Introduction

Subclinical thyroid dysfunction is an abnormal level of thyroid-stimulating hormone (TSH) in serum with normal free thyroxine (fT4) and triiodothyronine (fT3) levels [1]. Subclinical hyperthyroidism and subclinical hypothyroidism are more common than overt dysfunction, particularly in older adults. The prevalence rates of subclinical hyperthyroidism and subclinical hypothyroidism have been reported to be 0.7% and 4.3% to 8.5%, respectively [2, 3], and are higher in women than in men [4].

Clinical thyroid dysfunction affects the bone remodeling cycle and adult bone maintenance. Both increased and decreased bone turnover increase the risk of bone fracture. Excess thyroid hormone induces high bone turnover. Overt hyperthyroidism and subclinical hyperthyroidism are associated with high bone turnover and loss of bone density, which have been linked to osteoporosis and a high risk of fractures [5]. In overt hypothyroidism, lower bone turnover also increases the risk of bone fractures [6, 7]. However, the association between subclinical hypothyroidism and loss of bone density on the one hand and risk of fracture on the other is unclear. Some studies show an association between subclinical hypothyroidism and bone loss [8], but others report no association with bone loss or fracture risk [9, 10]. Some reports show no association between any subclinical thyroid dysfunction and bone loss or fracture risk [11, 12].

Serum TSH levels increase with age, and subclinical thyroid dysfunction is common in elderly adults [13, 14]. The effects of subclinical thyroid dysfunction on bone status occur largely in older adults. Although previous studies have considered subclinical thyroid dysfunction and bone mineral density (BMD) in older adults, few studies have considered middle-aged adults. In the present study, we use data from a large cohort of healthy middle-aged adults to determine whether subclinical thyroid dysfunction affects BMD or is a risk factor for osteoporosis.

Materials and methods

Study population

The Kangbuk Samsung Health Study is a cohort study of all adults who underwent a comprehensive annual or biennial examination at the Kangbuk Samsung Hospital Total Healthcare Centers in Seoul and Suwon, South Korea [15]. A total of 59,827 subjects in the cohort underwent a BMD test as part of their comprehensive health examination between January 2011 and December 2016.

For the present study, 34,317 subjects were excluded for the following reasons: lack of BMD measurement with a T-score (n = 23,444) and serum level for TSH and fT4 (n = 59); overt hypothyroidism or overt hyperthyroidism, with an fT4 level outside the 0.84–1.74 ng/dL range (n = 966); a history of heart disease, stroke, cancer, diabetes mellitus, chronic renal failure, osteoarthritis, or thyroid disease; or use of medication for osteoporosis, including calcium or a steroid (n = 9848). For those who visited our institution more than once, we used data from their most recent visit. Therefore, the present analysis included 25,510 subjects (15,761 women and 9749 men). Among them, 23,124 subjects (91%) visited our institution only once, and the rest visited at least twice, up to six times. This study was approved by the Institutional Review Board of Kangbuk Samsung Hospital (IRB No. 2018-05-062), and the requirement for informed consent was waived because this was deemed a de-identified retrospective study involving data routinely collected during the health screening process.

Data collection

At each comprehensive health examination, we collected data on demographic characteristics, smoking status, alcohol intake, physical activity, medical history, and medication use through standardized, self-administered questionnaires. Smoking status was categorized as never, past smoker, or current smoker. Alcohol consumption was categorized as ≤ 20 g/day or > 20 g/day. Physical activity was based on the metabolic equivalent of task-minutes/week (MET-min/wk) and categorized as low (< 600 MET-min/wk), moderate (600–2999 MET-min/wk), or high (≥ 3000 MET-min/wk). Blood pressure and anthropometry items were measured by trained registered nurses as part of a health check-up program. Body mass index (BMI) was calculated as weight (kg) divided by the square of height (m2), and obesity was defined as a BMI ≥ 25 kg/m2, which is the proposed threshold for Asian populations [16].

