Effect of Seasonal Changes on the Transition Between Subclinical Hypothyroid and Euthyroid Status

Transcription

1 ORIGINAL Endocrine ARTICLE Research Effect of Seasonal Changes on the Transition Between Subclinical Hypothyroid and Euthyroid Status Tae Hyuk Kim, Kyung Won Kim, Hwa Young Ahn, Hoon Sung Choi, Hojeong Won, Yunhee Choi, Sun Wook Cho, Jae Hoon Moon, Ka Hee Yi, Do Joon Park, Kyong Soo Park, Hak C. Jang, Seong Yeon Kim, and Young Joo Park Department of Internal Medicine (T.H.K., H.S.C., S.W.C., D.J.P., K.S.P., S.Y.K., Y.J.P.), Seoul National University College of Medicine, and Healthcare System Gangnam Center (K.W.K.) and Medical Research Collaborating Center (H.W., Y.C.), Seoul National University Hospital, Seoul , Korea; Department of Internal Medicine (H.Y.A.), Chung-Ang University Hospital, Seoul , Korea; Department of Internal Medicine (J.H.M., H.C.J.), Seoul National University Bundang Hospital, Seongnam , Korea; and Department of Internal Medicine (K.H.Y.), Seoul National University Boramae Medical Center, Seoul , Korea Context: The widespread use of thyroid tests in asymptomatic individuals identifies many patients with transient subclinical hypothyroidism. Objective: Our objective was to determine the effect of seasonal change on serum TSH levels and the transition between subclinical hypothyroid and euthyroid status. Design, Setting, and Subjects: This was a retrospective longitudinal study of 1751 subclinical hypothyroid and euthyroid subjects aged over 18 years who underwent serial thyroid function tests at a health screening center between October 2003 and May Main Outcome Measures: Age-adjusted geometric mean values of the TSH level by month were calculated using linear mixed models. Adjusted odds ratios of test season and multiple baseline clinical factors were determined using generalized estimating equations. Results: During a median 36 months of follow-up, 57.9% of subclinical hypothyroid subjects reverted to euthyroidism, and 4.3% of euthyroid subjects developed subclinical hypothyroidism. The monthly distribution of follow-up TSH levels indicated a biphasic pattern, ie, an increase during the winter-spring season and a decrease during the summer-fall season, with a maximal TSH difference of 0.69 miu/l in subclinical hypothyroid and 0.30 miu/l in euthyroid subjects. Normalization of subclinical hypothyroidism was increased 1.4-fold in follow-up tests during the summer-fall followup, whereas subclinical hypothyroidism increased 1.4-fold in euthyroid subjects during the winterspring follow-up. Conclusions: The season in which thyroid testing was performed was independently related to the transition between subclinical hypothyroid and euthyroid status. Seasonal variations in TSH concentration should be considered before deciding on treatment of subclinical hypothyroidism, particularly in the areas with a wide annual temperature range. (J Clin Endocrinol Metab 98: , 2013) ISSN Print X ISSN Online Printed in U.S.A. Copyright 2013 by The Endocrine Society Received March 11, Accepted May 31, First Published Online June 14, 2013 Abbreviations: BMI, body mass index; CI, confidence interval; FT 4, free T 4 ; IQR, interquartile range; TPO, thyroid peroxidase; TFT, thyroid function test jcem.endojournals.org J Clin Endocrinol Metab, August 2013, 98(8): doi: /jc

2 doi: /jc jcem.endojournals.org 3421 Subclinical hypothyroidism is characterized by the isolated elevation of serum TSH levels without obvious symptoms. The clinical significance of this mild thyroid dysfunction is controversial, and the recommendations for screening differ considerably, even among professional committees (1, 2). TSH measurements are among the most frequently requested tests in primary care (3, 4), and the most common clinical problem is subclinical hypothyroidism, which occurs in 4.3% of the U.S. population (5) and 11.7% of the Korean population (6). Management of this condition is complicated because subclinical hypothyroidism is reversible in more cases than previously thought; one study of more than individuals showed that in 60% of patients with mild TSH elevation ( 5.5 to 10 miu/l), TSH levels normalized without any intervention (4). Although some predictors of persistent TSH elevation have been suggested, including a higher TSH level with the presence of anti-thyroid peroxidase (TPO) antibodies (7), much of the variability in TSH remains unexplained, especially in cases with mild TSH elevation. Interestingly, prolonged Antarctic residence was shown to increase TSH by approximately 30%, which suggests that prolonged cold exposure could affect TSH levels in human adults (8). This modest effect on TSH could partly explain changes in thyroid function in patients whose TSH levels are near the upper limit of normal. Because Korea has a continental climate characterized by substantial temperature differences between winter and summer, an analysis of Korean inhabitants would permit an assessment of the effect of temperature differences on thyroid status. In this study, we estimated the effect of seasonal changes on serum TSH levels and the transition between subclinical hypothyroid and euthyroid status in a large institutional cohort. We also evaluated the effects of the degree of baseline TSH elevation, age, gender, smoking habits, body mass index (BMI), and follow-up duration on the transition between these 2 statuses. Subjects and Methods Study population and data sources These longitudinal analyses were based on the database of the institutional cohort of Seoul National University Hospital Healthcare System, Gangnam Center, in Seoul, Korea. Most of the study subjects voluntarily paid for their health check-ups and the number of subjects during the study period determined the sample size. From October 2003 to May 2011, subjects visited the center. Of the adult subjects aged 18 years or Table 