbott), with a working range of 0.01–100 mU/l and laboratory-evaluated reference range of 0.4–4.1 mU/l. FT3 and FT4 were determined by non-analogue, two-step immunoassays from the same manufacturer, reference ranges being 2.23–5.36 and 9.5–25 pmol/l respectively. TPO Abs were determined by RIA (reference range <60 IU/ml) and TSH-R Abs by the second-generation TRAK assay (reference range <2 U/l; BRAHMS, Berlin, Germany). Quality management included standard laboratory evaluation procedures and regular participation in inter-laboratory tests.

Calculations, models and statistical methods

A physiologically based mathematical model (MiMe-NoCoDI model) of thyroid homoeostasis has been previously developed and validated in healthy volunteers and cohorts of patients with defined thyroid disorders, including hypothyroid and hyperthyroid conditions, by one of the authors (15, 16). This model has also been made available as an open-source software (SimThyr 3.2 for Mac OS X, Windows and Linux) at sourceforge.net. It allows mapping of the physiological and biochemical interplay to the various parameters involved. With this simulative approach, sensitivity analysis was performed to study the relationship between certain structural parameters such as type 2 deiodinase (D2) and thyroid parameters (17). The mathematical equation used for empirical curve fitting was as follows: TSH=s/(1+LS×FT4), with s referring to maximum secretory capacity and LS to the brake constant of non-competitive inhibition at the pituitary site. Theoretical sum activity of peripheral deiodinases (GD) was calculated from the equation:

article image

from equilibrium levels for FT4 and FT3 and from constant parameters for kinetics, plasma proteins and dissociation constants as previously defined and described (17).

Basic descriptive statistics, correlations and linear models were performed using the statistical software packages R 2.15.0 for Mac and Deducer 0.7 with JGR 1.7–10 (18, 19). Kruskal–Wallis rank sum test was used to compare values among groups and analysis of covariance (ANCOVA) was used for comparison of regression lines. For comparison of models, F-statistics and Akaike criteria were applied. In the setting of multiple testing, P values were corrected with the Benjamini–Hochberg method. P values <0.05 were considered significant.

Results

The 1994 patients studied were grouped together according to their treatment. Group 1 comprised untreated subjects, with groups 2–4 subjects on 50–75, 100–125 and 150–200 μg/day l-T4 respectively. Group 5 comprised patients on iodine supplementation. Table 1 provides an overview of treatment-related functional states in the five groups. Thyroid parameters differed significantly among groups in the euthyroid state, not in hypothyroidism, and partly in the other conditions shown (Table 1).

Table 1

Treatment groups and thyroid function states.

Thyroid statusParameterGroup 1 no Txn=1159Group 2 l-T450–75 μg/day n=254Group 3 l-T4100–125 μg/day n=343Group 4 l-T4150–200 μg/day n=188Group 5 iodinen=50
Hypothyroidn4115710
FT3 (pmol/l)2.252.412.093.67
1.49–3.291.65–2.911.85–2.32
FT4 (pmol/l)5.917.917.649.04
4.45–8.257.01–8.846.55–7.93
GD (nmol/s)35.928.224.437.5
27.6–45.024.0–35.520.9–34.7
TSH (mU/l)18.628.315.322.8
7.24–43.09.35–55.487.28–38.55
Subclinically hypothyroidn54282960
FT3 (pmol/l)3.342.783.192.89
2.65–3.832.30–3.362.12–3.822.79–3.60
FT4* (pmol/l)10.913.113.914.9
10.4–12.311.9–14.112.1–15.513.8–16.0
GD* (nmol/s)27.519.218.916.6
19.8–32.417.5–21.813.8–25.516.4–18.7
TSH (mU/l)5.627.245.875.32
4.88–7.535.11–15.35.32–9.544.49–18.10
Euthyroidn7311541533142
FT3* (pmol/l)3.643.473.293.653.81
3.13–4.232.97–3.942.82–3.773.13–4.013.21–4.18
FT4* (pmol/l)13.014.116.1417.013.3
11.8–14.612.9–15.814.5–18.715.4–19.113.0–14.4
GD* (nmol/s)25.821.418.019.726.0
21.8–30.118.6–26.314.6–22.017.1–22.022.8–29.0
TSH* (mU/l)1.071.211.090.921.01
0.64–1.640.88–2.000.75–1.910.59–1.870.86–1.59
Subclinically hyperthyroidn314551351148
FT3* (pmol/l)3.773.563.934.074.13
3.03–4.412.99–4.213.25–4.433.52–4.533.63–4.66
FT4* (pmol/l)14.715.419.421.514.5
12.9–16.513.4–17.517.0–21.619.7–23.512.6–16.7
GD* (nmol/s)23.521.718.317.424.2
19.1–27.717.7–24.216.2–21.315.5–19.521.7–31.9
TSH (mU/l)0.210.210.050.010.27
0.07–0.310.08–0.300.01–0.190.01–0.060.20–0.37
Hyperthyroidn19219360
FT3* (pmol/l)7.684.494.714.77
4.90–12.04.28–4.713.93–5.204.31–5.54
FT4 (pmol/l)31.834.526.528.4
26.1–38.430.8–38.225.9–28.326.5–29.8
GD* (nmol/s)22.512.916.315.5
17.6–30.310.9–14.813.1–17.813.5–17.6
TSH (mU/l)0.010.010.010.01
0.01–0.180.01–0.010.01–0.010.01–0.04

GD, sum activity of peripheral deiodinases. Values indicate median and interquartile range (IQR). An asterisk indicates significant differences (P<0.05) of the parameter among the groups using Kruskal–Wallis rank sum test and P value correction for multiple testing according to Benjamini–Hochberg method.

