Study on diurnal variation in TSH and freeT 4 levels of healthy adults

Thyroid stimulating hormone (TSH) secretion shows a circadian rhythm, affecting free thyroxine (fT4) secretion in the same manner. Samples are collected randomly throughout the day for serum TSH & fT4 measurement in current practice. This may affect interpretation of results if the follow up samples are collected at different times of the day from the same individual. We aimed to assess the diurnal variation in serum TSH and fT4 levels in healthy adults. Healthy adult volunteers with Sinhalese ethnicity aged 21 -50 years were selected for the study. Subjects with thyroid or any other disease, on medications including oral contraceptives and postmenopausal women were excluded. Two blood samples were drawn from each subject at 8-9 am and 3-4 pm on the same day. TSH and fT4 were measured using VITROS ECi Immunodiagnostic system. Paired T test was used to assess whether there was a statistically significant diurnal variation in hormone values. Data were analyzed using IBM SPSS Statistics Version 20 and Microsoft Excel 2013. A total number of 36 subjects with an equal number of males & females participated in the study. There was no variation in hormone levels due to age or gender. Log transformed TSH and fT4, showed a statistically significant difference for both TSH (p=0.013) and freeT4 (p<0.001) values in the morning and afternoon. Standardization of sample collection time may be important for TSH and fT4 hormone assays as this study revealed a statistically significant diurnal variation for both hormones. However further studies are needed in patients with thyroid diseases and individuals of other age groups as well to ascertain the clinical significance. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution and reproduction in any medium provided the original author and source are credited


Study Conclusions
Weeke J, Gundersen H J [7] (1978 • fT3 and TSH levels low during the day time & high at night.
• The same tendency observed for fT4; not clinically significant.
• Sample collection done between 8.00 am to 9.00 am and 2.00 pm to 4.00 pm.
• Median of morning TSH level in patients with subclinical hypothyroidism, 5.83 mU/L whereas 3.79 mU/L in the afternoon.• The range of above TSH circadian variability is 73%.
• Median of morning TSH level in patients, who take L-thyroxine, was 3.27 mU/L whereas 2.18 mU/L in the afternoon.
in TSH and fT4 levels.For the TSH immunoassay test, serum or plasma can be used.If plasma is used, heparin or Ethylenediaminetetraacetic acid (EDTA) can be used as the anticoagulant.The sample should be free of haemolysis and lipaemia.Turbid samples cannot be used.Serum/plasma separated from the clot/red cells, is stable for 5-7 days at 2-8 °C and one month if frozen at -20 °C.Repeated freezing and thawing should be avoided (5).
Serum is the preferred sample for fT4, similarly for TSH.But plasma with EDTA or heparin can also be used.T4 is stable up to 7 days at room temperature, but the best is 2-8 °C.Frozen (at -20 °C) specimens are stable up to 30 days.Repeated freezing and thawing should be avoided.Minor haemolysis and lipaemia have no significant effect on most fT4 immunoassays.However heavy haemolysis should be avoided as it can dilute the sample.Turbid samples should be centrifuged before testing, because turbidity may affect the test result.T4 auto-antibodies can interfere with the assay depending on the method (4,6).
TSH secretion occurs in a circadian rhythm.Highest concentration occurs between 2.00 am to 4.00 am, whereas the lowest is seen in 5.00 pm to 6.00 pm.Low amplitude oscillations occur throughout the day (4).This oscillation may be lost in critical illnesses and after surgery (4).There are no significant differences when race and sex are considered.TSH increases immediately after the birth and decline back to the cord blood level by three days (4).It reaches the adult range in the first week.According to some other authors, TSH secretion