Blood samples were obtained in the morning after 10 h of fasting. Serum fT3, fT4, and TSH levels were measured by an automated electrochemiluminescent immunoassay on a Cobas 8000 e602 analyzer (Roche Diagnostics, Tokyo, Japan), and lipid profile, ferritin, and homocysteine levels were measured using the Cobas 8000 c702 chemistry analyzer (Roche Diagnostics). Subclinical hyperthyroidism was defined as a TSH level < 0.27 μIU/mL, and subclinical hypothyroidism was defined as TSH > 4.2 μIU/mL, both with an fT4 level within the normal range (0.84–1.74 ng/dL). We defined euthyroid status based on the normal TSH reference range of 0.27–4.20 μIU/mL. The Laboratory Medicine Department has been accredited by the Korean Society of Laboratory Medicine and participates annually in inspections and surveys by the Korean Association of Quality Assurance for Clinical Laboratories and the College of American Pathologists Proficiency Testing program.

BMD measurement

BMD was measured by dual-energy X-ray absorptiometry (Lunar Prodigy; GE, Madison, WI, USA), and the results are expressed in g/cm2. Calibration with a machine-specific phantom was performed each day prior to the examination of patient samples. Each subject’s lumbar spine, femur neck (FN), and total proximal femur (FT) were measured. Lumbar spine BMD values were calculated as the means of four measured values for L1–L4. Following World Health Organization guidelines [17], postmenopausal women were divided into normal, osteopenia (those with a T-score from − 2.4 to − 1.0) and osteoporosis (those with a T-score ≤ − 2.5) subgroups according to their lumbar spine, femoral-neck, or total femoral T-scores. Osteoporosis cannot be diagnosed in men under age 50 on the basis of BMD alone [18]. So, we performed multiple logistic regression analysis of 4275 men older than 50 years. They were divided into lower BMD groups according to T-scores (− 2.5 or less, and − 2.4 to less than − 1.0), which was the only numerical information available for either group. Each T-score was calculated based on the reference lumbar spine BMD of young Korean adults.

Statistical analysis

Continuous variables are presented as the mean ± standard deviation or the median, and categorical variables are presented as numbers and percentages. Comparisons of variables among subjects with subclinical hypothyroidism, subclinical hyperthyroidism, and euthyroid status were conducted using an analysis of variance (ANOVA) or χ2 test. Tukey’s post-hoc test was performed for all multiple comparisons by ANOVA. Adjustments for age and other variables were performed with analysis of covariance. Stepwise multiple regression analyses were performed with lumbar spine BMD as the dependent variable. Multiple logistic regression analyses were performed with osteoporosis and osteopenia as dependent variables and subclinical hyperthyroidism and hypothyroidism as independent variables adjusted for age and BMI. We estimated adjusted odds ratios (ORs) with 95% confidence intervals (CIs) as endpoints. P values of 0.05 or less were considered statistically significant. All statistical analyses were done using SPSS version 24 software (IBM Inc., Chicago, IL, USA).

Results

Characteristics of the subjects

Baseline characteristics of the subjects are presented in Table 1. The mean age was 44.7 ± 10.6 years for women and 48.0 ± 11.3 years for men. Mean serum levels of TSH and fT4 were 2.3 ± 1.6 μIU/mL and 1.2 ± 0.1 ng/dL, respectively, for women and 2.0 ± 1.5 μIU/mL and 1.3 ± 0.2 ng/dL, respectively, for men, with no significant differences between female and male subjects. The frequencies of subclinical hyperthyroidism and hypothyroidism were 0.6% and 8.4% for women, 0.5% and 5.7% for men, respectively.