The relationships of log TSH serum levels and peripheral hormone concentrations were analysed over a comparable functional range from the euthyroid to mildly hyperthyroid spectrum defined by FT4 ≥9.5 pmol/l and TSH >0.01 and <100 mU/l (Fig. 1). The correlations revealed marked differences in the relationships among the groups, particularly between patient groups untreated and on l-T4. Gradients of the log TSH–FT3 regressions were relatively flat in untreated patients and those taking iodine, but much steeper in l-T4 treatment (Fig. 1). The log TSH–FT4 relationship also displayed significantly higher gradients in treatment groups 3 and 4, but compared with log TSH vs FT3, changes were less pronounced (Fig. 1). GD was positively correlated with log TSH only in untreated subjects (Fig. 1).

Figure 1
Figure 1

Correlations of log TSH vs FT4, log TSH vs FT3 and GD vs log TSH in various treatment groups. Group 1: untreated subjects, Groups 2–4: l-T4-treated patients with doses of 50–75 μg/day (Group 2), 100–125 μg/day (Group 3) or 150–200 μg/day (Group 4), Group 5: patients taking iodine supplementation. P values given were corrected for multiple testing by the Benjamini–Hochberg method. Equations of regression lines (±s.e.m.) depicted are as follows: left panels, Group 1: log TSH=−0.069 (±0.005) * FT4+0.80 (±0.07), r=−0.39, P<0.0001, n=1061. Group 2: log TSH=−0.058 (±0.015) * FT4 +0.86 (±0.21), r=−0.25, P<0.0001, n=234. Group 3: log TSH=−0.11 (±0.011) * FT4 +1.64 (±0.19), r=−0.52, P<0.0001, n=284. Group 4: log TSH=−0.095 (±0.016) * FT4 +1.23 (±0.32), r=−0.51, P<0.0001, n=109. Group 5: log TSH=−0.06 (±0.025) * FT4 +0.75 (±0.35), r=−0.32, P=0.02, n=50. Middle panels, Group 1: log TSH=−0.019 (±0.016) * FT3 −0.066 (±0.06), r=−0.04, P=0.27, n=1061. Group 2: log TSH=−0.18±0.040 FT3 +0.65±0.14, r=−0.29, P<0.0003, n=234. Group 3: log TSH=−0.32 (±0.044) * FT3 +0.87 (±0.16), r=−0.40, P<0.0003, n=284. Group 4: log TSH=−0.17 (±0.086) * FT3 −0.05 (±0.33), r=−0.19, P=0.08, n=109. Group 5: log TSH=−0.05±0.052 FT3 +0.12±0.20, r=−0.13, P=0.37, n=50. Right panels, Group 1: GD=3.2 (±0.44) * log TSH +25.60 (±0.22), r=0.22, P=0.0005, n=1061. Group 2: GD=−1.5 (±0.70) * log TSH +21.86 (±0.40), r=−0.14, P=0.06, n=234. Group 3: GD=−0.10 (±0.46) * log TSH +18.90 (±0.36), r=0.01, P=0.83, n=284. Group 4: GD=1.74 (±0.54) * log TSH +18.98 (±0.56), r=0.30, P=0.004, n=109. Group 5: GD=0.69 (±2.68) * log TSH +26.21 (±0.91), r=0.04, P=0.83, n=50. Slopes were significantly different compared to the untreated Group 1 (ANCOVA), for FT4 in Group 3 (P<0.001) and Group 4 (P<0.03), for FT3 in Group 2 (P<0.001), Group 3 (P<0.001) and Group 4 (P<0.01), for GD in Group 2 (P<0.001), Group 3 (P<0.03) and Group 4 (P<0.001).

Citation: European Journal of Endocrinology 168, 2; 10.1530/EJE-12-0819

Download Figure

Group differences in TSH, FT3, FT4 and GD were further assessed in defined bands of TSH concentrations (Fig. 2). Over a wide functional range, medians and 95% confidence interval (CIs) for FT4 and partly FT3 differed significantly from untreated or iodine-treated subjects, as did GD, because FT3 and FT4 trends were moving in opposite directions (Fig. 2). In the euthyroid TSH range, represented by TSH bands 4 and 5, medians of FT3 for patients receiving moderate l-T4 doses (Groups 2 and 3) declined and the lower 95% CI limits were placed outside the reference range (Fig. 2). Only high l-T4 doses (Group 4), frequently accompanied by suppressed TSH, placed FT3 values within its reference limits. Patients regarded euthyroid by TSH criteria (TSH bands 4 and 5) would sometimes be classified differently by FT3 and considered hypothyroid in 13/164 (8%) of the cases in Group 2 on 50–75 μg/day l-T4, 18/154 (12%) in Group 3 on 100–125 μg/day l-T4 and 0/33 (0%) in Group 4 on 150–200 μg/day l-T4.