Introduction
Thyroxine (T4) and triiodothyronine (T3) hormones are produced by the thyroid gland.T4 is the pro-hormone of T3 which is the active form.T4 production is much higher than the T3 production.In the thyroid gland about 40% of the T4 converts to T3 by deiodinase enzyme at cellular level.In plasma more than 99% of T3 and T4 are bound to hormone binding proteins (albumin, thyroid binding pre-albumin and thyroid binding globulin).Hence, total hormone (T4 and T3) levels change with the changes of binding protein levels due to various physiological and pathological conditions.The free fraction is the physiologically active form and independent of effects due to variation of binding proteins.Clinicians now prefer free T4 (fT4) and free T3 (fT3) levels over total hormone levels (1-4).
Thyrotropin Releasing Hormone (TRH) is synthesised in the hypothalamus and stimulates the secretion of thyroid stimulating hormone (TSH) from the anterior pituitary gland.TSH stimulates the growth and the activity of thyroid follicular cells which increase the release of T3 and T4.TRH and TSH secretion is controlled by negative feedback mechanism of T3 and T4.Some other hormones and drugs such as glucocorticoid, dopamine and somatostatin inhibit TSH secretion (1)(2)(3)(4).The hypothalamus-pituitary-thyroid axis is the main regulator of TSH and T4 levels in body (4).
TSH and fT4 are the commonly used tests in evaluating the functional status of the thyroid gland.Patients come to the laboratory to test their serum TSH and fT4 levels throughout the day and specimens are collected without considering the time of the day.Hence, sample collection time is important if there is a significant diurnal variation in TSH and fT4 levels.

Table 1: Review of previous studies
• The range of above TSH circadian variability is 64.7%.

Lucke C et al. [9] (1977)
• TSH peaks from 8.00 pm to 2.00 am and nadir from 7.00 am to 2.00 pm.
• Multiple, short-lived fluctuations observed for TSH.
• T4 levels were highest from 8.00 am to 12.00 noon and lowest from 11.00 pm to 3.00 am.• Multiple short-lived fluctuations observed for T4.

Weeke J Laurberg P [10] (1976)
• TSH level low during day and high during the night, in mild hypothyroidism.
• No significant variation seen in serum T3 level.
• Patients with severe hypothyroidism showed no significant variation in both TSH and T3.• Patients with severe hypothyroidism, treated, showed circadian periodicity in serum TSH levels.
The main objective of this study was to assess the diurnal variation in serum TSH and fT4 levels in healthy adults.

Materials and Methodology
This was an evaluation study and all specimens were analysed at the Department of Chemical Pathology, Teaching Hospital, Karapitiya.after removing outliers, the normality test results revealed that the data were skewed.So that log10 transformation was done for the remaining data to get rid of the skewness and make the data set normally distributed.This log values were used to apply paired T-test, to identify whether there was any statistically significant variation between morning and afternoon data.A p-value of <0.05 was considered as statistically significant.For the data analysis IBM SPSS Statistics Version 20 and Microsoft Excel 2013 software packages were used.

Results
A total of 36 (n=36) individuals were enrolled for this study.They were categorized in to three age groups (age 21 to 30, age 31 to 40 and age 41 to 50) and equal number of males and females were selected for each age group as it represented healthy adult population.
Figure 1 shows the diurnal changes of each observed values for both TSH and fT4. Figure 2 illustrates the distribution of TSH and fT4 values, morning and afternoon, after removing outliers.This reveals the changes of distribution before and after log10 transformation.All the data obtained from morning and afternoon TSH and fT4 values were analysed to find whether the data were normally distributed.Basic statistical analysis revealed that the skewness and kurtosis values were >+1.000 for morning values of both TSH and fT4 and kurtosis value for afternoon fT4 was < -1.000.Normality tests were performed to assess the normal distribution of observed data, and that results also revealed the data sets were not normal.
Therefore the, Resistant Rules for Outlier Labelling was applied to identify possible outliers; two outliers were detected and removed from the data set (11).Remaining  A statistically significant difference was observed only for fT4 morning and afternoon data.P-values obtained from paired T-test for fT4 (p<0.001) are lower than p=0.05 at 95% confidence interval level, and p-value is higher than p=0.05 for TSH (p=0.052).Even though the F-test results showed no significant difference between the variances, results of skewness and kurtosis of morning TSH and fT4 data, and both tests of normality for TSH and fT4 morning and afternoon data, were not normally distributed, as all the p-values for normality tests were lower than p=0.05 at 95% confidence interval level.
Therefore, the data for both TSH and fT4 (n=34) were log transformed to assess whether the values were normally distributed.The log transformed values were analysed to assess the significance of diurnal variation of both TSH and fT4 values.After log10 transformation, the skewness and kurtosis values of morning TSH and fT4 levels were within +1.000 to -1.000, which indicated the data sets were normally distributed.
The normality test results after log10 transformation, revealed that both tests of normality for TSH morning and afternoon data were normally distributed, as all the pvalues were higher than p=0.05 at 95% confidence limits.
Even after log10 transformation, only Shapiro-Wilk test gave results as normally distributed, for fT4 morning and afternoon data, whereas Kolmogorov-Smirnova test with Lilliefors Significance Correction indicate the data set still deviated from normal distribution.
After log10 transformation, remaining (n=34) values were analysed to find whether there is any significant variation between TSH and fT4 values with gender and with age groups.The results of Independent Samples Test for Gender and ANOVA analysis for three age groups revealed that the variations of log TSH and log fT4 with gender or with age group were not significant.All the pvalues are greater than p=0.05 at 95% confidence limits.
* Upper and lower values of observed data of the study were given as reference ranges.As there is no significant variation of TSH and fT4 values with gender or with age groups even after log10 transformation, the results were analysed for diurnal variation.Before applying paired T-test to find diurnal variation of the morning and afternoon log10 transformed data, they were analysed to check whether there is any significant difference between the group variances by applying F-test (12).
Calculated F-values for both logTSH (1.0566) and logfT4 (1.0051) were lower than critical F-value (1.84), at the 95% confidence interval level.Thus, paired T-test was directly for log10 transformed morning and afternoon TSH and fT4 values (n=34).Table 3 shows that a statistically significant difference is observed for both TSH and fT4 morning and afternoon data, after log10 transformation.P-values obtained from paired T-test for logTSH (p=0.013) and logFT4 (p<0.001) are lower than p=0.05 at 95% confidence interval level.
Diurnal variation observed in this study is applicable only for a healthy population and is statistically significant.However, the observed values were found within the manufacturer's reference limits for both TSH & fT4.(Table 4) There was a strong, significant and positive correlation between morning and afternoon logTSH data, and also between morning and afternoon logfT4 data.However, when the logTSH and logfT4 were considered, there was a poor correlation between the values of the two hormones, which were not significant (Table 5).