1. Characteristics of Study Subjects With Subclinical Hypothyroidism and Euthyroidism at Baseline Total Males Females Subclinical Subclinical Subclinical Hypothyroid Euthyroid Hypothyroid Euthyroid Hypothyroid Euthyroid Characteristics (n 1751) (n ) (n 721) (n ) (n 1030) (n ) Age (y), 48.6 (11.6) 46.9 (10.4) 49.5 (12.0) 47.5 (10.3) 47.9 (11.3) 45.9 (10.5) mean (SD) Female, n (%) 1030 (58.8) (41.4) Current smoker, 146 (8.3) 4488 (16.0) 119 (16.5) 4162 (25.3) 27 (2.6) 326 (2.8) n (%) BMI (kg/m 2 ), 23.0 (3.0) 23.5 (3.0) 24.4 (2.7) 24.5 (2.6) 22.1 (2.8) 22.0 (2.8) mean (SD) a TSH (miu/l) Geometric mean (95% CI) 5.46 ( ) 1.56 ( ) 5.45 ( ) 1.46 ( ) 5.47 ( ) 1.69 ( ) Median (IQR) 5.00 ( ) 1.58 ( ) 4.95 ( ) 1.48 ( ) 5.03 ( ) 1.75 ( ) Free T 4 (ng/dl), 1.14 (0.24) 1.24 (0.25) 1.21 (0.25) 1.30 (0.24) 1.08 (0.22) 1.17 (0.24) mean (SD) Season of baseline test, n (%) Winter (December February) 494 (28.2) 6834 (24.3) 221 (30.7) 3986 (24.2) 273 (26.5) 2848 (24.5) Spring (March May) Summer (June August) Fall (September November) 404 (23.1) 6526 (23.2) 167 (23.2) 3814 (23.2) 237 (23.0) 2712 (23.3) 438 (25.0) 7219 (25.7) 176 (24.4) 4162 (25.3) 262 (25.4) 3057 (26.3) 415 (23.7) 7517 (26.8) 157 (21.8) 4513 (27.4) 258 (25.0) 3004 (25.8) a Calculated as weight in kilograms divided by height in meters squared.

3 3422 Kim et al Season and Thyroid Function J Clin Endocrinol Metab, August 2013, 98(8): older who underwent thyroid function tests (TFTs), including TSH and free T 4 (FT 4 ), we identified subjects who had at least 2 or more test results. At baseline, we excluded subjects with a history of thyroid disease, those on thyroxine or antithyroid drug treatment at initial visit, and those taking medications that could interfere with thyroid function (ie, amiodarone, antidepressants, and hormonal replacement) or who had suffered from serious comorbidities or cancer. In total, 1601 subjects were excluded according to these criteria. Among the subjects remaining, 1751 (5.7%) subjects with subclinical hypothyroidism and (91.5%) subjects with euthyroidism at baseline were included in the final analysis. Baseline characteristics are presented in Table 1. The mean air temperature for each day and month from 2003 through 2011 was obtained from the Korea Meteorological Administration, which has a recording station in Seoul (9). Seasons were defined according to the meteorological definition of winter as December to February, spring as March to May, summer as June to August, and fall as September to November (10). Details regarding the number of patients and samples drawn according to each year and season are shown in Supplemental Table 1 (published on The Endocrine Society s Journals Online web site at For ease of interpretability, seasonal categories were combined into winter-spring and summer-fall periods for further analysis. This study was approved by the Institutional Review Board of Seoul National University Hospital with a waiver for informed consent. Statistical analysis We stratified subjects with subclinical hypothyroidism and euthyroidism by baseline values of age (18 50 or 50 years), sex, smoking habits (current smoker or not), BMI ( 25.0 or 25.0 kg/m 2 ) (12), TSH ( , , and miu/l for the euthyroid group; , , and miu/l for the subclinical hypothyroid group), season of the baseline and last follow-up test (summer-fall or winterspring), and follow-up duration (0 3, 4 24, 25 48, and 48 months). The 2 test was used to compare the last follow-up status across strata. The age-adjusted geometric mean and 95% confidence interval (CI) of TSH levels by month of baseline and follow-up tests were calculated using a linear mixed model to include all TFT results in longitudinal data. In our mixed model, the intercept and the regression coefficient for the age of the sub- Clinical and laboratory assessments Each subject completed a structured questionnaire that included health-related questions addressing history of thyroid disease, smoking habits, and pharmaceutical information at the visit. Current smoking was defined as smoking regularly for the previous 12 months. Height and body weight were measured using a digital scale, and BMI was calculated. A blood sample was drawn from each subject after a 12- hour overnight fast. The concentrations of TSH and FT 4 were measured at the central laboratory of Seoul National University Hospital using an immunoradiometric assay with an RIAmat 280 device (Stratec, Birkenfeld, Germany). The TSH assay had a functional sensitivity of miu/l, a reference range of 0.40 to 4.10 miu/l, and 2.5% intra-assay and 3.5% interassay coefficients of variation. The FT 4 assay had a functional sensitivity of 0.08 ng/dl, a reference range of 0.7 to 1.8 ng/dl, and 2.7% intra-assay and 3.8% interassay coefficients of variation. The Korean population is considered to have sufficient iodine intake (11). Classification of thyroid status The subjects were placed into 4 categories based on their TFTs: 1) overt hypothyroidism was defined as TSH 4.10 miu/l with FT ng/dl or thyroxine treatment; 2) subclinical hypothyroidism was defined as TSH 4.10 miu/l with normal FT 4 ( ng/dl); 3) euthyroidism was defined as normal TSH ( miu/l); and 4) others included both subclinical hyperthyroidism (defined as TSH 0.40 miu/l with normal FT 4 ) and overt hyperthyroidism (defined as TSH 0.40 miu/l with FT ng/dl or antithyroid drug treatment). Figure 1. Age-adjusted mean TSH level by month of the baseline and follow-up tests in subclinical hypothyroid (A) and euthyroid (B) subjects. The age-adjusted geometric mean (solid line) and 95% CI (error bar) of TSH levels were assessed by a linear mixed model using the baseline and all of the follow-up data. The white (E) and black (F) circles indicate baseline and follow-up TSH levels, respectively. The dotted line indicates the monthly mean temperature of Seoul during the study period.