Figure 2
Figure 2

Comparison of medians and 95% confidence intervals (95% CI) of thyroid parameters in various treatment groups and functional bands defined by TSH values. The TSH bands on the x-axis were defined as follows: Band 1: TSH >50 mU/l, Band 2: TSH 10–50 mU/l, Band 3: TSH 4–10 mU/l, Band 4: TSH 2–4 mU/l, Band 5: TSH 0.4–2 mU/l, Band 6: TSH 0.1–0.4 mU/l and Band 7: TSH <0.1 mU/l. The error bars (95% CI) depicted in each TSH band refer to Groups 1–5 respectively. Group 1: untreated subjects, Group 2: l-T4-treated patients with doses of 50–75 μg/day, Group 3: l-T4-treated patients with doses of 100–125 μg/day, Group 4: l-T4-treated patients with doses of 150–200 μg/day and Group 5: patients taking iodine supplementation. An asterisk indicates the presence of a significant difference (P<0.05) among the groups, as established by Kruskal–Wallis test and Benjamini–Hochberg correction for multiple testing. Full colour version of this figure available via http://dx.doi.org/10.1530/EJE-12-0819.

Citation: European Journal of Endocrinology 168, 2; 10.1530/EJE-12-0819

Download Figure

Theoretical mathematical modelling was performed independently from the analysis of the clinical sample. We have initially assessed the MiMe-NoCoDI model by fitting it to our data sample. This may best be done in an open-loop situation, i.e. in overtly hypothyroid untreated patients, as shown in Fig. 3. For comparison, fitting with the conventional log TSH FT4 model is also shown in Fig. 3. As the MiMe-NoCoDI model is based on TSH values rather than their log transformed derivatives, which are commonly applied and used in the latter model, the two models were depicted in terms of TSH vs FT4 (Fig. 3). When statistically comparing the compatibility of the two models with the data (F-statistics and Akaike criteria), their goodness of fit did not significantly alter (Fig. 3).

Figure 3
Figure 3

Fit of models to observed data in the subgroup of untreated overtly hypothyroid patients: i) MiMe-NoCoDI model (solid line), ii) log linear standard model (dashed line). The fitting curves to the data points (n=50) were described by the following equations and parameters (±s.e.m.): a) TSH=88 (±22)/(1+0.35 (±0.16) * FT4) and b) log TSH=−0.12 (±0.02) * FT4 +2 (±0.12). For reasons of comparability to a, b is also shown in a non-logarithmic representation of TSH vs FT4. Comparisons by F-statistics (sum of residuals 31312 vs 30934, P=NS) and Akaike criteria (AIC 469 vs 466, P=NS) revealed no significant differences in the goodness of the fits of the two models.

Citation: European Journal of Endocrinology 168, 2; 10.1530/EJE-12-0819

Download Figure

When applying the MiMe-NoCoDI model, sensitivity analysis revealed that pituitary TSH secretion, in addition to responding to circulating FT3 and FT4 concentrations, is subject to the influence of a variety of structural parameters (Fig. 4, left panel). This importantly includes pituitary D2 (Fig. 4, left panel). In this model, the divergence of central T3 and peripheral T3 production becomes evident. Increased central T3 by D2 activation suppresses TSH and indirectly (via TSH) reduces FT4 and FT3 production (Fig. 4, right panel).

Figure 4
Figure 4

Mathematical model of sensitivity of TSH, central T3, FT3 and FT4 to changes in structural parameters. According to sensitivity analysis (see Materials and methods section), shown in the left panel, a 20% variation in various structural parameters' changes, in the absence of any change in circulating FT3 or FT4 concentrations, pituitary TSH secretion by about 5 to 20%. Type 2 deiodinase (D2) represented by GD2 and KM2 is among the more influential parameters in this respect. In the right panel, the relationships between GD2 activity and central T3 and circulating FT3 and FT4 concentrations are shown. Central T3 rises with increasing GD2 activity, thereby suppressing TSH and, as a consequence, circulating FT3 and FT4 levels decline. GD1, maximum activity of type I deiodinase; KM1, dissociation constant of 5′-deiodinase I; GD2, maximum activity of type II deiodinase; KM2, dissociation constant of 5′-deiodinase II; GT, secretory capacity of thyroid gland; DT, damping constant (EC50) of TSH at the thyroid gland; GH, secretory capacity of the pituitary; DH, damping constant (EC50) of TRH at the pituitary; SS, brake constant of TSH ultrashort feedback; DS, EC50 for TSH at the pituitary; GR, maximum gain of TRβ receptors; DR, EC50 for central T3, betaS, clearance exponent for TSH; betaS2, clearance exponent for central TSH; betaT, clearance exponent for T4; beta31, clearance exponent for T3P; beta32, clearance exponent for central T3; LS, brake constant (adopted from Dietrich et al. (15, 17)). Full colour version of this figure available via http://dx.doi.org/10.1530/EJE-12-0819.

Citation: European Journal of Endocrinology 168, 2; 10.1530/EJE-12-0819

Download Figure

Discussion

Originating from a basic understanding of pituitary thyroid feedback, the perspective of the diagnostic role of TSH has shifted in importance and the parameter has been promoted into a separate statistical measure in its own right (5, 6, 7, 8, 9, 10, 11, 12, 20). Limitations in this approach include analytical issues, genetic variability, uncertainty about the reference range, high ratio of the basic intra-individual to inter-individual variability and considerable natural intra-individual fluctuations (10, 11, 12, 13, 21, 22, 23, 24, 25, 26, 27, 28, 29). A lack of correlation with clinical symptom scores occurs particularly in patients with subclinical hypothyroidism (30). However, the implicit assumption of a fixed relationship between peripheral thyroid hormone homoeostasis and the pituitary set point in health and disease has not been seriously questioned. In this study, we show that as a result of the attainment of a hypothyroid state, there is an imbalance and disjoint involving peripheral thyroid hormone homoeostasis and the pituitary set point that persists and is particularly evident under T4 monotherapy. The treatment-related disjoint between peripheral thyroid hormone status and pituitary response became apparent by retrospective correlative analysis of a large clinical sample and was further corroborated by fundamental modelling and independent simulation.