Discussion
Studies vary in their conclusions regarding the diurnal variation of TSH and fT4.The current evidence supports that the highest TSH level is observed in midnight and early morning, whereas the lowest is seen in the evening (3, 4) (10,11).There is a similar tendency for fT4 as well [7] though short-lived fluctuations of fT4 are observed throughout the day (9).This is the first study of this nature done in Sri Lanka to observe the diurnal variation in TSH and fT4.No variation was observed in hormone levels due to age or sex, both before and after log10 transformation.Free thyroxine values showed a statistically significant diurnal variation (p<0.001),whereas diurnal variation of TSH showed no statistical significance (p=0.052)prior to log10 transformation.
TSH and fT4, showed a statistically significant difference for both TSH (p=0.013) and fT4 (p<0.001)values for the morning and afternoon after log10 transformation the study population was derived from two state institutions and therefore might have introduced a sampling bias.

Table 5: Pearson Correlation
The results observed is dependent on the performance characteristics of the analytical method used; immunochemiluminescence technology.The applicability of the study findings need to be verified by repeating the study in a general healthy population using other immunoassay techniques and platforms as well.In addition, the study should be extended to patients with thyroid dysfunction, and individuals of other age groups to investigate the pattern of diurnal variation in those sub groups.
However, as the study showed a statistically significant variation for both TSH and fT4 following log transformation, for morning and afternoon samples, it's best to collect samples at a defined time period in the day, preferably morning, to minimize the changes in test results due to diurnal variation.This needs to be considered during repeat and follow-up testing especially.Establishing reference ranges for TSH and fT4 for morning and afternoon separately would also be a useful step.

Conclusion
There is a diurnal variation in log transformed TSH and fT4 values of healthy adults from 20 -50 years and it is statistically significant at 95% confidence limits.Therefore, standardization of sample collection time may be important for TSH and fT4 hormone assays.However, further studies are needed in patients with thyroid diseases and individuals of all age groups to ascertain the clinical significance of the diurnal variation observed.

Table 2 : Paired Samples T-test for TSH and FT4, morning and afternoon data
Calculated F-values for both TSH (1.2800) and FT4 (1.0811) are lower than critical F-value (1.84), which were obtained from F-table at 95% confidence interval level.Because of that the null hypothesis is accepted as true which is that, there is no significant difference between the variances.So paired T-test can be applied directly for remaining morning and afternoon TSH and FT4 values (n=34).Table2shows the results of paired T-test for TSH and FT4, morning and afternoon data.