Table 1 Characteristics of the study population
Full size table

For both women and men, subjects with subclinical hyperthyroidism and hypothyroidism had significant differences in characteristics compared with euthyroid subjects. Women with subclinical hyperthyroidism had higher fT4 and lower total cholesterol and low-density lipoprotein (LDL) levels than euthyroid women. Meanwhile, men with subclinical hyperthyroidism had higher fT4 and calcium levels and frequently achieved more than 3000 MET-min/wk than euthyroid men. On the other hand, women with subclinical hypothyroidism were older and obese, more likely to be postmenopausal, and had lower fT4 and LDL but higher total cholesterol, triglycerides, LDL, and ferritin levels than euthyroid women. Men with subclinical hypothyroidism were older, less likely to consume > 20 g/day of alcohol or be a current smoker, and had lower fT4 levels than euthyroid men. Serum phosphate, C-reactive protein, and homocysteine levels did not differ among the three groups (subclinical hyperthyroidism, euthyroid, and subclinical hypothyroidism).

Subclinical thyroid dysfunction and BMD

The adjusted means of the lumbar spine BMD and T-scores are presented in Table 2. Neither the unadjusted BMD nor the T-score differed significantly among the subclinical hyperthyroid, subclinical hypothyroid, and euthyroid groups. The BMD and T-scores of women with subclinical hypothyroidism were statistically higher than those of euthyroid women after adjustment for age, BMI, and postmenopausal status, both including and not including serum total cholesterol, triglycerides, high-density lipoprotein (HDL), LDL, and ferritin. No association was found between the adjusted BMD and T-scores and thyroid status for men.

Table 2 Lumbar spine BMD and T-scores by adjusted mean (standard error) of the subclinical hyperthyroidism, euthyroid, and subclinical hypothyroidism groups
Full size table

Table 3 shows the results of the stepwise multiple regression analyses between lumbar spine BMD and clinical variables. Age and postmenopausal status showed significant negative correlations for women, and smoking showed a significant negative correlation for men. The variables of BMI and MET-min/wk ≥ 3000 showed a significant positive correlation in both women and men. In terms of standardized beta (β) values, postmenopausal status showed the strongest relationship in women. BMI contributed the second strongest relationship in women and the strongest relationship in men. Age contributed the third strongest relationship in women, but it was removed from the stepwise analysis in men.

Table 3 Assessment of clinical variables related to spine BMD using stepwise multiple regression analysis
Full size table

Among those who visited our institute more than twice during the study period, 2082 subjects (1445 women and 637 men) showed persistent euthyroid status, and 88 subjects (54 women and 34 men) showed persistent subclinical hypothyroid status. Neither the mean lumbar spine BMD nor the T-scores differed between those two groups (euthyroid versus subclinical hypothyroid): 1.140 g/cm2 vs. 1.119 g/cm2 (P = 0.324) and − 0.08 vs. − 0.25 (P = 0.326), respectively, in women; 1.185 g/cm2 vs. 1.183 g/cm2 (P = 0.946) and 0.04 vs. 0.02 (P = 0.927), respectively, in men. Table 4 shows the percent changes in lumbar spine BMD from the first to last visits. The median intervals were 1.95 years in persistent euthyroid group and 2.01 years in persistent subclinical hypothyroid group. BMD had decreased at the last visit in both groups. The persistent subclinical hypothyroidism group exhibited greater BMD reduction than the persistent euthyroid group, but this finding was not statistically significant.

Table 4 Percent changes in lumbar spine BMD between persistent euthyroid and persistent subclinical hypothyroidism groups
Full size table

Subclinical thyroid dysfunction and osteoporosis

Of our study subjects, 4710 women were postmenopausal and 4275 men were above the age of 50 years. We conducted further analyses to confirm the association between osteoporosis and subclinical thyroid dysfunction in these subjects. Both postmenopausal women and men age 50 and older showed no difference in BMD according to subclinical thyroid dysfunction. As shown in Table 5, subclinical hypothyroidism was negatively associated with the risk of osteoporosis (T-score ≤ − 2.5) in postmenopausal women. The OR (95% CI) of the subclinical hypothyroidism group was 0.657 (0.464–0.930). Subclinical hypothyroidism was not significantly associated with osteopenia status (T-score from − 2.4 to − 1.0). Among men older than 50, the low BMD (T-score ≤ − 2.5 or T-score from − 2.4 to − 1.0) groups were not significantly associated with subclinical thyroid status.