4 doi: /jc jcem.endojournals.org 3423 ject at each test was treated as a random effect such that each subject had a unique intercept and regression coefficient. Associations between baseline TSH level (log-transformed value) and mean daily temperature when each sample was taken as well as 1, 2, and 3 months before the test were also investigated using the Pearson correlation coefficient. Significant differences between correlations were assessed using the Hotelling-Williams test (13). Finally, we constructed generalized estimating equation models to estimate the effects of season at baseline and follow-up as well as the effect of other baseline clinical factors on transitions between the 2 conditions using all of the follow-up data. In these models, each subject was considered as a clustering variable, and the duration between the baseline and each follow-up test was adjusted for unequal follow-up. We performed a correlation analysis between risk factors to rule out multicollinearity. With the exception of female sex/higher BMI ( 0.30) and female sex/current smoking ( 0.31), all correlation coefficients were below 0.30; thus, we also performed sex-stratified generalized estimating equation analysis. Statistical analysis was performed using SAS version 9.2 (SAS Institute, Cary, North Carolina) software. A P value.05 was considered statistically significant. Results Monthly mean TSH level at baseline and entire follow-up Overall, 1751 subclinical hypothyroid and euthyroid subjects had TFT assessments (mean, 3.1 tests per subject) during a median (interquartile range [IQR]) follow-up of 36 (22 53) months. The TSH levels of subclinical hypothyroid subjects during the entire fol- Table 2. Changes in Thyroid Status at Last Follow-up by Baseline Characteristics and Season a Subclinical Hypothyroid Group at Baseline Euthyroid Group at Baseline Characteristics Total Euthyroid Subclinical Hypothyroid Overt Hypothyroid Others Total Euthyroid Subclinical Hypothyroid Overt Hypothyroid Others n (%) 1751 (100) 1013 (57.9) 656 (37.5) 55 (3.1) 27 (1.5) (100) (93.7) 1195 (4.3) 124 (0.4) 439 (1.6) Mean follow-up (months), 29 (13 47) 30 (13 47) 28 (13 48) 24 (12 37) 38 (26 54) 36 (22 53) 36 (22 53) 41 (24 62) 42 (23 61) 37 (23 55) median (IQR) Age (y) (100) 605 (61.1) 346 (34.9) 21 (2.1) 18 (1.8) (100) (94.3) 673 (3.7) 79 (0.4) 281 (1.5) (100) 408 (53.6) b 310 (40.7) c 34 (4.5) b 9 (1.2) 9877 (100) 9152 (92.7) d 522 (5.3) d 45 (0.5) 158 (1.6) Sex Male 721 (100) 417 (57.8) 276 (38.3) 21 (2.9) 7 (1.0) (100) (95.1) 545 (3.3) 44 (0.3) 217 (1.3) Female 1030 (100) 596 (57.9) 380 (36.9) 34 (3.3) 20 (1.9) (100) (91.8) d 650 (5.6) d 80 (0.7) d 222 (1.9) d Current smoker No 1605 (100) 920 (57.3) 604 (37.6) 54 (3.4) 27 (1.7) (100) (93.2) 1098 (4.7) 112 (0.5) 384 (1.6) Yes 146 (100) 93 (63.7) 52 (35.6) 1 (0.7) 0 (0) 4488 (100) 4324 (96.3) d 97 (2.2) d 12 (0.3) 55 (1.2) c BMI (kg/m 2 ) e (100) 744 (57.8) 473 (36.8) 48 (3.7) 22 (1.7) (100) (93.4) 892 (4.6) 90 (0.5) 298 (1.5) (100) 257 (58.4) 172 (39.1) 6 (1.4) c 5 (1.1) 8491 (100) 8032 (94.6) d 292 (3.4) d 34 (0.4) 133 (1.6) TSH (miu/l) f,g Category (100) 858 (64.1) 439 (32.8) 20 (1.5) 21 (1.6) (100) (96.6) 106 (0.8) 43 (0.3) 280 (2.2) Category (100) 119 (37.3) 173 (54.2) 23 (7.2) 4 (1.3) 9314 (100) 8890 (95.4) 298 (3.2) 32 (0.3) 94 (1.0) Category 3 94 (100) 36 (38.3) d 44 (46.8) d 12 (12.8) d 2 (2.1) 5997 (100) 5092 (84.9) d 791 (13.2) d 49 (0.8) d 65 (1.1) d Season of baseline test Summer-Fall 853 (100) 506 (59.3) 312 (36.6) 26 (3.0) 9 (1.1) (100) (93.7) 621 (4.2) 63 (0.4) 249 (1.7) Winter-Spring 898 (100) 507 (56.5) 344 (38.3) 29 (3.2) 18 (2.0) (100) (93.8) 574 (4.3) 61 (0.5) 190 (1.4) Season of last follow-up Summer-Fall 767 (100) 494 (64.4) 241 (31.4) 20 (2.6) 12 (1.6) (100) (94.8) 404 (3.1) 56 (0.4) 214 (1.6) Winter-Spring 984 (100) 519 (52.7) d 415 (42.2) d 35 (3.6) 15 (1.5) (100) (92.8) d 791 (5.2) d 68 (0.5) 225 (1.5) Follow-up duration (months) g (100) 85 (55.6) 59 (38.6) 8 (5.2) 1 (0.7) 76 (100) 72 (94.7) 1 (1.3) 0 (0) 3 (3.9) (100) 326 (57.4) 218 (38.4) 21 (3.7) 3 (0.5) 9022 (100) 8555 (94.8) 300 (3.3) 36 (0.4) 131 (1.5) (100) 373 (59.0) 227 (35.9) 17 (2.7) 15 (2.4) (100) 9413 (93.8) 426 (4.2) 36 (0.4) 163 (1.6) (100) 229 (57.5) 152 (38.2) 9 (2.3) c 8 (2.0) c 8960 (100) 8298 (92.6) d 468 (5.2) d 52 (0.6) 142 (1.6) a Data are shown as n (percent) unless indicated otherwise. Each column represents the last follow-up status of thyroid function. b P.01 for 2-group comparisons from 2 tests or for linear trend across strata. c P.05. d P.001. e Subjects with missing data were excluded. f For the subclinical hypothyroid group, categories 1, 2, and 3 mean TSH , , and 10.00, respectively. For the euthyroid group, categories 1, 2, and 3 mean TSH , , and , respectively. g P values for trends across strata were obtained from 2 tests for linear trends.