In untreated euthyroid subjects, the gradient of the correlation line relating FT3 and log TSH is flat, but in subjects treated by T4 monotherapy, the relationship is different in all cases, the gradient increasing with hormone dosage. This suggests that the balanced relationship between the thyroid hormones themselves and TSH, evident in untreated euthyroids, no longer applies and shows that in monotherapy added increments of T4 are progressively more effective in suppressing TSH production, but simultaneously less effective in restoring FT3. Obviously, this is a consequence of the decreasing ability of thyroid homoeostasis to maintain the normal interrelationships between T3, T4 and TSH.

Hence, the results indicate that the observed disjoint between the thyroid/pituitary FT4–TSH feedback mechanism and T3 production noted in hypothyroidism (whether inadequately or untreated) is not fully restored even when sufficient T4 is given to regain an apparently euthyroid state. In several cases, this results in a classification mismatch, for example placing ∼10% of patients on moderate T4 doses that are judged euthyroid according to their TSH measurements below the FT3 reference range. Higher l-T4 doses maintained FT3 within its reference limits, though frequently only in conjunction with suppressed TSH. Gullo et al. (31) have recently reported the same phenomenon in a large homogeneous group of athyreotic patients, whereas Jonklaas et al. (32) reported that total T3 was not reduced in such patients, but did not study FT3. In contrast, Ito et al. (33) showed that in patients who have undergone thyroidectomy, a moderately TSH-suppressive dose of l-T4 is required to postoperatively restore serum FT3 to the preoperative level. Interestingly, earlier studies have also indicated that l-T4 replacement aiming at a normal serum TSH may not invariably result in an appropriate normalisation of serum FT3 (34).

Additionally, we observed a positive correlation between TSH and GD in untreated patients. This could possibly stem from the stimulation by TSH of both type 1 (D1) and type 2 (D2) deiodinases mediated by a signalling process using cAMP (35, 36, 37, 38, 39). Interestingly, the correlation was strongest in untreated subjects and impaired in T4-treated patients, suggesting a role of TSH-modulated deiodination within the thyroid gland itself.

The pituitary responds primarily to changes in circulating FT4 by the action of type 2 deiodinase (D2) on the local production of T3 (central T3) from T4 (2). Circulating FT3 derived mainly from peripheral deiodinase activities has less influence (2). Remarkably, the pituitary retains its ability to respond to increasing FT4 levels in T4-treated subjects, permitting appropriate TSH feedback responses. This may be achieved by a mechanism unique to thyrotrophs, which in contrast to other cells expressing D2, are capable of increasing D2 expression to maintain high T3 production at high T4 concentrations (2). By sensitivity analysis, we have shown that central T3 levels at the pituitary level, as opposed to T3 in the periphery, are up-regulated by the activation of D2. This, in turn, results in suppression of TSH concentrations and, as a further consequence, in a decrease in circulating FT4 and FT3 concentrations. The divergence in the central response from the peripheral equilibrium may explain the observed phenomenon that FT3 serum levels remain disproportionally low even when a supposedly normal TSH has been achieved by means of exogenous l-T4 administration. This means that the thyroid status and dose adequacy of T4 substitution cannot readily be defined by the putative gold standard TSH in these conditions because TSH concentrations within the accepted reference range may not invariably signal optimum concentrations of the peripherally active hormone FT3.

The clinical findings prompted us to seek independent confirmation on theoretical grounds. To this end, we applied a physiologically based mathematical model of thyroid homoeostasis, which had previously been developed by one of the authors (15, 16). Apart from the fact that this model permits simulation of the influence of various structural parameters and sensitivity analysis, which the standard log linear model does not, another advantage is that it does not make a preconceived assumption that the TSH–FT4 relationship is log linear but relies on Michaelis Menten kinetics and non-competitive divisive inhibition, thus reflecting the physiological and biochemical interplay of the various parameters involved (15). The model has previously been validated in a clinical setting (16). It was briefly re-evaluated in the present data sample and provided a reasonably good fit to the data, as did the log linear model. Proof of superiority of one or the other model is a more complex task and beyond the scope of this article, but the question is currently explored in an ongoing prospective trial (NOMOTHETICOS study), the protocol of which has been published (ClinicalTrials.gov NCT01145040, German Clinical Trials Register DRKS00003153, UTN U1111-1122-3273). Sensitivity analysis revealed a dependency of TSH levels on multiple structural parameters of the feedback control system, including the sum activity of pituitary type 2 deiodinase (GD2), which in turn is modulated by a whole variety of factors (2). TSH-stimulated deiodination, on the other hand, seems to be largely restricted to patients with a functional thyroid gland. Finally, our observation could explain the existence of a subgroup of patients with normal TSH but low FT3 levels.

Although our study may be affected by heterogeneity and its retrospective design, we have been careful to exclude those patients with known problems of assay interpretation. No patient was considered to have a systemic illness of a magnitude likely to initiate a significant ‘low T3 syndrome’ typical of this condition, though such possibilities must always be envisaged when using FT3 as a diagnostic tool. Although the study group was heterogeneous, relevant comparisons were made in defined TSH segments. While TSH disequilibrium may have been present in some subjects, it does not account for a low FT3/‘normal’ TSH pair because peripheral hormone levels are normalised before TSH in the treatment of hypothyroidism. Apart from acknowledged limitations of a retrospective study design and assay techniques, importantly, there is a second line of independent supportive evidence. Mathematical modelling was found to be in good accord with our empirical findings, supporting a significant disparity between the central FT3/TSH and the peripheral T3 under l-T4.

