The relationship between the thyroid and the skeleton was first suggested in 1891 when Von Recklinghausen (1891) reported a patient with hyperthyroidism and multiple fractures. Thyroid disease is common; more than 1% of British women receive thyroxine treatment, increasing to over 3% in those over 50 years of age, and the concern is that patients with a history of hyperthyroidism or those receiving thyroxine treatment may be at increased risk of osteoporosis and fractures (Parle et al., 1993). While the clinical consequences of overt hyperthyroidism on bone have been known for many years the molecular mechanism of action of thyroid hormone on bone has remained incompletely understood. Furthermore, earlier recognition of thyroid disease and effective treatment for clinical symptoms has meant that severe bone disease is now uncommonly seen. It is only in more recent years, with the development of more sensitive TSH assays and the recognition that even so-called subclinical thyroid disease, associated with a suppressed TSH but normal FT4 and FT3 levels, may have effects on fracture risk that there has been a renewed interest in the role of thyroid hormone in bone (Klee & Hay, 1988; Bauer et al., 2001). The molecular actions of thyroid hormone in bone have recently been comprehensively reviewed (Harvey et al., 2002; Bassett & Williams, 2003). This review will therefore only briefly summarize this information and instead concentrate in detail on some of the clinical aspects of thyroid hormone effects on bone growth and development in childhood and bone turnover and maintenance in adults. Tables are used to illustrate the effects of thyroid hormone suppressive therapy for thyroid cancer and thyroid hormone replacement on bone mineral density (BMD), and the effects of thyroid hormone on fracture risk. Each clinical section concludes with a summary of the salient points discussed. The role of calcitonin, a hormone also synthesized and secreted by the thyroid gland, and its effects on bone has recently been reviewed and will not be discussed further here (Inzerillo et al., 2002). In adults, the mechanical integrity of the skeleton is maintained by bone remodelling, which results from the localized removal of old bone and its replacement with newly formed bone (Ng et al., 1997). Remodelling enables bone to adapt to mechanical stress and to repair micro-damage, thus maintaining its strength. Alterations in bone remodelling are responsible for most metabolic bone diseases, including osteoporosis and the reduction in BMD seen with uncontrolled hyperthyroidism. Osteoporosis is defined as a BMD value 2·5SDs or more below the mean value for young adults (T score ≤−2·5) (Anonymous, 1991). As well as low bone mass, osteoporosis is characterized by microarchitectural deterioration of bone, with a consequent increase in bone fragility. The major end point of osteoporosis is fracture, especially of the distal forearm, vertebrae or hip, with a two- to threefold increase in fracture incidence for each SD reduction in BMD. It is not clear what initiates remodelling at a particular site on bone but it may be in response to paracrine factors secreted by osteocytes when subjected to mechanical stimuli. During the initial 'activation' phase, interactions between osteoblast (bone-forming) and osteoclast (bone-resorbing) precursors leads to the differentiation, migration and fusion of large multinucleated osteoclasts. Activated osteoclasts adhere closely to the mineralized bone surface. They then secrete hydrogen ions through a proton pump system, and digestive lysosomal enzymes that dissolve the protein matrix of bone at a low pH. Once resorption is complete, osteoclasts undergo programmed cell death (apoptosis) during the 'reversal' phase. Once a certain resorption depth is achieved, the final 'formation' phase begins. Osteoblasts invade the area, lay down new matrix and begin mineralization. The activities of the two cell types, osteoclasts and osteoblasts, are thus tightly coupled. The activation–resorption–formation sequence normally lasts about 200 days. However, when there is either enhanced osteoclastic activity or decreased osteoblastic activity each resorption cavity is only partially filled, resulting in net bone loss. Reconstruction of the bone remodelling sequence in uncontrolled hyperthyroid patients indicates that the duration of remodelling is significantly reduced with a negative balance between resorption and formation, resulting in a net loss of about 10% of mineralized bone per cycle, with ultimately a reduction in BMD and a predisposition to fractures (Eriksen et al., 1985). Calcium kinetic studies in hyperthyroid patients have shown that intestinal calcium and phosphate absorption is reduced, while urinary, faecal and dermal calcium excretion is increased (Mosekilde et al., 1990). Thus thyrotoxicosis is a condition of negative calcium balance and this negative balance may also contribute to the increased fracture risk. In addition, levels of biochemical markers of both bone formation [alkaline phosphatase, osteocalcin, carboxyterminal propeptide of type 1 collagen (P1NP)] and bone resorption [pyridinoline and deoxypyridinoline collagen cross-links, hydroxyproline, carboxyterminal cross-linked telopeptide of type 1 collagen (CTX)] are elevated in thyrotoxicosis and