5 3424 Kim et al Season and Thyroid Function J Clin Endocrinol Metab, August 2013, 98(8): low-up were significantly decreased compared with the baseline value (3.34 vs 5.41 miu/l, P.001). The monthly distribution of follow-up TSH levels indicated a biphasic pattern: they increased during winter-spring and decreased during summer-fall, with a maximal TSH difference of 0.69 miu/l, but this pattern was not observed in baseline tests (Figure 1A). For euthyroid subjects, a similar biphasic pattern was also observed, with modest TSH differences at both baseline (0.26 miu/l) and in follow-up tests (0.30 miu/l) (Figure 1B). There was a significantly negative correlation between the baseline TSH level and the temperature of the day on which the sample was taken (r 0.091, P.001 for all correlation coefficients). This correlation magnitude was similar for temperatures from 1 month earlier (r 0.083) and significantly greater for temperatures from two months earlier (r 0.049, P.001 for difference). Transitions of thyroid status at last follow-up relative to baseline At the last follow-up, of the 1751 baseline subclinical hypothyroid subjects, 57.9% (annual rate of 21.7%) reverted to euthyroidism, 37.5% remained subclinically hypothyroid, and 3.1% (annual 1.2%) progressed to overt hypothyroidism during a median 29 months of follow-up (Table 2). Normalization of subclinical hypothyroidism (n 1013) was more common in the younger group and in those with the lowest TSH level at baseline but did not differ according to sex, smoking status, BMI, or follow-up duration. Of the euthyroid subjects, during a median 36 months of follow-up, the majority (93.7%) remained euthyroid, and 4.3% (annual 1.3%) developed subclinical hypothyroidism, which represented the highest prevalence of any TFT abnormality. The development of subclinical hypothyroidism (n 1195) was more common in older subjects, nonsmokers, nonobese subjects, women, and individuals with a higher TSH category (P for trend.001) and longer follow-up duration (P for trend.001). Transition of thyroid status at last follow-up by season Among subjects with subclinical hypothyroidism, those who had their last follow-up test in the summer-fall season were significantly more likely to revert to euthyroidism than those who had their last follow-up test in the winter-spring season (64.4% vs 52.7%). In contrast, the development of subclinical hypothyroidism among euthyroid subjects was more likely in individuals who had their last assessment in the winter-spring season (5.2% vs 3.1%). Figure 2. Thyroid status at last follow-up by month in subjects with subclinical hypothyroidism (A) and euthyroidism (B). Other categories include both subclinical hyperthyroidism and overt hyperthyroidism. At the bottom of the figure, the number of subjects who reverted to euthyroid status from subclinical hypothyroid status (A) and developed subclinical hypothyroidism from euthyroidism (B) is shown. Abbreviation: SHypo, subclinical hypothyroid. Figure 2 shows the monthly distribution of thyroid status at last follow-up. In accordance with the monthly TSH distribution (Figure 1), the normalization of subclinical hypothyroidism (range 49.4% 71.4%) was greater in the summer-fall season (Figure 2A). However, the transition from euthyroid to subclinical hypothyroid status ranged from 2.2% to 6.4% and was more common in the winterspring season (Figure 2B). There was no difference in the proportion of transitions between subclinical hypothyroid, and euthyroid subjects according to the season of the baseline test (Table 2). The results of additional analyses according to the follow-up durations (4 24, 25 48, and 48 months) were similar to those in Table 2 (data not shown).