The described disjoint between FT4–TSH feedback and T3 production implies that T4 treatment constitutes a novel compensatory adaptive state of its own and therefore cannot be considered a complete re-enactment and restoration of the previous euthyroid state. This poses the question as to the possible long-term consequences of this disparity and whether thyroid hormone replacement, as it is currently performed, is truly optimal in all cases. Combined T3/T4 therapy has generally not been found superior in many short-term trials including a meta-analysis (40, 41). However, it is possible that certain subjects, e.g. those with limited endogenous thyroid reserve and consecutive hypodeiodination, may benefit, and most studies have used more heterogeneous populations (40, 41). Genetic variation, in particular D2 polymorphism, could also play a role (25, 42). On the other hand, a TSH value within the reference limits does not guarantee the absence of an adverse clinical outcome (43). This question remains unanswered, but our study encourages the need for further examination of adjunctive T3 therapy in selected patients, as suggested by Biondi & Wartofsky (40) in a recent review. Further, preferably prospective, studies are required to corroborate and extend these findings: i) to relate the described ‘disjoint’ to a host of clinically important parameters (e.g. metabolic rate, thermogenesis, lipid profiles and psychological well-being); ii) to elaborate on the possible influence of genomic factors and other influences (such as age, sex, BMI, thyroid volume, disease etiology, iodine intake, selenium status); and iii) to address the remaining methodological concerns by applying equilibrium dialysis or mass spectrometry for hormone measurement (29). Theoretical modelling, as demonstrated here, may prove helpful in guiding the design of adequate clinical trials. There are a few randomised controlled trials or cross-sectional studies that have addressed some of these aspects. Appelhof et al. (44) in a double-blind randomised trial on combined T3/T4 therapy made the point that satisfaction with study medication was not correlated with either absolute TSH values or with the change in TSH. Saravanan et al. (45) showed that strong correlations between FT4 levels and well-being persisted even when TSH values were within the reference range. To complicate things further, a recent study by Wang et al. (46) suggests that not only may thyroid hormones act independently of TSH but TSH within its reference limits may also exert more direct thyroid hormone-independent effects, for instance on lipid profiles.

In conclusion, our data suggest that, in order to fully understand the implications of T4 monotherapy in the hypothyroid patient, the normalisation of T3 production for maintenance of the metabolic processes should be considered to be equally as relevant as the control of the mechanisms of T4–TSH feedback that currently dominate functional diagnostic procedures. The two cannot be regarded as equivalent owing to their treatment-related disparity of response. This challenges the role of TSH as an exclusive standard in assessing dose adequacy in thyroid hormone replacement and invites further study.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding

This research did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.

Acknowledgements

The authors wish to thank Walter Eckl for technical assistance with the data set.

References

 

Official journal of

European Society of Endocrinology

Sections

Figures

  • Correlations of log TSH vs FT4, log TSH vs FT3 and GD vs log TSH in various treatment groups. Group 1: untreated subjects, Groups 2–4: l-T4-treated patients with doses of 50–75 μg/day (Group 2), 100–125 μg/day (Group 3) or 150–200 μg/day (Group 4), Group 5: patients taking iodine supplementation. P values given were corrected for multiple testing by the Benjamini–Hochberg method. Equations of regression lines (±s.e.m.) depicted are as follows: left panels, Group 1: log TSH=−0.069 (±0.005) * FT4+0.80 (±0.07), r=−0.39, P<0.0001, n=1061. Group 2: log TSH=−0.058 (±0.015) * FT4 +0.86 (±0.21), r=−0.25, P<0.0001, n=234. Group 3: log TSH=−0.11 (±0.011) * FT4 +1.64 (±0.19), r=−0.52, P<0.0001, n=284. Group 4: log TSH=−0.095 (±0.016) * FT4 +1.23 (±0.32), r=−0.51, P<0.0001, n=109. Group 5: log TSH=−0.06 (±0.025) * FT4 +0.75 (±0.35), r=−0.32, P=0.02, n=50. Middle panels, Group 1: log TSH=−0.019 (±0.016) * FT3 −0.066 (±0.06), r=−0.04, P=0.27, n=1061. Group 2: log TSH=−0.18±0.040 FT3 +0.65±0.14, r=−0.29, P<0.0003, n=234. Group 3: log TSH=−0.32 (±0.044) * FT3 +0.87 (±0.16), r=−0.40, P<0.0003, n=284. Group 4: log TSH=−0.17 (±0.086) * FT3 −0.05 (±0.33), r=−0.19, P=0.08, n=109. Group 5: log TSH=−0.05±0.052 FT3 +0.12±0.20, r=−0.13, P=0.37, n=50. Right panels, Group 1: GD=3.2 (±0.44) * log TSH +25.60 (±0.22), r=0.22, P=0.0005, n=1061. Group 2: GD=−1.5 (±0.70) * log TSH +21.86 (±0.40), r=−0.14, P=0.06, n=234. Group 3: GD=−0.10 (±0.46) * log TSH +18.90 (±0.36), r=0.01, P=0.83, n=284. Group 4: GD=1.74 (±0.54) * log TSH +18.98 (±0.56), r=0.30, P=0.004, n=109. Group 5: GD=0.69 (±2.68) * log TSH +26.21 (±0.91), r=0.04, P=0.83, n=50. Slopes were significantly different compared to the untreated Group 1 (ANCOVA), for FT4 in Group 3 (P<0.001) and Group 4 (P<0.03), for FT3 in Group 2 (P<0.001), Group 3 (P<0.001) and Group 4 (P<0.01), for GD in Group 2 (P<0.001), Group 3 (P<0.03) and Group 4 (P<0.001).

