Main

It is now well established that TSH release is pulsatile in nature and has a circadian periodicity, characterized by a nocturnal surge that begins in the late afternoon and reaches a peak around midnight, remains relatively stable for several hours, and declines thereafter(1–7). Although neither the neuroendocrine mechanisms governing the pulsatile release of TSH nor its circadian rhythm in man is established, considerable evidence indicates that the dopaminergic system, timing of sleep, serum cortisol, and somatostatin tone inhibit TSH secretion(8–15). Abnormal basal serum TSH concentrations and altered response of TSH to TRH in patients suffering from GHD have been reported(16–19). Due to these abnormal responses, such patients have been diagnosed as having pituitary hypothyroidism (hyporesponsiveness) or hypothalamic hypothyroidism (delayed, prolonged, and/or exaggerated TSH response to TRH).

The increased basal TSH levels and exaggerated TSH response to TRH could result from decreased inhibition of TSH release due to a reduction in somatostatin and/or dopamine tone associated with growth hormone deficiency(19, 20). In experimental animals hypothalamic deafferentation abolishes the circadian TSH rhythm(21), explaining the absence of the nocturnal surge of TSH(22) in hypothalamic (central) hypothyroidism.

Recently Municchi et al.(23) assessed the nocturnal TSH surge in GHD children during the afternoon and night periods, but to our knowledge the analysis of a circadian and pulsatile secretion of serum TSH during the 24-h period in GHD has not been previously reported. The purpose of the current study was to characterize pulsatile and circadian TSH and PRL release, examine the nocturnal secretion pattern of TSH and PRL in GHD, and evaluate the relationships among thyroid hormone, GH, PRL, and TSH secretion in children with GH deficiency.

METHODS

Subjects. Twelve children were investigated and divided in three groups according to the 24-h serum GH, GH stimulation tests, and growth velocity. 1) Control subjects consisted of four healthy subjects(one girl and three boys) with constitutional delay(24)(24-h GH greater than 3 μg/L), without endocrinologic abnormalities. All but one (male subject with pubic hair development no greater than Tanner stage II) were considered to be in prepuberty development, and all boys' present plasma testosterone was less than 20 ng/mL. 2) Classical GHD including four prepuberal patients (one boy and three girls), who were diagnosed as GHD by at least two stimulation tests: insulin-induced hypoglycemia(25), clonidine(25), and L-dopa(26). Patients 1-4 had been previously treated with human GH, but therapy was interrupted at least 3 mo before our evaluation. Two patients (nos. 2 and 4) had received T4 because of clinical and laboratory diagnosis of central hypothyroidism during treatment with GH. In these patients thyroid hormone was stopped at least 60 d before this study. 3) NSD according to Spiliotis et al.(27) including four prepubertal patients (nos. 5-8) with short stature (one girl and three boys). Two of these NSD patients (nos. 6 and 7) did not have a delay in bone age, and patients 7 and 8's present IGF-I levels were in the normal range. The clinical and radiologic data for all groups of subjects are presented in Table 1. Bone age was assessed according to Greulich and Pyle(28) standards and pubertal development according to Tanner(29).

Table 1 Clinical characteristics of 4 GHD and 4 NSD children
Full size table

Experimental design. Written informed consent for the study from at least one parent of every patient (approved by the University's Institutional review board). The children were admitted to the Hospital on the day before testing to allow one night for accommodation to the setting and were encouraged to maintain their usual diet and ambulatory activity.

A blood sample was drawn at 0700 h for the baseline measurement of serum T4 and T3 and plasma IGF-I. Beginning at 0800 h, blood samples(1.5 mL) were obtained every 20 min for 24 h for GH, TSH, and PRL determination, through an indwelling venous catheter, with special effort made not to disturb sleep. The children were allowed to ambulate, and a record of their meals and sleep was kept. Serum was immediately separated after blood withdrawal and stored at -20°C until assay. On the following day, TRH (200μg; Biophysics Laboratory, Escola Paulista de Medicina, São Paulo, Brazil), regular insulin (0.1 IU/kg; Lilly, Indianapolis, IN) were administered by i.v. bolus. Blood samples were drawn -30 min and immediately before and 15, 30, 60, 90, and 120 min after the provocative stimulus for GH, TSH, and cortisol determinations. All of the tests were started at 0800 h, at least 10 h after the last meal.

Assays. All measurements were made in duplicate. TSH and GH concentrations were determined by IRMA methods (Pharmacia, Uppsala, Sweden), the detection limit being, respectively, 0.3 mIU/L and 0.1 μg/L for TSH and GH. Serum T3, T4, and cortisol were measured by kits provided by Clinical Assays Diagnostics (Boston, MA). The sensitivities for these methods were 0.3 nmoL/L (T3), 10.3 nmoL/L (T4), and 0.14 μg/dL(cortisol). PRL was measured with kits obtained from Diagnostics Products Corp. (Los Angeles, CA), with a sensitivity of 0.37 μg/L. For measurement of IGF-I, kits from Nichols Laboratories (San Juan Capristano, CA) were used with a sensitivity of 0.10 IU/mL. For all RIA methods the intrassay coefficients of variation were below 11% considering samples of low and high values.