6 doi: /jc jcem.endojournals.org 3425 Table 3. Effects of Clinical Characteristics and Seasonal Change on Transition from Subclinical Hypothyroid to Euthyroid Status a Total Males Females Characteristics OR (95% CI) P OR (95% CI) P OR (95% CI) P Age (y) ( ) ( ) ( ).09 Sex Male Female 1.09 ( ).37 Current smoker No Yes 1.40 ( ) ( ) ( ).42 BMI (kg/m 2 ) ( ) ( ) ( ).51 TSH (miu/l) b Category 1 Category ( ) ( ) ( ).001 Category ( ) ( ) ( ).001 Season of baseline test Summer-Fall Winter-Spring 1.02 ( ) ( ) ( ).22 Season of follow-up test Summer-Fall Winter-Spring 0.70 ( ) ( ) ( ).001 Follow-up duration (1 y) 1.04 ( ) ( ) ( ).55 Abbreviation: OR, odds ratio. a Each column represents one generalized estimating equation model using all the follow-up data. All models include all variables listed, except for sex in sex-stratified analysis. b Categories 1, 2, and 3 mean TSH , , and 10.00, respectively. Multivariable-adjusted models of the transition between subclinical hypothyroid and euthyroid status As shown in Tables 3 and 4, the season of the follow-up test remained significantly and independently associated with transition of thyroid status in the multivariable-adjusted model. Compared with summer-fall, the follow-up test in the winter-spring season was associated with 40% increased odds of incident subclinical hypothyroidism in individuals with euthyroidism and 30% lower odds of normalization in individuals with subclinical hypothyroidism; this difference in odds of transition became more marked, to 52% and 38%, respectively, when we compared the winter and summer follow-up (Table 5). Euthyroid subjects who had a baseline test in the winter-spring season were significantly less likely to develop subclinical hypothyroidism than those who had the test in the summer-fall after adjusting for baseline TSH category, which was not significant in the univariate analysis. Other significant baseline predictors of the normalization of subclinical hypothyroidism were younger age (in males), current smoking, and lowest TSH elevation. Current smoking was not significant in the sex-stratified analysis. The transition from euthyroidism to subclinical hypothyroidism was associated with older age, female sex, lower BMI (in males), higher TSH category, and longer follow-up period. Discussion In this study, we showed that the season in which the thyroid test was performed was related to the serum TSH concentration and played an independent role in the transition between subclinical hypothyroid and euthyroid status. Overall, the TSH levels increased during winter-spring and decreased during the summer-fall season. The normalization of subclinical hypothyroidism increased 1.4- fold on follow-up tests performed during the summer-fall, whereas subclinical hypothyroidism increased 1.4-fold in euthyroid subjects during the winter-spring follow-up. To the best of our knowledge, this is the first longitudinal study to show a seasonal effect on thyroid function in a large population of subclinical hypothyroid and euthyroid subjects. The modest effect of test season on TSH level is particularly important because it may affect the TSH distribution of the whole population and result in periodic changes in the relative prevalence of subclinical

7 3426 Kim et al Season and Thyroid Function J Clin Endocrinol Metab, August 2013, 98(8): Table 4. Effects of Clinical Characteristics and Seasonal Change on Transition from Euthyroid to Subclinical Hypothyroid Status a Total Males Females Characteristics OR (95% CI) P OR (95% CI) P OR (95% CI) P Age (y) ( ) ( ) ( ).001 Sex Male Female 1.25 ( ).001 Current smoker No Yes 0.88 ( ) ( ) ( ).91 BMI (kg/m 2 ) ( ) ( ) ( ).40 TSH (miu/l) b Category 1 Category ( ) ( ) ( ).001 Category ( ) ( ) ( ).001 Season of baseline test Summer-Fall Winter-Spring 0.80 ( ) ( ) ( ).02 Season of follow-up test Summer-Fall Winter-Spring 1.40 ( ) ( ) ( ).001 Follow-up duration (1 y) 1.08 ( ) ( ) ( ).001 Abbreviation: OR, odds ratio. a Each column represents one generalized estimating equation model using all the follow-up data. All models include all variables listed, except for sex in sex-stratified analysis. b Categories 1, 2, and 3 mean TSH , , and , respectively. hypothyroidism and euthyroidism determined by a fixed upper TSH reference limit. The present study s results are notable because of the large number of subjects and the analysis of all serial test results. Based on this longitudinal analysis, we found the reciprocal relationship of subclinical hypothyroid and euthyroid prevalence that could provide a broader view of the natural history of transient subclinical hypothyroidism. We refined the TSH categories based on FT 4 measurements with TSH in timed blood samples. We also adjusted multiple clinical factors that influence the TSH level and estimated their effects on transitions of thyroid status. Subclinical hypothyroidism has been reported to occur in 4% to 18% of the adult population (2). However, at least 75% of these individuals have mild TSH elevation (TSH 10.0 miu/l) (14), and such patients often revert to euthyroidism without treatment (7). A previous large Israeli study reported that 3.0% had mildly elevated TSH ( 5.5 to 10.0 miu/l) at baseline; of these individuals, 62.1% of them became normal, 34.6% remained elevated, and 2.9% progressed at a second TSH measurement after a mean interval of 19 months (4). These results are consistent with ours. For euthyroid subjects, a