    View in gallery
  • Comparison of medians and 95% confidence intervals (95% CI) of thyroid parameters in various treatment groups and functional bands defined by TSH values. The TSH bands on the x-axis were defined as follows: Band 1: TSH >50 mU/l, Band 2: TSH 10–50 mU/l, Band 3: TSH 4–10 mU/l, Band 4: TSH 2–4 mU/l, Band 5: TSH 0.4–2 mU/l, Band 6: TSH 0.1–0.4 mU/l and Band 7: TSH <0.1 mU/l. The error bars (95% CI) depicted in each TSH band refer to Groups 1–5 respectively. Group 1: untreated subjects, Group 2: l-T4-treated patients with doses of 50–75 μg/day, Group 3: l-T4-treated patients with doses of 100–125 μg/day, Group 4: l-T4-treated patients with doses of 150–200 μg/day and Group 5: patients taking iodine supplementation. An asterisk indicates the presence of a significant difference (P<0.05) among the groups, as established by Kruskal–Wallis test and Benjamini–Hochberg correction for multiple testing. Full colour version of this figure available via http://dx.doi.org/10.1530/EJE-12-0819.

    View in gallery
  • Fit of models to observed data in the subgroup of untreated overtly hypothyroid patients: i) MiMe-NoCoDI model (solid line), ii) log linear standard model (dashed line). The fitting curves to the data points (n=50) were described by the following equations and parameters (±s.e.m.): a) TSH=88 (±22)/(1+0.35 (±0.16) * FT4) and b) log TSH=−0.12 (±0.02) * FT4 +2 (±0.12). For reasons of comparability to a, b is also shown in a non-logarithmic representation of TSH vs FT4. Comparisons by F-statistics (sum of residuals 31312 vs 30934, P=NS) and Akaike criteria (AIC 469 vs 466, P=NS) revealed no significant differences in the goodness of the fits of the two models.

    View in gallery
  • Mathematical model of sensitivity of TSH, central T3, FT3 and FT4 to changes in structural parameters. According to sensitivity analysis (see Materials and methods section), shown in the left panel, a 20% variation in various structural parameters' changes, in the absence of any change in circulating FT3 or FT4 concentrations, pituitary TSH secretion by about 5 to 20%. Type 2 deiodinase (D2) represented by GD2 and KM2 is among the more influential parameters in this respect. In the right panel, the relationships between GD2 activity and central T3 and circulating FT3 and FT4 concentrations are shown. Central T3 rises with increasing GD2 activity, thereby suppressing TSH and, as a consequence, circulating FT3 and FT4 levels decline. GD1, maximum activity of type I deiodinase; KM1, dissociation constant of 5′-deiodinase I; GD2, maximum activity of type II deiodinase; KM2, dissociation constant of 5′-deiodinase II; GT, secretory capacity of thyroid gland; DT, damping constant (EC50) of TSH at the thyroid gland; GH, secretory capacity of the pituitary; DH, damping constant (EC50) of TRH at the pituitary; SS, brake constant of TSH ultrashort feedback; DS, EC50 for TSH at the pituitary; GR, maximum gain of TRβ receptors; DR, EC50 for central T3, betaS, clearance exponent for TSH; betaS2, clearance exponent for central TSH; betaT, clearance exponent for T4; beta31, clearance exponent for T3P; beta32, clearance exponent for central T3; LS, brake constant (adopted from Dietrich et al. (15, 17)). Full colour version of this figure available via http://dx.doi.org/10.1530/EJE-12-0819.

    View in gallery

References

1

Braverman LE & Utiger RD. Introduction to thyrotoxicosis. In The Werner and Ingbar's Thyroid, 6th edn, ch 30, pp 645–647. Eds LE Braverman & RD Utiger. Philadelphia: J.B. Lippincott Company, 1995

2

GerebenBZavackiAMRibichSKimBWHuangSASimonidesWSZeoldABiancoAC. Cellular and molecular basis of deiodinase-regulated thyroid hormone signaling. Endocrine Reviews200829898938. (doi:10.1210/er.2008-0019).

3

ArrojoEDrigoRBiancoAC. Type 2 deiodinase at the crossroads of thyroid hormone action. International Journal of Biochemistry & Cell Biology20114314321441. (doi:10.1016/j.biocel.2011.05.016).

4

PiloAIervasiGVitekFFerdeghiniMCazzuolaFBianchiR. Thyroidal and peripheral production of 3,5,3′ -triiodothyronine in humans by multicompartmental analysis. American Journal of Physiology1990258E715E726.

5

DemersLMSpencerCA. Thyrotropin/thyroid stimulating hormone (TSH) measurement. Thyroid2003133344. (doi:10.1089/105072503321087015).

6

FlynnRWVMacdonaldTMJungRTMorrisADLeeseGP. Mortality and vascular outcomes in patients treated for thyroid dysfunction. Journal of Clinical Endocrinology and Metabolism20069121592164. (doi:10.1210/jc.2005-1833).

