Association between subclinical thyroid dysfunction and change in bone mineral density—an individual participant data analysis of prospective cohorts.

Segna D, Bauer D, Aubert CE, Collet TH, Da Costa B, Fischer K, Peeters R, Fink HA, Cappola AR, Blum MR, van Dorland HA, Robbins J, Naylor K, Eastell R, Uitterlinden AG, Ramirez FR, Gogakos A, Gussekloo J, Williams GR, Schwartz A, Cauley JA, Aujesky D, Bischoff-Ferrari HA, Rodondi N. Association between subclinical thyroid dysfunction and change in bone mineral density—an individual participant data analysis of prospective cohorts. Poster session presented at: Society of General Internal Medicine Annual Meeting; 2016 May 12; Hollywood, FL.

Background: Subclinical hyperthyroidism (SHyper) is associated with an increased risk for hip and other fractures, but the mechanism is not established. Previous population-based cohort studies found an average annualized bone loss of 0.2-0.6% at the femoral neck or total hip in the general population with even higher rates in the elderly. Overt hyperthyroidism decreases bone mineral density (BMD), but data on the association between subclinical thyroid dysfunction (SCTD) and bone loss are controversial. Considering the heterogeneity of current literature, we aimed to clarify whether SCTD was associated with bone loss by pooling individual participant data from population-based prospective cohort studies. Methods: Two independent reviewers conducted a systematic literature search in MEDLINE/ EMBASE (from inception to May 2015) without language restrictions identifying prospective cohorts with baseline thyroid status and serial BMD measurements. We collected individual participant data by contacting the authors of each cohort. We classified thyroid status into euthyroidism (Thyroid Stimulating Hormone [TSH] 0.45-4.49 mIU/L), SHyper (TSH < 0.45 mIU/L) and subclinical hypothyroidism (SHypo, TSH 4.50-19.99 mIU/L), both with normal free thyroxine concentrations. For our primary outcome of bone loss, defined as annualized percentage change in BMD (% BMD), we pooled location-specific data on serial dual x-ray absorptiometry scans of the femoral neck, total hip and lumbar spine. For the main analysis, we excluded individuals taking bone-active medications at baseline, and conducted two sensitivity analyses: 1) additional exclusion of patients on bone-influencing drugs during follow-up; and 2) excluding patients on bone- or thyroid-influencing medications at any time point. We used established methods with a 2-step approach, first analyzing the association in each cohort of SCTD with bone loss using multivariable linear regression models controlling for age, gender, body mass index, smoking, diabetes mellitus and menopausal status, and in a second step calculating pooled estimates using random-effects models. We used I2 indices for assessing heterogeneity across cohorts. Results: In our individual data analysis of 6 prospective cohorts (from the United States, United Kingdom, Germany, France and the Netherlands) including 7636 participants (median age: 72 years, 56.9% women), the prevalence was 5.9% for SHyper and 7.8% for SHypo at baseline. During follow-up, pooled mean % BMD was -0.59 (95% confidence interval [CI]: -0.63,-0.54) at the femoral neck, -0.55 (CI: -0.61,-0.49) at the total hip and 0.32 (CI:-0.21,0.84) at the lumbar spine. Relative to euthyroidism, SHyper was associated with greater annualized bone loss at the femoral neck (pooled multivariable % BMD = -0.18; CI:-0.34,-0.02; I2 = 0.0%), with a similar pattern at the total hip (% BMD = -0.14; CI:-0.38,0.10; I2 = 52.7%), but not at the lumbar spine (% BMD = 0.03; CI:-0.30,0.36; I2 = 24.8%). In contrast, SHypo was not associated with changes in % BMD at the femoral neck (0.00; CI:-0.12,0.13; I2 = 0.0%), total hip (% BMD = 0.02; CI:-0.08,0.12; I2 = 0.0%), or lumbar spine (% BMD = -0.01; CI:-0.34,0.32, I2 = 37.7%). In stratified analyses (Table), there was a significant trend for increased bone loss at the femoral neck associated with lower TSH subgroups (p for trend = 0.03), especially in the subgroup with TSH < 0.1 mIU/l (% BMD = -0.58; CI:-0.96,-0.20), and a comparable, but non-significant trend at the total hip (p for trend = 0.11). Bone loss in SHyper relative to euthyroidism was greater among patients not on any thyroid- and bone-influencing medication at any time during follow-up, bone loss at the femoral neck (% BMD = -0.36; CI:-0.71,-0.00; I2 = 45.9%) and total hip (% BMD = -0.40; CI:-0.96,0.16; I2 = 81.9%) was increased, but SHypo remained not associated with bone loss at any BMD site. Conclusions: Compared to euthyroid individuals, hip bone loss was increased in those with SHyper, which may account for the observed increase in fracture risk, whereas SHypo was not associated with bone loss at any measured body site. These results reinforce the importance of fracture risk assessment and initiation of preventive strategies among older adults with SHyper. Randomized trials are needed to confirm that treatment of SHyper prevents bone loss and reduces fractures.