longitudinal study reported that 9.3% developed hypothyroidism, defined as a TSH increase greater than 4.0 miu/l or use of thyroxine treatment, after 13 years (15). Clinical factors known to influence the prevalence of subclinical hypothyroidism include age (16), sex (5), BMI (17), smoking status (18), ethnicity (16), dietary iodine intake (19), underlying autoimmune thyroid diseases (20), and upper TSH reference limits used to define the status. The effect of seasonal variation in thyroid function has been reported in healthy individuals (21 23), but not in another small study (24). In a study of 26 healthy Belgian adults, significant annual variation of TSH was observed; the lowest TSH values were observed in spring. The difference between peak (December) and trough (June) was 29%, expressed as a percentage of the mean (25). Similar, but greater, seasonal variation of TSH related to changes in temperature was reported in a small number of hypothyroid patients on thyroxine treatment (26). In addition, Hamada et al (27) found a slight fall in thyroid hormone levels seasonally in 7 hypothyroid patients on treatment, but not in 8 normal controls. A recent study in a polar area showed that prolonged cold exposure led to modest TSH elevation in healthy adults accompanied by an increase in thyroglobulin and T 3 (8), presumably due to the increased need for thyroid

8 doi: /jc jcem.endojournals.org 3427 Table 5. Status a Effects of 4 Seasons on Transitions of Thyroid Odds ratio (95% CI) a P Subclinical hypothyroid to euthyroid Season of baseline test Winter Spring 0.81 ( ).10 Summer 0.86 ( ).20 Fall 0.95 ( ).66 Season of follow-up test Winter Spring 1.07 ( ).48 Summer 1.61 ( ).001 Fall 1.39 ( ).001 Euthyroid to subclinical hypothyroid Season of baseline test Winter Spring 1.01 ( ).92 Summer 1.35 ( ).001 Fall 1.17 ( ).04 Season of follow-up test Winter Spring 0.95 ( ).34 Summer 0.66 ( ).001 Fall 0.72 ( ).001 a Odds ratio (95% CI) was calculated using multiple models. Each season was used for the analyses instead of 2 categories (summer-fall and winter-spring). hormone by peripheral tissues. In our study, the TSH levels were inversely related to changes in air temperature up to 2 months before blood sampling, predominantly in the colder period. We speculate that people who spend almost of their time in heated rooms would have a smaller TSH change in reaction to cold exposure, because adaptive thermogenesis is minimized at thermoneutrality temperature, which is approximately 23 C in adult humans (28, 29). Other studies of subarctic populations suggested that seasonal differences in thyroid function could be related not only to ambient temperature but also to photoperiod (30, 31). Alterations in daylight hours may affect the seasonal TSH change, because melatonin drives changes in multiple neuroendocrine pathways in mammals (32). However, the lack of a seasonal pattern in baseline TSH levels in subclinical hypothyroid subjects in our study suggests a minimal effect when the test is performed in the case of definite TSH elevation, which is representative of autoimmune thyroid disease (7). The effects of conventional clinical factors were generally consistent with our findings. A positive correlation between age and TSH concentration was previously reported, particularly in an iodine-sufficient population (16, 33). The present study also found the positive relationship in both men (r 0.033, P.001) and women (r 0.050, P.001). In a study of subclinical hypothyroid elderly individuals aged over 65 years, 35% became euthyroid and 9% progressed to overt hypothyroidism or started thyroid hormone replacement after 2 years (34). The present study also demonstrated that older subjects had 1.5- fold higher odds of incident subclinical hypothyroidism and were 1.3-fold more likely to remain at that status. In the Whickham Survey (35), women who had elevated TSH alone or positive TPO antibody alone had a 2.6% and 2.1%, respectively, annual risk of developing overt hypothyroidism. This risk increased to 4.3% in women who had both conditions. A U.S. population study reported that the TSH level and prevalence of antithyroid antibodies were higher in females (5). Similarly, our results show that the development of subclinical hypothyroidism was more common in females, although we could not evaluate the effects of TPO antibody in this study. In contrast, Walsh et al (15) reported that age was no longer significant in a multivariate analysis and that thyroid antibodies were of borderline significance in euthyroid subjects. The TSH concentration at baseline was a useful predictor for the transition of thyroid status in this study. The association of upper normal TSH ( miu/l) with long-term hypothyroidism was reported in a 13-year follow-up study (15). An Israeli study also reported that resolution of subclinical hypothyroidism was more common in subjects with mildly elevated TSH than in those with highly elevated TSH ( 10.0 miu/l) (62.1% vs 27.2%) (4). The association of current smoking with lower TSH level and a lower prevalence of subclinical hypothyroidism were reported in cross-sectional studies (18, 36). In the present study, the TSH level of current smokers was lower than that of nonsmokers, although the impact of smoking on the transition of thyroid status was not significant in a sex-specific model. Epidemiologic studies also reported positive correlations between TSH level and BMI, which