Pulse and statistical analysis. To detect significant serum TSH and PRL hormone excursions (pulses) cluster analysis(30) was used. This program defines a pulse as a statistically significant increase in a cluster of hormone values, followed by a statistically significant decrease in a second cluster value. For TSH, the number of points for nadir and peak clusters, respectively, were two and two, with t statistics of 2.08 for the peak upstroke and 2.08 for peak downstroke. For PRL the number of points for nadir and peak clusters were two and one, respectively, with t statistics of 3.95 and 1.5 for the upstrokes and downstrokes, respectively. These parameters were chosen to constrain the false-positive pulse detection rate to less than 5%. The TSH and PRL data series were analyzed as the intact 24-h set and as two 12-h segments(0800-2000 and 2000-0800 h). If the pulse fell on a sampling time between these two segments, it was assigned to the preceding one.

Comparisons were made by Scheffe's test after analysis of variance or byt test proceeded by logarithmic transformation for nonnormal data. The normality of distribution was examined by using the Kolmogorov-Smirnov test. Statistical significance was assumed when p was less than 0.05. Results were expressed as the mean ± SE. Circadian variations in TSH and PRL were analyzed using the method of least squares to fit the mean hormone values to a cosine function(31). The mesor(value about which the oscillations occur), amplitude (half the difference between the highest and lowest values), and acrophase (timing of high point in clock hours) were studied in group control, GHD, and NSD groups.

RESULTS

The 24-h GH concentrations. The mean 24-h serum concentrations measured in the control, GHD, and NSD subjects are indicated inTable 2. As can be seen, the mean 24-h concentration(μg/L) in the control children (6.1 ± 0.8) was significantly higher than that documented in the GHD (0.9 ± 0.1; p = 0.0003) or NSD (2.1 ± 0.3; p = 0.0015) groups.

Table 2 Laboratory data of 4 GHD and 4 NSD children
Full size table

The 24-h TSH and PRL concentrations and pulse characteristics. The results of the characterization of pulsatile TSH and PRL release using cluster analysis are shown in Tables 3 and 4 and Figure 1 (A and B), respectively. With regard to TSH, the 24-h mean serum concentration observed in the GHD and NSD groups was higher than in the control subjects (p < 0.05). This elevation could not be explained by a change in TSH pulse frequency or duration, given that there were no differences between the control and GHD or NSD groups. However, data from the GHD and NSD subjects revealed increases in TSH pulse amplitude(p = 0.0004), pulse area (p = 0.0059), and interpulse valley mean (p = 0.001) when compared with the control subjects. Concerning PRL, the 24-h mean serum concentration observed in the GHD and NSD group was no different from that observed in the control subjects. Moreover, there were no differences in the frequency of PRL pulses or in any PRL pulse characteristic when the control and GHD and NSD subjects were compared.

Table 3 Characteristics of 24-h TSH secretion in 12 children (control, GHD, and NSD)
Full size table
Table 4 Characteristics of 24-h PRL secretion in 12 children (control, GHD, and NSD)
Full size table
Figure 1
figure 1

Twenty-four-hour TSH (A) and PRL(B) secretion pattern (sampling every 20 min). The closed squares(▪) indicate the mean serum TSH and PRL concentration of control(n = 4), the open squares (□) and the closed diamonds (♦) represent the mean serum TSH and PRL concentrations of GHD (n = 4) and NSD (n = 4), respectively.

Full size image

The 12-h TSH and PRL concentrations and pulse characteristics. When the data were analyzed according to 12-h daytime (0800-2000 h) and 12-h nighttime (2000-0800 h) segments, both control and NSD subjects demonstrated a rise in mean serum concentrations during the night but no change in pulse frequency. Data from control subjects revealed an increase at night in TSH pulse amplitude, pulse area, and interpulse valley, but not in pulse duration. Data from the NSD subjects demonstrated an increase at night in the pulse area, pulse duration, and amplitude mean; GHD subjects demonstrated a rise in the mean of interpulse valley. Analysis of the day versus night PRL data from control, GHD, and NSD subjects demonstrated a rise in mean serum concentration during the night but no day/night changes in pulse frequency or duration. The control, GHD, and NSD groups demonstrated nighttime increases in PRL pulse amplitude and pulse area mean.

TSH and PRL circadian rhythms. When the 24-h TSH and PRL data were fitted to a cosinor function, clear circadian TSH and PRL rhythms were disclosed in all groups (control, GHD, and NSD). The acrophases were similar with a range 23:37 to 02:23 h and 01:46 to 04:20 h to TSH and PRL, respectively.

DISCUSSION

Both the circadian rhythms and pulsatile patterns of TSH and PRL release were markedly consistent in control children and agree with previous studies that identified episodic release of TSH and PRL(1–8);i.e. hormonal concentrations exhibited a nocturnal rise beginning in the late afternoon or evening for TSH and PRL, respectively. The acrophase occurs around midnight for TSH and approximately 2 h later for PRL. The circadian TSH and PRL periodicities are composed of series of secretory pulses, with greater magnitudes during the nocturnal hours.