Table 5 Multiple logistic regression of 4710 postmenopausal women and 4275 men older than 50 years
Full size table

Discussion

We undertook this study to determine whether subclinical thyroid dysfunction influences BMD or is a risk factor for osteoporosis. To the best of our knowledge, we enrolled the second largest number of subjects among reports about subclinical thyroid dysfunction and BMD [19], and we enrolled the largest number of postmenopausal women. Our main finding in this study of healthy women and men is that no apparent association exists between subclinical hypothyroidism or subclinical hyperthyroidism and BMD, although age and postmenopausal status can affect BMD in women, and BMI can affect BMD in both women and men. We contend that it is another piece of evidence that subclinical thyroid dysfunction and BMD are not associated in healthy middle-aged adults. We also found that subclinical hypothyroidism in postmenopausal women is independently associated with a lower risk of osteoporosis and that subclinical hyperthyroidism does not affect osteoporosis in women or men.

Some previous studies suggested a relationship between subclinical thyroid dysfunction or TSH level and BMD. The risk of fractures increases with subclinical hyperthyroidism [20, 21], as does the risk of osteoporosis [7]. The upper-normal TSH group in a previous study had a decreased risk of osteoporosis, and the frequency of osteoporosis decreased inversely with TSH increases [22]. A high risk of osteoporosis was reported in patients with low TSH levels or subclinical hyperthyroidism [23, 24]. Also, subclinical thyroid dysfunction in perimenopausal women [8] and low-normal TSH (0.4–1.2 mIU/L) in men [25] have been found to be associated with low BMD at the lumbar spine or FN. Also, another similar study performed in Korean individuals [19] reported that variations in TSH levels within the euthyroid and subclinical range are positively correlated with BMD at the lumbar spine, FN, and total hip, and a negative relationship between TSH and fT4, whereas total T3 and TSH correlated positively in women and did not correlate in men. Their study participants were older with an average age of 50.1 ± 8.3 in women and 58.2 ± 9.2 in men than in our study (44.7 ± 10.6 years for women and 48.0 ± 11.3 years for men). In particular, since the women in the subclinical hyperthyroid group were youngest in our study (43.8 ± 11.1 years), we thought that they showed lower BMD results of 1117 mg/cm2 than our study (1162 mg/cm2). They used a TSH reference range with 0.51–5.00 μIU/mL, which was a higher range than used in our study (0.27–4.20 μIU/mL).

However, other reports including ours have shown no association, so controversy remains. The risk of fractures increases with subclinical hyperthyroidism [20, 21], as does the risk of osteoporosis [7]. The upper-normal TSH group in a previous study had a decreased risk of osteoporosis, and the frequency of osteoporosis decreased inversely with TSH increases [22]. A high risk of osteoporosis [23, 24] was reported in patients with low TSH levels or subclinical hyperthyroidism. Also, subclinical thyroid dysfunction in perimenopausal women [8] and low-normal TSH (0.4–1.2 mU/L) in men [25] have been found to be associated with low BMD at the lumbar spine or FN. Conversely, TSH was not associated with statistically significant changes in BMD in a previous study of postmenopausal women, but TSH levels above the reference range were associated with a 35% reduced risk of non-vertebral fractures [26]. In middle-aged women, the effect of abnormal TSH on BMD changes or risk of fracture may be small. Saler et al. [27] suggested that subclinical hyperthyroidism does not lead to loss of BMD in premenopausal women (mean, 33–34 years old). Abrahamsen et al. reported that a group of subjects aged 18–49 years showed no association between fracture risk and low TSH [28] or high TSH [29], which differed from a group aged 50+ years. Garin et al. [9] also found no association between subclinical thyroid dysfunction and BMD at the lumbar spine, total hip, or FN in elderly women and men older than 65 years. In our study, adjusted BMD values did not differ in the lumbar spine, FN, or FT among subjects with subclinical hyperthyroidism and subclinical hypothyroidism. Adjusted lumbar spine BMD in women with subclinical hypothyroidism was statistically higher than in euthyroid women, but the absolute difference was only 0.01 g/cm2. Subclinical hypothyroidism was associated with a decreased risk of osteoporosis (OR, 0.657) only in postmenopausal women. However, we performed subgroup analysis among postmenopausal women, and neither BMD nor T-score differed between subjects with subclinical thyroid dysfunction and those with euthyroid status, although postmenopausal women from our study had a high frequency (57%) of osteopenia or osteoporosis.

