Discussion

We have confirmed that in humans there is a circadian rhythm of TSH with a peak level occurring at around 0240 h and levels remaining above the mesor from 20200820 h. We have also confirmed that FT3 shows a circadian rhythm (2, 3, 5), although with lower amplitude than TSH. We have now demonstrated that FT3 levels peak approximately 90 min after TSH levels at around 0404 h and remain above the median level from 22001000 h. Although FT4 showed significant rhythmicity, this was not evident from the raw data and did not correlate with either the TSH or FT3 rhythm.

We used single cosinor analysis as our primary tool for interrogating the data for a circadian rhythm. Our objective was to detect the presence of a circadian rhythm, should one exist, without regard to its detailed shape. A necessary condition for the existence of a periodic function of a given period is the presence of its fundamental. Therefore, the detection of a single cosinor was adequate for testing our hypotheses. A limitation of this analysis was the availability of only a single 24-h profile from each subject. Although longer sampling periods would improve statistical accuracy, this restriction only precludes detection of rhythms of a longer period, outside the current hypothesis. In addition, it was assumed that the period of any rhythm would be 24 h instead of estimating it directly from the data. Exploratory studies with T ranging from 23.524.5 h made small quantitative differences (data not shown). TSH levels were measured hourly and FT4 and FT3 every 20 min because there was an insufficient sample to remeasure TSH. Previous studies using more frequent sampling have shown that TSH has an ultradian pulsatile secretion in addition to its diurnal rhythm (19, 20).

Despite the aforementioned limitations, it is evident from the data that the TSH rhythm is well described by a single sinusoid. For TSH, 100% of the profiles showed a significant sinusoidal component, and when comparing the group cosinor prediction with the mean levels of TSH, there was a close fit for a single sinusoid. The FT3 data showed a similar profile to TSH but with smaller amplitude and a time lag of approximately 90 min. For FT4 cosinor analysis showed that in 75.9% of subjects, a sinusoid was better than a straight line to describe the data, but the rhythm was not evident from inspection of the raw data, and the amplitude was low.

We observed that the peak in FT3 lagged behind that in TSH by approximately 90 min, and there was a strong correlation between the time-adjusted FT3 and TSH levels (? = 0.80; P < 0.0001). The strong temporal relationship between TSH and FT3 levels and the positive correlation between time-adjusted FT3 and TSH levels suggest that the variation in FT3 levels is determined by TSH. This would be consistent with the observation that the increase in TSH and T3 levels correlates after treatment with the dopamine antagonist metoclopramide (12). Only 20% of T3 is derived from thyroid secretion and the remainder through peripheral conversion to T3. Thus, it is possible that TSH either stimulates T3 release from the thyroid or increases conversion to T3 in the tissues. A previous study addressing this issue concluded that the change in FT3 was not due to peripheral conversion (11), and, therefore, it seems likely that the change in FT3 relates to TSH stimulation of thyroid hormone release. The failure to show a circadian rhythm of T3 in previous studies may relate to small sample size and the assay sensitivity at that time (7, 8).

There are many variables that determine the biological action of a hormone: rhythmicity, absolute level, affinity for receptor, and receptor occupancy required for maximal intracellular signaling. The observation that FT3 levels are dependent on an approximate 72% change in TSH levels from nadir to peak confirms that at these serum concentrations, the variation in level of TSH has a biological action. The change in FT3 from nadir to peak was only 11.2% of the mean FT3 level. The low amplitude could be explained by the fact that only 20% of T3 is derived from thyroid gland secretion, that 97% of T3 is bound to thyroid binding globulin, and that the serum half-life of T3 is 0.75 d. The biological significance of this variation in FT3 is not known. For T4 and T3 sampled at the same time of day at monthly intervals, there is little variation in serum T4 and T3 within individual subjects compared with the variation within the population, and maybe a small change in thyroid hormone has physiological significance for the individual subject (21). Many hormones in addition to having a circadian rhythm may also show pulsatility. This is true for TSH, which has an ultradian rhythm with frequent small amplitude pulses (19), although the physiological significance of this is not established.

The clinical significance of any circadian rhythm in thyroid hormones has yet to be established. All the anterior pituitary hormones have a circadian rhythm with an increase in hormone levels overnight. ACTH and, thus, cortisol have a very distinct circadian rhythm with levels low before sleep and increasing from about 02000400 h (22). Although the role of the cortisol circadian rhythm in normal physiology has not been established, the loss of this rhythm does result in significant pathology. This is seen in congenital adrenal hyperplasia, in which recent attempts to replace the circadian rhythm have resulted in better biochemical disease control (23), and oral preparations of hydrocortisone that deliver circadian therapy have been developed (24). The TSH rhythm is distinct from ACTH, and TSH levels appear to increase earlier around the time of normal bedtime, and from our data, FT3 levels are above the median levels from 22001000 h. Current preparations of T4 and T3 result in immediate hormone release, with an increase in FT4 levels within the first 2-h administration.

Most patients take their T4 dose in the morning, although a recent study examined biochemical control from taking T4 in the evening and found better control (25). Combination therapies of T4 and T3 have been tested, but none has reproduced the circadian rhythm of thyroid hormones, and there is no clear evidence that these treatments benefit patients (14). We have carefully examined and defined the normal circadian rhythm of thyroid hormones in anticipation of testing a more physiological replacement, but it remains to be proven that this would be of any benefit to patients.

Click here to READ THE FULL ARTICLE