generally correlate with thyroid hormone concentrations (Allain & McGregor, 1993; Miyakawa et al., 1996; Nagasaka et al., 1997). The thyroid gland synthesizes and secretes the prohormone 3,5,3′,5′-l-tetraiodothyronine (thyroxine, T4) together with a small amount of the active hormone 3,5,3′-l-triiodothyronine (T3) in response to TSH stimulation. However, most circulating T3 is generated by the iodothyronine deiodinase enzymes D1 and D2, which convert T4 to T3 by 5′ monodeiodination. A D3 deiodinase enzyme is responsible for monodeiodination of T4 and T3 at the 5 position resulting in generation of the inactive metabolites reverse T3 (rT3) and rT2. It is the relative activities of these three deiodinases, D1, D2 and D3, that determines the intracellular level of the physiologically active hormone, T3 (Bianco et al., 2002). The cellular actions of T3 are mediated by nuclear T3 receptors, which are members of the steroid receptor superfamily and function as T3 inducible transcription factors that regulate the expression of hormone-sensitive target genes (Lazar, 1993). The T3 receptors (TRs) α and β are encoded by two separate genes, THRA and THRB, located on chromosomes 17 and 3 respectively in humans. At physiological levels, T4 is inactive because it possesses 100-fold lower affinity than T3 for binding to the TR and does not enter the cell nucleus at high enough concentrations to occupy the ligand binding site of the TR (Bianco et al., 2002). Within the nucleus, TRs bind as heterodimers with the retinoid X receptor (RXR) to specific hormone response elements located in the promoter regions of target genes (Zhang & Lazar, 2000; Fig. 1). In the absence of T3 unliganded RXR-TR heterodimers mediate transcriptional repression. Binding of T3 leads to the dissociation of co-repressors and the binding of co-activators and thyroid receptor-associated proteins, thereby promoting active thyroid responsive gene transcription. Mechanism of action of thyroid hormone. The prohormone T4 is converted to active T3 by type 1 (D1) or type 2 (D2) deiodinase. Type 3 (D3) deiodinase inactivates T4 and T3. Thyroid hormone receptors (TR) preferentially form heterodimers with retinoid X receptors (RXR), but can also form homodimers. Unliganded RXR-TR heterodimers bind thyroid-response elements (TREs) and mediate transcriptional repression via a co-repressor complex (nuclear receptor co-repressor (NCoR) or silencing mediator of retinoic acid receptor and TR (SMRT) proteins), which possess histone deacetylase activity. Binding of T3 leads to the dissociation of the co-repressor complex and recruitment of a co-activator complex [steroid receptor co-activators (p160/SRCs)], possessing histone deacetylase activity. This allows chromatin remodelling. In addition, further thyroid receptor-associated proteins (TRAPs) are recruited, promoting active gene transcription. Both the THRA and THRB genes are transcribed as multiple RNA isoforms (for details of the molecular mechanism of TR action see Lazar, 1993, Harvey et al., 2002 and O'Shea & Williams, 2002). Functional TRα and TRβ proteins have been shown to be expressed in bone marrow stromal cells, osteoblasts and bone growth plate chondrocytes both in vitro and in vivo (Robson et al., 2000; Stevens et al., 2000). Data from cell and organ cultures have identified roles for T3 in the production of cytokines, growth factors, markers of bone turnover and proangiogenic factors involved in bone development and growth, bone turnover and bone maintenance (Milne et al., 1998, 2001; Siddiqi et al., 1998; Kim et al., 1999; Pereira et al., 1999; Salto et al., 2001; Stevens et al., 2003). Analysis of information gained from genetically modified mice has recently been summarized (O'Shea & Williams, 2002; Bassett et al., 2003). TRα is expressed at about 10-fold higher levels than TRβ in bone and is functionally predominant in the skeleton, though TRβ also plays a role (O'Shea et al., 2003; Stevens et al., 2003). TRα(0/0) mice, which lack all TRα isoforms but are biochemically euthyroid have growth retardation with delayed bony ossification and impaired mineralization, features that are very similar to the skeletal consequences of childhood hypothyroidism (Gauthier et al., 2001). TRβ(–/–) mice, which lack all TRβ isoforms and display thyroid hormone resistance are reported to have normal growth and no skeletal abnormalities have been identified to date (Forrest et al., 1996; Gauthier et al., 1999). Nevertheless, TRα(0/0)TRβ(–/–) have a more severe phenotype than TRα(0/0) mice indicating the ability of TRβ to partially compensate for the lack of TRα in bone (Gauthier et al., 1999). Classical thinking has always attributed the reduction in BMD associated with overt thyrotoxicosis to the high circulating levels of thyroid hormone. The concomitant suppressed TSH levels had not been considered to have any direct physiological effects on bone. Data from a recently published paper have challenged this viewpoint by suggesting a direct role for TSH in the maintenance of bone mass. A TSH receptor (TSHR–/–) knockout mouse model, in which thyroid hormone replacement was given from weaning, was generated (Abe et al., 2003). These mice have high bone turnover osteoporosis, suggesting that TSH is a negative regulator of bone turnover. This is the first evidence to suggest that