We applied the reference range of 0.27 to 4.20 μIU/mL suggested by the manufacturer (Roche Diagnostics), although subclinical hypothyroidism is commonly defined using a TSH level of 0.45 μIU/mL [30]. It was lower than that of 0.62 and 6.86 μIU/mL range [31], or 0.59 to 7.03 μIU/mL range [14]. TSH levels in a Korean reference population were higher than those reported in other countries, presumably because of excess intake of dietary iodine in Korea [13]. In our data, when the lower limit of the TSH reference range was changed from 0.27 to 0.45 μIU/mL, the frequency of subclinical hyperthyroidism increased from 0.6 to 1.5%, but there were no significant differences in BMD among three groups of men and women, and the OR for subclinical hyperthyroidism for osteoporosis remained nonsignificant (data not shown).

Previous studies found that subclinical hypothyroidism is associated with lipid abnormalities, especially increased total cholesterol and LDL, whereas there are conflicting data on HDL cholesterol [32, 33]. Also, levels of ferritin differed in overt hypothyroidism (43.7 ± 52.6 ng/mL), subclinical hypothyroidism (42.1 ± 47.4 ng/mL), and the control group (48.1 ± 58.8 ng/mL), but those differences were not statistically meaningful (P = 0.623) [34]. Moreover, subclinical hypothyroidism (12.77 ± 8.63 mmol/L) produced slightly higher homocysteine levels compared with their control group (8.29 ± 1.35 mmol/L) [35]. In our study, we found higher levels of total cholesterol, triglycerides, LDL, and ferritin in women with subclinical hypothyroidism. However, those differences were not observed in men except for hypertriglyceridemia. Homocysteine level did not differ between groups.

The present study has several limitations. Our study population contained mainly middle-aged women and men in Korea. Therefore, our results might not apply to other age groups or populations from other countries. Selection bias could be present because we enrolled individuals whose spine BMD and T-scores were measured. Not all visitors to our health promotion center receive BMD measurements. The visitors whose data we used wanted to know their BMD and chose programs that included BMD study. Therefore, our cohort might not represent the general middle-aged population of Korea. The present study was a cross-sectional study, and was not designed as long-term prospective study. Therefore, the cumulative skeletal effects of subclinical thyroid dysfunction cannot be affirmed using our data. In our study, most subjects visited our institution only once, so we defined subclinical hypothyroidism or hyperthyroidism on a single measurement of TSH, similar to other studies based on single TSH values [19, 24, 25]. Like Abrahamsen et al. [28, 29], our study warrants follow-up observations of how long-term exposure to asymptomatic subclinical thyroid dysfunction might affect osteoporosis or fracture.

In conclusion, this study found no apparent association between subclinical hypothyroidism or subclinical hyperthyroidism defined based on a single TSH measurement and BMD at the lumbar spine or femur in healthy middle-aged adults. Subclinical hypothyroidism was independently associated with a lower risk of osteoporosis in postmenopausal women.