In all subjects studied, mean TSH and PRL concentrations were higher at night than during the day, consistent with the known nocturnal rise in TSH and PRL release(6, 32). Accordingly, increases in pulse amplitude and area means were observed in the nocturnal segment in control and NSD, with GHD subjects having a significantly interpulse valley mean at night. The greater 24-h TSH pulse amplitude, area, and interpulse valley mean found in GHD and NSD compared with those in the control children suggest that the increased TSH release of GHD and NSD is independent from changes in pulse frequency and width (duration).

It has been proposed that the nocturnal surge of TSH is mediated by increased hypothalamic secretion of TRH(3), given the significant correlation between the value of nocturnal peaks and TRH-induced TSH peaks(7). As a group, these data suggest that patients with hypothalamic GHD, responsive to exogenous GH-releasing hormone, may have TRH pulses of greater magnitude than in normal subjects.

Our results demonstrate that children with GHD and NSD have increased circadian TSH secretion. These findings suggest that the GH secretory reserve status is a determinant of or at least contributes to the TSH release profile. Somatostatin, the inhibitory physiologic hormone for GH(33, 34), has been shown to inhibit TSH secretion(35). Experiments in animals have suggested that GH may exert short-loop positive feedback control over hypothalamic secretion of somatostatin(36–38). If a similar GH-dependent feedback mechanism operates in man, as has been suggested by Cobbet al.(19), it is reasonable to assume that GHD may result in reduced hypothalamic somatostatin production that, in turn, may result in increased TSH secretion. Several investigators have administered TRH to GHD and acromegalic subjects with normal basal serum T3 and T4 and have demonstrated that the TSH responses are significantly greater in the former group but blunted in the latter(16, 18, 19). In addition, in GHD patients who exhibit an exaggerated TSH response to TRH, a blunting of the TSH response has been reported during treatment with GH with a return of TSH responsiveness to control when GH treatment is discontinued(17).

In central idiopathic hypothyroidism, Faglia et al.(39) have found production of biologically inactive TSH due to an excess of TSH β subunits and have suggested that TRH signals the secretion of TSH with biologic potency. Morrow et al.(40) have shown that multiple TRH injections correct both quantitative and qualitative defects in TSH secretion in hypothalamic hypothyroidism and concluded that TRH regulates not only TSH secretion but also its bioactivity. It has been recently demonstrated that patients with central hypothyroidism do not have a normal nocturnal increase in TSH pulse amplitude(41). In contrast, the present observations reveal a nocturnal TSH increase in two patients with hypothalamic hypothyroidism (cases 2 and 4) as assessed by cosinor function (acrophase 0.03 and 23.46 h, respectively). That TSH of low biologic activity is being secreted is a possibility with the preservation of circadian rhythms in patients with hypothalamic hypothyroidism representing a less severe degree of tertiary hypothyroidism.

Considerable evidence indicates that the dopaminergic system inhibits TSH and PRL secretion(42, 43). Acute dopamine receptor blockade resulting from i.v. metoclopramide administration has been shown to induce a greater increase in serum TSH and PRL levels during the night as opposed to during the day in men(44, 45). Rossmanith et al.(46) confirmed previous observations that dopamine infusion or bromocriptine administration inhibits TSH secretion and abolishes the nocturnal surge of TSH, and proposed a decrease in dopaminergic activity as assessed by TSH pulse amplitude in response to an infusion of metoclopramide for a 24-h period. These investigators concluded that dopamine inhibits TSH secretion. Our current data suggested that there are no differences between mean PRL concentrations and any pulse parameters in NSD GHD and control subjects, rendering it unlikely that increased TSH secretion in both groups of patients reflects a decrease in dopaminergic tone, different from Shulman et al.(47) data. If this was the case, increased prolactin secretion should be observed, because lactotrophs are remarkably sensitive to dopamine blockage.

Thyroid hormone may influence pulsatile TSH secretion. In effect, T3 rapidly decreases TSH secretion in vitro and in vivo(46). Short-term variations in serum T3 and T4 have been previously described(5) despite the very long half-lives of these hormones. Similar short-term variations were found in patients receiving exogenous T4 replacement with little correlation between peripheral serum T4 or T3 concentrations and pulsatile TSH secretion patterns(32). Brabant et al.(9) found a decrease in TSH pulse amplitude but no change in pulse frequency in response to infusion of T3 and T4. These observations, taken together, suggest that circadian variations in pulsatile TSH release could not account for short-term thyroid hormone variations. Consistent with these findings, when free T4 values were fitted to a cosinor function, there were no circadian rhythms in control and GHD patients (data not shown).

In summary, we have investigated the profile of TSH and PRL secretion in control GHD and NSD children using frequent sampling in conjunction with recently developed techniques to quantitate statistically hormone rhythmicity and pulsatility. These combined approaches demonstrate circadian and episodic pulsatility for TSH and PRL secretion. Furthermore, the degree of integrity of GH secretion from adenohypophysis has an important role in the control of TSH secretion, with GHD patients demonstrating increased TSH release and reflecting increased pulse amplitude and interpulse concentrations rather than changes in pulse frequency and duration.