TSH may have a direct effect on the skeleton but requires further exploration before it can be related to human physiology. Thus, it is unclear at this stage whether bone loss in hyperthyroidism results from thyroid hormone excess, TSH deficiency or a combination of both. In this context the role of TSH receptor stimulating antibodies in Graves' thyrotoxicosis needs to be explained. Paradoxically, it would be predicted that TSH receptor stimulation by such antibodies would mitigate against bone loss, whereas in fact, patients with Graves' disease are more at risk of bone loss and osteoporosis. The clinical effects of childhood thyroid dysfunction are well documented. Untreated congenital hypothyroidism is characterized by growth arrest, epiphyseal dysgenesis, delayed bone age and short stature (Bucher et al., 1985; Virtanen, 1988; Leger & Czernichow, 1989; Virtanen & Perheentupa, 1989; Newland et al., 1991; Chiesa et al., 1994). Early treatment with thyroxine replacement increases growth velocity, allowing children to reach their predicted adult height (Bucher et al., 1985; Chiesa et al., 1994; Dickerman & De Vries, 1997; Salerno et al., 2001). Follow-up of children with congenital hypothyroidism treated with thyroxine replacement for a mean of 8·5 years, indicates that they have normal BMD, as measured by dual-energy X-ray absorptiometry (DXA), when compared with controls (Kooh et al., 1996; Leger et al., 1997). However, adult women treated since the neonatal period for congenital hypothyroidism have been shown to have a 10% reduction in radial BMD (Demeester-Mirkine et al., 1990). No data exist on lifetime fracture risk for individuals with congenital hypothyroidism treated with long-term thyroxine replacement. Similarly, patients with juvenile acquired hypothyroidism have growth arrest, delayed bone maturation and short stature. Following thyroxine replacement, rapid catch-up growth occurs but these individuals may fail to achieve final predicted height and the resulting permanent height deficit is related to the duration of thyroid hormone deficiency prior to replacement (Rivkees et al., 1988). Conversely, childhood thyrotoxicosis results in accelerated growth, advanced bone age and short stature, resulting from premature closure of the epiphyseal growth plates. In severe cases, premature closure of the cranial sutures results in craniosynostosis (Segni & Gorman, 2001). Resistance to thyroid hormone (RTH) is an autosomal dominant condition of impaired tissue responsiveness to thyroid hormone caused by dominant negative mutations of the thyroid hormone receptor β (Weiss & Refetoff, 1996). The clinical presentation and skeletal phenotype of RTH is highly variable, although few patients have been studied in detail. The syndrome results in a complex mixed hyper- and hypothyroid phenotype depending on the target tissue studied. Thus an individual patient can have symptoms of both thyroid hormone deficiency and excess. Skeletal features including short stature, advanced bone age, delayed bone age, increased bone turnover, osteoporosis, fractures, craniofacial abnormalities and craniosynostosis have all been recorded. The clinical variability of this condition is thought to be related to the precise properties of the mutant TRβ and the genetic heterogeneity of the cofactors (co-activators and co-repressors) that modulate the receptor-mediated action of thyroid hormone (Weiss & Refetoff, 1999). Physiological concentrations of thyroid hormone are essential for normal growth and skeletal development. Thyroid hormones influence the formation of long bones (endochondral ossification) and flat bones (intramembranous ossification). The remainder of this review will focus on the relationships between thyroid status, BMD and fracture risk. Current knowledge regarding the effects on BMD of TSH suppressive therapy, subclinical hyperthyroidism and thyroid hormone replacement is summarized. Much of the literature in this field is heterogenous and Tables 1–3 can only give an overview of findings. Factors such as careful case–control matching, dietary calcium intake, smoking, alcohol intake, physical activity, co-existent hypoparathyroidism and variable techniques in bone mass measurement are not always adequately addressed within and between studies. Nonetheless, it is useful to examine the available literature in order to establish consistent findings and identify those questions that still require an answer. Current practice for the treatment of well-differentiated thyroid cancer, including follicular and papillary carcinoma, is total or near total thyroidectomy with or without ablative radio-iodine therapy. Oral thyroxine, in doses sufficient to cause suppression of TSH, is then given to prevent the stimulating effects of TSH on tumour cell proliferation. The 10 year disease free survival for patients with well-differentiated thyroid cancer is over 80–90%, so that any deleterious effects of suppressive thyroxine treatment could potentially affect a significant number of survivors (Gilliland et al., 1997). The British Thyroid Association (2002) recommends that lifelong thyroxine should be given in a dose sufficient to suppress TSH to < 0·1 mU/l in a sensitive assay. For most patients this equates to a dose of 175–200 µg thyroxine daily. They also suggest that the consequences