was consistent with our data (17, 37). We found that obese subjects who were euthyroid at baseline had a 16% lower chance of becoming subclinically hypothyroid than their nonobese counterparts. Because subclinical hypothyroidism is frequently reversible and a similar number of euthyroid individuals transit into subclinical hypothyroidism, an adequate follow-up scheme that considers the multiple clinical factors mentioned above is necessary. During follow-up for subclinical hypothyroidism, resolution occurred constantly over time in this study. In cases with a high likelihood of normalization, such as young subjects with minimal TSH elevation ( miu/l), it is prudent to perform follow-up tests in the summer-fall season before deciding on treatment. It is noteworthy that individuals who converted to euthyroidism were still prone to develop subclinical hypothyroidism again because their final TSH was higher (geometric mean 2.43, 95% CI ) than that of the original euthyroid population. The final TSH levels

9 3428 Kim et al Season and Thyroid Function J Clin Endocrinol Metab, August 2013, 98(8): reached by these patients were reported in the range of 3 to 5 miu/l in an earlier study (7). Initially, euthyroid subjects were found to remain in the same status in most the population but developed subclinical hypothyroidism in proportion to follow-up time. Therefore, a selective long-term follow-up scheme is recommended for those with a high risk of developing subclinical hypothyroidism, particularly with TSH levels of 2.5 to 4.1 miu/l. It is important to account for seasonal variation when comparing serial TSH results. Seasonal variation of TSH has clinical relevance particularly for asymptomatic individuals whose TSH values are near the upper limit of the reference range because diagnoses may shift between euthyroidism and subclinical hypothyroidism. However, a transient fluctuation in TSH level does not necessarily mean that the patient s thyroid function is either deteriorated or improved because the mildly elevated TSH in the colder period would return to the previous level in the subsequent warm season. In this situation, we support the previous recommendation to serially monitor thyroid function before considering thyroid hormone replacement unless there is a compelling indication, such as pregnancy (38). Another important point to consider is the regression to the mean, which is a natural statistical phenomenon in repeated data (39). Because TSH has significant day-today and diurnal variation, a single result that is slightly outside the reference range is likely to revert to normal on repeat testing. In our data, the frequent normalization of subclinical hypothyroidism when tested in the same season probably reflected this phenomenon (51.6% in winter-spring and 64.0% in summer-fall season). There are several limitations to this study. We could not develop a more sophisticated model to predict the normalization of subclinical hypothyroidism because our data did not allow us to determine the TPO antibody status, which is frequently associated with female sex, old age, and high TSH elevation (5). Positive TPO antibody was reported in 11.7% of Korean adults and 17.3% of the elderly population (6), which was similar to the findings of a U.S. population study (5). This study was performed in an iodine-sufficient area with distinct annual variations in temperature. We cannot be sure that the transition of thyroid status would be similar in iodine-deficient areas or whether the seasonal change in TSH would be decreased in areas with a narrow annual temperature range. Finally, a medium to high socioeconomic status of the study population also might lead to selection bias. However, socioeconomic status has not been established as a major determinant for the thyroid function. In conclusion, in a large cohort of healthcare recipients, we demonstrated a seasonal impact on the transition between subclinical hypothyroidism and euthyroidism. Seasonal variations in TSH concentration should therefore be considered before deciding on treatment of subclinical hypothyroidism, particularly in the areas with a wide annual temperature range. Acknowledgments Address all correspondence and requests for reprints to: Young Joo Park, MD, PhD, Department of Internal Medicine, Seoul National University College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul , Korea. yjparkmd@snu.ac.kr. This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Disclosure Summary: The authors have nothing to disclose. References 1. Cooper DS, Biondi B. Subclinical thyroid disease. Lancet. 2012; 379: Villar HC, Saconato H, Valente O, Atallah AN. Thyroid hormone replacement for subclinical hypothyroidism. Cochrane Database Syst Rev. 2007;3:CD Beckett GJ, Toft AD. First-line thyroid function tests: TSH alone is not enough. Clin Endocrinol (Oxf). 2003;58: Meyerovitch J, Rotman-Pikielny P, Sherf M, Battat E, Levy Y, Surks MI. Serum thyrotropin measurements in the community: five-year follow-up in a large network of primary care physicians. Arch Intern Med. 2007;167: Hollowell JG, Staehling NW, Flanders WD, et al. Serum TSH, T(4), and thyroid antibodies in the United States population (1988 to 1994): National Health and Nutrition Examination Survey (NHANES III). J Clin Endocrinol Metab. 2002;87: Choi HS, Park YJ, Kim HK, et al. Prevalence of subclinical hypothyroidism in two population based-cohort: Ansung and KLoSHA Cohort in Korea. J Korean Thyroid Assoc. 