7

VölzkeHSchwahnCWallaschofskiHDörrM. Review: the association of thyroid dysfunction with all-cause and circulatory mortality: is there a causal relationship?Journal of Clinical Endocrinology and Metabolism20079224212429. (doi:10.1210/jc.2007-0179).

8

WaringACHarrisonSSamuelsMHEnsrudKELe BlancESHoffmanAROrwollEFinkHABarrett-ConnorEBauerDC. Thyroid function and mortality in older men: a prospective study. Journal of Clinical Endocrinology and Metabolism201297862870. (doi:10.1210/jc.2011-2684).

9

WeissIABloomgardenNFrishmanWH. Subclinical hypothyroidism and cardiovascular risk: recommendations for treatment. Cardiology in Review201119291299. (doi:10.1097/CRD.0b013e318227df87).

10

DickeyRAWartofskyLFeldS. Optimal thyrotropin level: normal ranges and reference intervals are not equivalent. Thyroid20051510351039. (doi:10.1089/thy.2005.15.1035).

11

LaurbergPAndersenSCarléAKarmisholtJKnudsenNPedersenIB. The TSH upper reference limit: where are we at?Nature Reviews. Endocrinology20117232239. (doi:10.1038/nrendo.2011.13).

12

SurksMIGoswamiGDanielsGH. The thyrotropin reference range should remain unchanged. Journal of Clinical Endocrinology and Metabolism20059054895496. (doi:10.1210/jc.2005-0170).

13

AndersenSPedersenKMBruunNHLaurbergP. Narrow individual variations in serum T(4) and T(3) in normal subjects: a clue to the understanding of subclinical thyroid disease. Journal of Clinical Endocrinology and Metabolism20028710681072. (doi:10.1210/jc.87.3.1068).

14

HoermannREcklWHoermannCLarischR. Complex relationship between free thyroxine and TSH in the regulation of thyroid function. European Journal of Endocrinology201016211231129. (doi:10.1530/EJE-10-0106).

15

DietrichJWTescheAPickardtCRMitzdorfU. Thyrotropic feedback control: evidence for an additional ultrashort feedback loop from fractal analysis. Cybernetics and Systems200435315331. (doi:10.1080/01969720490443354).

16

Dietrich JW. Der Hypophysen-Schilddrsen-Regelkreis. Entwicklung und klinische Anwendung eines nichtlinearen Modells. In Spektrum Medizinische Forschung, pp 1–188. Ed F Schardt. Berlin: Logos Verlag, 2002

17

DietrichJWStachonAAnticBKleinHHHeringS. The AQUA-FONTIS study: protocol of a multidisciplinary, cross-sectional and prospective longitudinal study for developing standardized diagnostics and classification of non-thyroidal illness syndrome. BMC Endocrine Disorders2008813. (doi:10.1186/1472-6823-8-13).

18

R Development. Core Team R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing 2012 ISBN 3-900051-07-0. Available from http://cran.r-project.org.

19

Fellows I. Deducer: Deducer R package version 0.7-0. 2012. Available from http://cran.r-project.org.

20

LarsenPR. Thyroid–pituitary interaction: feedback regulation of thyrotropin secretion by thyroid hormones. New England Journal of Medicine19823062332. (doi:10.1056/NEJM198201073060107).

21

AndersenSBruunNHPedersenKMLaurbergP. Biologic variation is important for interpretation of thyroid function tests. Thyroid20031310691078. (doi:10.1089/105072503770867237).

22

AokiYBelinRMClicknerRJeffriesRPhillipsLMahaffeyKR. Serum TSH and total T4 in the United States population and their association with participant characteristics: National Health and Nutrition Examination Survey (NHANES 1999–2002). Thyroid20071712111223. (doi:10.1089/thy.2006.0235).

23

ColletT-HGusseklooJBauerDCElzen denWPJCappolaARBalmerPIervasiGAsvoldBOSgarbiJAVölzkeH. Subclinical hyperthyroidism and the risk of coronary heart disease and mortality. Archives of Internal Medicine2012172799809. (doi:10.1001/archinternmed.2012.402).

24

HamiltonTEDavisSOnstadLKopeckyKJ. Thyrotropin levels in a population with no clinical, autoantibody, or ultrasonographic evidence of thyroid disease: implications for the diagnosis of subclinical hypothyroidism. Journal of Clinical Endocrinology and Metabolism20089312241230. (doi:10.1210/jc.2006-2300).

25

HoftijzerHCHeemstraKAVisserTJle CessieSPeetersRPCorssmitEPMSmitJWA. The type 2 deiodinase ORFa-Gly3Asp polymorphism (rs12885300) influences the set point of the hypothalamus–pituitary–thyroid axis in patients treated for differentiated thyroid carcinoma. Journal of Clinical Endocrinology and Metabolism201196E1527E1533. (doi:10.1210/jc.2011-0235).

26

HollowellJGStaehlingNWFlandersWDHannonWHGunterEWSpencerCABravermanLE. Serum TSH, T(4), and thyroid antibodies in the United States population (1988 to 1994): National Health and Nutrition Examination Survey (NHANES III). Journal of Clinical Endocrinology and Metabolism200287489499. (doi:10.1210/jc.87.2.489).

27

OverRMannanSNsouli-MaktabiHBurmanKDJonklaasJ. Age and the thyrotropin response to hypothyroxinemia. Journal of Clinical Endocrinology and Metabolism20109536753683. (doi:10.1210/jc.2010-0281).

