I. INTRODUCTION
Although there usually is a good negative correlation between free T4 and TSH levels, there are some notorious exceptions in which the feedback system between fT4 and TSH seems to be disrupted. Most clinicians will be aware of the fact that TSH levels can remain low, despite clinical euthyroidism and normal concentrations of T4 and T3, in patients treated for Graves’ hyperthyroidism.¹ This situation is attributed to a delayed recovery of the pituitary-thyroid axis after a prolonged state of thyrotoxicosis.²
This explanation seemed to us unlikely for various reasons. First, TSH levels increase within weeks after discontinuation of T4 therapy in patients treated with TSH supressive doses of T4 for thyroid cancer, suggesting that the pituitary-thyroid axis can revive fast. Secondly, not all patients treated for Graves’ hyperthyroidism show this phenomenon of long-term TSH suppression. And thirdly, a decreased level of TSH after a one year course of antithyroid drugs is an independent risk factor for recurrence of hyperthyroidism upon discontinuation of antithyroid drugs.³ These clinical observations suggest that the long-term TSH suppression in Graves’ disease is a specific feature of a subset of Graves’ patients: those likely to relapse after antithyroid drug therapy, e.g. patients with persisting TSH Receptor Stimulating Immunoglobulins (TSI).
We hypothesized that TSI are responsible for the prolonged TSH suppression observed in a subset of Graves’ patients. The implication would be that TSI could decrease TSH secretion independently of thyroid status, at the central (hypothalamus/pituitary) level. Because the pituitary gland is outside the blood-brain barrier and the hypothalamus is not, we postulated that the TSI may act on the pituitary level decreasing TSH secretion. A requisite for this hypothesis is that the pituitary contains a TSH R.

II. PHYSIOLOGICAL RELEVANCE OF A PITUITARY TSH-RECEPTOR: A PITUITARY ULTRA-SHORT LOOP FEEDBACK.
The hypothesis of the existence of a pituitary TSH-Receptor implies that this receptor might have a function in human physiology, since it would be present in all subjects and not only in Graves’ patients. What would the physiological role of such a receptor be? If present, this TSH-Receptor might sense directly the pituitary TSH secretion. This is logical in the sense that such a receptor would enable fine-tuning of pituitary TSH secretion.
In physiological circumstances a decline in T4 production by the thyroid would be sensed in the pituitary, because of the negative feedback system between T4 concentrations and TSH production. Such a decline would be followed by an increase in TSH production leading to a stimulation of the thyroid gland to produce more thyroid hormones. However, there is a certain lag-time between the increase in TSH and a rise in plasma T4 and T3 levels. If during this lag-time TSH would remain elevated, an overshoot in thyroid hormone production will ensue. This would not occur, if the pituitary would be able to anticipate this effect of TSH on the thyroid gland, for instance by measuring its own TSH output. For this it would need a TSH-Receptor.
The hypothesis thus is that secreted TSH binds to an intrapituitary TSH-R, which upon activation then signals back (e.g. via a cytokine) to the thyrotroph to diminish its TSH secretion in an ultra-short loop feed-back system. Such a short-loop feedback is not limited to TSH secretion. Prolactin receptors,⁴ and GH Receptors5 have also been found in the pituitary, and evidence is accumulating that PRL and GH down-regulate their own secretion.^6,7

III. THE TSH-RECEPTOR IS PRESENT IN THE PITUITARY
We therefore embarked on a series of studies to demonstrate the presence of the TSH-R in human pituitaries. We first showed by RT-PCR the presence of mRNA encoding for the receptor in a human anterior pituitary library. This was confirmed by using in situ hybridization on human anterior pituitary slices. The presence of TSH-R protein was demonstrated using immunohistochemistry on human anterior pituitaries and using double labeling techniques we could show that the TSH-R was expressed by folliculo-stellate cells (Fig. 1).⁸

Fig.1

These findings were later confirmed by others, who also found that folliculo-stellate cells contain TSH-R protein.⁹
These folliculo-stellate cells make up 10% of the pituitary cell population. They are a kind of dendritic cells and thought to have a function in the signalling with the other pituitary cell populations.¹⁰ Now that we know that they express the TSH-R, their function becomes more clear in that they possibly form part of the ultra-short loop feed back on TSH secretion.
It is likely that they are also involved in a paracrine regulation of the secretion of other hypophyseal hormones, because we later showed that they also contain mRNA coding for the GH receptor and the ACTH receptor (manuscript submitted).

IV. CLINICAL RELEVANCE OF THE PITUITARY TSH-RECEPTOR
As mentioned, the pituitary lies outside the blood-brain barrier and this means that TSI may interact with this pituitary binding site. We postulated that the physiological ligand, TSH, would downregulate TSH secretion to some extent and hence TSI would do the same (Fig 2).


Fig. 2

To test this hypothesis, we treated rats with methimazole and thyroxine (mimicking the block and replacement therapy used in Graves patients) to switch off their thyroid glands, while maintaining euthyroidism. The rats were then infused with human TSI containing IgG’s or a control human IgG preparation, in some ways as in the old LATS assay. TSI infusion indeed resulted in a significant decrease in TSH concentrations compared to controls, without differences in T4 or T3 levels between both groups (Fig 3).¹¹


Fig. 3

This study thus showed that TSI is capable of reducing TSH secretion by rat pituitaries independently of thyroid hormone concentrations, indicative of a direct action of TSI on the pituitary TSH-R.
We then put our hypothesis to the proof in patients with Graves’ disease treated with block and replacement therapy. We followed a cohort of 45 patients who were rendered euthyroid by methimazole and thyroxine therapy. Three months after having achieved biochemical euthyroidism (defined as normal free T4 and total T3 levels), 22 patients still had detectable TSI (measured as TSH Binding Inhibiting Immunogolbulins, TBII assay) levels, whereas in 23 patients TBII levels had become negative. We then compared these two groups, which had similar free T4 and total T3 levels by that time, for their TSH values. The TSI positive group had significantly lower TSH values than the group who had become TSI negative (Fig 4).¹²


Fig. 4

In addition, we found that TSH levels in these treated Graves’ patients only correlated with TBII titers and not with free T4 or total T3 concentrations. These observations clearly support our notion, that TSI suppress TSH secretion via a direct central effect, most likely at the pituitary level when euthyroidism is restored.

V. CONCLUDING REMARKS
Long-term TSH suppression during otherwise successful treatment of Graves’ disease has always been attributed to a delayed recovery of the pituitary-thyroid axis. Less experienced clinicians regard it as proof for still existing “subclinical” hyperthyroidism and act accordingly by increasing the methimazole dosage or decreasing T4 substitution.

The above mentioned experiments have clearly shown that prolonged TSH suppression is very likely to be caused by an interaction between the pituitary TSH-R and circulating TSH-R autoantibodies, which can remain present in about half of treated Graves’ patients. Low TSH levels in clinically euthyroid patients with normal T4 and T3 levels thus do not indicate persisting low-grade hyperthyroidism, but should instead be seen as an indication for continued TSI activity.
A low TSH value in such patients may be regarded as a positive “bio-assay” for TSI activity and explain why decreased TSH values are an independent risk factor for a relapse of Graves’ hyperthyroidism after a course of antithyroid drugs.

 

REFERENCES