2010;3: Díez JJ, Iglesias P, Burman KD. Spontaneous normalization of thyrotropin concentrations in patients with subclinical hypothyroidism. J Clin Endocrinol Metab. 2005;90: Do NV, Mino L, Merriam GR, et al. Elevation in serum thyroglobulin during prolonged Antarctic residence: effect of thyroxine supplement in the polar 3,5,3 -triiodothyronine syndrome. J Clin Endocrinol Metab. 2004;89: Korea Meteorological Administration surface observation data. Korea Meteorological Administration website. weather/observation/past_table.jsp. Accessed March 10, American Meteorological Society classification of seasons. American Meteorological Society website. wiki/seasons. Accessed March 10, Kim JY, Moon SJ, Kim KR, Sohn CY, Oh JJ. Dietary iodine intake and urinary iodine excretion in normal Korean adults. Yonsei Med J. 1998;39: World Health Organization, International Obesity Task Force The Asian-Pacific perspective: redefining obesity and its treatment. Geneva, Switzerland: WHO Western Pacific Region; Williams EJ. The comparison of regression variables. J R Stat Soc Ser B. 1959;21: Canaris GJ, Manowitz NR, Mayor G, Ridgway EC. The Colorado thyroid disease prevalence study. Arch Intern Med. 2000;160: Walsh JP, Bremner AP, Feddema P, Leedman PJ, Brown SJ, O Leary

10 doi: /jc jcem.endojournals.org 3429 P. Thyrotropin and thyroid antibodies as predictors of hypothyroidism: a 13-year, longitudinal study of a community-based cohort using current immunoassay techniques. J Clin Endocrinol Metab. 2010;95: Surks MI, Boucai L. Age- and race-based serum thyrotropin reference limits. J Clin Endocrinol Metab. 2010;95: Knudsen N, Laurberg P, Rasmussen LB, et al. Small differences in thyroid function may be important for body mass index and the occurrence of obesity in the population. J Clin Endocrinol Metab. 2005;90: Asvold BO, Bjøro T, Nilsen TI, Vatten LJ. Tobacco smoking and thyroid function: a population-based study. Arch Intern Med. 2007; 167: Buchinger W, Lorenz-Wawschinek O, Semlitsch G, et al. Thyrotropin and thyroglobulin as an index of optimal iodine intake: correlation with iodine excretion of 39,913 euthyroid patients. Thyroid. 1997;7: Spencer CA, Hollowell JG, Kazarosyan M, Braverman LE. 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. J Clin Endocrinol Metab. 2007;92: Andersen S, Bruun NH, Pedersen KM, Laurberg P. Biologic variation is important for interpretation of thyroid function tests. Thyroid. 2003;13: Fisher DA. Physiological variations in thyroid hormones: physiological and pathophysiological considerations. Clin Chem. 1996; 42: Plasqui G, Kester AD, Westerterp KR. Seasonal variation in sleeping metabolic rate, thyroid activity, and leptin. Am J Physiol Endocrinol Metab. 2003;285:E338 E Pasquali R, Baraldi G, Casimirri F, et al. Seasonal variations of total and free thyroid hormones in healthy men: a chronobiological study. Acta Endocrinol. 1984;107: Maes M, Mommen K, Hendrickx D, et al. Components of biological variation, including seasonality, in blood concentrations of TSH, TT 3,FT 4, PRL, cortisol and testosterone in healthy volunteers. Clin Endocrinol (Oxf). 1997;46: Konno N, Morikawa K. Seasonal variation of serum thyrotropin concentration and thyrotropin response to thyrotropin-releasing hormone in patients with primary hypothyroidism on constant replacement dosage of thyroxine. J Clin Endocrinol Metab. 1982;54: Hamada N, Ohno M, Morii H, et al. Is it necessary to adjust the replacement dose of thyroid hormone to the season in patients with hypothyroidism? Metabolism. 1984;33: Donaldson GC, Tchernjavskii VE, Ermakov SP, Bucher K, Keatinge WR. Winter mortality and cold stress in Yekaterinburg, Russia: interview survey. BMJ. 1998;316: Silva JE. The thermogenic effect of thyroid hormone and its clinical implications. Ann Intern Med. 2003;139: Leppaluoto J, Sikkila K, Hassi J. Seasonal variation of serum TSH and thyroid hormones in males living in subarctic environmental conditions. Int J Circumpolar Health. 1998;57(Suppl 1): Levine M, Duffy L, Moore DC, Matej LA. Acclimation of a nonindigenous sub-arctic population: seasonal variation in thyroid function in interior Alaska. Comp Biochem Physiol A Physiol. 1995; 111: Bechtold DA, Loudon AS. Hypothalamic thyroid hormones: mediators of seasonal physiology. Endocrinology. 2007;148: Bjøro H, Alte D, Kohlmann T, Lüdemann J, Nauck M, John U, Meng W. Reference intervals of serum thyroid function tests in a previously iodine-deficient area. Thyroid. 2005;15: Somwaru LL, Rariy CM, Arnold AM, Cappola AR. The natural history of subclinical hypothyroidism in the elderly: the cardiovascular health study. J Clin Endocrinol Metab. 2012;97: Vanderpump MP, Tunbridge WM, French JM, et al. The incidence of thyroid disorders in the community: a twenty-year follow-up of the Whickham Survey. Clin Endocrinol (Oxf). 1995;43: Cho NH, Choi HS, Kim KW, et al. Interaction between cigarette smoking and iodine intake and their impact on thyroid function. Clin Endocrinol (Oxf). 2010;73: Svare A, Nilsen TI, Bjøro T, Asvold BO, Langhammer A. Serum TSH related to measures of body mass: longitudinal data from the HUNT Study, Norway. Clin Endocrinol (Oxf). 2011;74: Karmisholt J, Andersen S, Laurberg P. Variation in thyroid function in subclinical hypothyroidism: importance of clinical follow-up and therapy. Eur J Endocrinol. 2011;164: Bland JM, Altman DG. Regression towards the mean. BMJ. 1994; 308:1499.