28

SpencerCAHollowellJGKazarosyanMBravermanLE. National Health and Nutrition Examination Survey III thyroid-stimulating hormone (TSH)-thyroperoxidase antibody relationships demonstrate that TSH upper reference limits may be skewed by occult thyroid dysfunction. Journal of Clinical Endocrinology and Metabolism20079242364240. (doi:10.1210/jc.2007-0287).

29

van DeventerHEMenduDRRemaleyATSoldinSJ. Inverse log-linear relationship between thyroid-stimulating hormone and free thyroxine measured by direct analog immunoassay and tandem mass spectrometry. Clinical Chemistry201157122127. (doi:10.1373/clinchem.2010.154088).

30

KarmisholtJAndersenSLaurbergP. Variation in thyroid function tests in patients with stable untreated subclinical hypothyroidism. Thyroid200818303308. (doi:10.1089/thy.2007.0241).

31

GulloDLatinaAFrascaFLe MoliRPellegritiGVigneriR. Levothyroxine monotherapy cannot guarantee euthyroidism in all athyreotic patients. PLoS ONE20116e22552. (doi:10.1371/journal.pone.0022552).

32

JonklaasJDavidsonBBhagatSSoldinSJ. Triiodothyronine levels in athyreotic individuals during levothyroxine therapy. Journal of the American Medical Association2008299769777. (doi:10.1001/jama.299.7.769).

33

ItoMMiyauchiAMoritaSKudoTNishiharaEKiharaMTakamuraYItoYKobayashiKMiyaA. TSH-suppressive doses of levothyroxine are required to achieve preoperative native serum triiodothyronine levels in patients who have undergone total thyroidectomy. European Journal of Endocrinology2012doi:10.1530/EJE-11-1029).

34

WoeberKA. Levothyroxine therapy and serum free thyroxine and free triiodothyronine concentrations. Journal of Endocrinological Investigation200225106109.

35

LarsenPRBerryMJ. Nutritional and hormonal regulation of thyroid hormone deiodinases. Annual Review of Nutrition199515323352. (doi:10.1146/annurev.nu.15.070195.001543).

36

BiancoACMaiaALda SilvaWSChristoffoleteMA. Adaptive activation of thyroid hormone and energy expenditure. Bioscience Reports200525191208. (doi:10.1007/s10540-005-2885-6).

37

MaiaALGoemannIMMeyerELSWajnerSM. Type 1 iodothyronine deiodinase in human physiology and disease: deiodinases: the balance of thyroid hormone. Journal of Endocrinology2011209283297. (doi:10.1530/JOE-10-0481).

38

MurakamiMKamiyaYMorimuraTArakiOImamuraMOgiwaraTMizumaHMoriM. Thyrotropin receptors in brown adipose tissue: thyrotropin stimulates type II iodothyronine deiodinase and uncoupling protein-1 in brown adipocytes. Endocrinology200114211951201. (doi:10.1210/en.142.3.1195).

39

NakaoNOnoHYamamuraTAnrakuTTakagiTHigashiKYasuoSKatouYKageyamaSUnoY. Thyrotrophin in the pars tuberalis triggers photoperiodic response. Nature2008452317322. (doi:10.1038/nature06738).

40

BiondiBWartofskyL. Combination treatment with T4 and T3: toward personalized replacement therapy in hypothyroidism?Journal of Clinical Endocrinology and Metabolism20129722562271. (doi:10.1210/jc.2011-3399).

41

Grozinsky-GlasbergSFraserANahshoniEWeizmanALeiboviciL. Thyroxine–triiodothyronine combination therapy versus thyroxine monotherapy for clinical hypothyroidism: meta-analysis of randomized controlled trials. Journal of Clinical Endocrinology and Metabolism20069125922599. (doi:10.1210/jc.2006-0448).

42

PanickerVSaravananPVaidyaBEvansJHattersleyATFraylingTMDayanCM. Common variation in the DIO2 gene predicts baseline psychological well-being and response to combination thyroxine plus triiodothyronine therapy in hypothyroid patients. Journal of Clinical Endocrinology and Metabolism20099416231629. (doi:10.1210/jc.2008-1301).

43

HeeringaJHoogendoornEHvan der DeureWMHofmanAPeetersRPHopWCJHeijer denMVisserTJWittemanJCM. High-normal thyroid function and risk of atrial fibrillation: the Rotterdam study. Archives of Internal Medicine200816822192224. (doi:10.1001/archinte.168.20.2219).

44

AppelhofBCFliersEWekkingEMScheneAHHuyserJTijssenJGPEndertEvan WeertHCPMWiersingaWM. Combined therapy with levothyroxine and liothyronine in two ratios, compared with levothyroxine monotherapy in primary hypothyroidism: a double-blind, randomized, controlled clinical trial. Journal of Clinical Endocrinology and Metabolism20059026662674. (doi:10.1210/jc.2004-2111).

45

SaravananPVisserTJDayanCM. Psychological well-being correlates with free thyroxine but not free 3,5,3′-triiodothyronine levels in patients on thyroid hormone replacement. Journal of Clinical Endocrinology and Metabolism20069133893393. (doi:10.1210/jc.2006-0414).

46

WangFTanYWangCZhangXZhaoYSongXZhangBGuanQXuJZhangJ. Thyroid-stimulating hormone levels within the reference range are associated with serum lipid profiles independent of thyroid hormones. Journal of Clinical Endocrinology and Metabolism20129727242731. (doi:10.1210/jc.2012-1133).

Information

Cited By

PubMed

Google Scholar

Related Articles

Altmetrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 182 182 182
PDF Downloads 35 35 35