of supraphysiological thyroxine therapy, including osteoporosis, need monitoring, particularly as patients age. The American Association of Clinical Endocrinologists/American Association of Endocrine Surgeons (AACE/AAES 2001) Clinical Practice Guidelines for the Management of Thyroid Carcinoma recognize that long-term thyroxine suppressive therapy may have adverse effects on bone, including accelerated bone turnover and osteoporosis, but stop short of any definitive recommendations for monitoring. Instead they suggest the use of good clinical judgement and individualization of thyroxine therapy for patients with thyroid cancer, as for those individuals in lower risk prognostic groups a lesser degree of long-term TSH suppression may suffice. Table 1 summarizes current data on the effects of suppressive doses of thyroxine in patients treated for thyroid cancer on bone. A recent systematic review has considered 11 of the more stringent of these studies, in which information was provided on gender, menopausal status and control subjects, as well as BMD measured at various sites (Quan et al., 2002). Four of these trials were longitudinal in design, seven cross-sectional. All patients had been on long-term suppressive therapy for at least 5 years. Eight trials included premenopausal women and found little or no change in BMD as a consequence of long-term suppressive thyroxine (Diamond et al., 1991; Franklyn et al., 1992; Pioli et al., 1992; Giannini et al., 1994; Marcocci et al., 1994; Muller et al., 1995; Gorres et al., 1996; Jodar et al., 1998). Similarly, in the three cross-sectional studies that included men, no change in BMD was seen in men taking suppressive doses of thyroxine (Franklyn et al., 1992; Gorres et al., 1996; Marcocci et al., 1997). A small longitudinal study (eight patients) found a significant reduction in BMD at the lumbar spine at a rate of 2·6% per year from the initial commencement of thyroxine in premenopausal women (Pioli et al., 1992) while a single cross-sectional study found a reduction in BMD at the femoral neck, but not in the lumbar spine or forearm when compared with controls (Diamond et al., 1991). Nine studies including postmenopausal women were reviewed. Six of these studies, two longitudinal and four cross-sectional, showed no difference in BMD between patients with thyroid cancer treated with suppressive doses of thyroxine and controls (Franklyn et al., 1992; Giannini et al., 1994; Hawkins et al., 1994; Muller et al., 1995; Gorres et al., 1996; Guo et al., 1997). One study showed a small decrease in BMD at the distal radius only (Jodar et al., 1998). Two studies, both achieving complete suppression of TSH in all patients, have reported decreased BMD at all measured sites in postmenopausal women receiving thyroxine (Diamond et al., 1991; Kung et al., 1993; Kung & Yeung, 1996). Given this conflicting evidence, particularly from the two most carefully controlled trials in which rigorous attempts were made to exclude confounding variables, it is difficult to reach a definitive conclusion as to the contribution of suppressive TSH therapy to reduction in BMD in postmenopausal women. The review authors conclude that, based on current evidence, no significant change in BMD occurs as a result of suppressive thyroxine treatment in premenopausal women or men (Quan et al., 2002). The situation in postmenopausal women is less clear, and recommendations include: screening for other risk factors for osteoporosis, BMD testing with DXA if warranted, dietary and lifestyle advice, plus or minus bisphosphonate therapy. The remaining studies summarized in Table 2 also give conflicting results with three studies showing an effect of suppressive thyroxine on BMD in postmenopausal women (Lehmke et al., 1992; McDermott et al., 1995; Kung et al., 1996) and an equal number of studies showing no effect (Ribot et al., 1990; Florkowski et al., 1993; Rosen et al., 1998). Suppressive doses of thyroxine, appropriate for thyroid cancer treatment, have little or no effect on the BMD of premenopausal women or men. The situation in postmenopausal women is less clear and it is recommended that all other risk factors for osteoporosis be considered in these patients, in whom a lower threshold for investigation of BMD by DXA scanning should also be considered. Subclinical hyperthyroidism is defined as clinical euthyroidism in the context of a normal free T4 and T3 level, with a TSH below the normal reference interval, usually undetectable in a sensitive assay. The prevalence increases with age and although many patients have a multinodular goitre other differential diagnoses such as silent thyroiditis, early overt thyrotoxicosis, nonthyroidal illness, drugs or pituitary dysfunction need to be considered (Shrier & Burman, 2002). The precise pathophysiology and long-term outcome of subclinical hyperthyroidism is unknown and interpretation of the available literature is difficult due to the inclusion of patients with differing aetiologies. A small number of published studies have considered the effects of endogenous subclinical hyperthyroidism on bone. One prospective, nonrandomized study showed that postmenopausal women treated with radio-iodine for subclinical hyperthyroidism had an increase in BMD measurement of the hip and spine after 2 years follow-up, whereas similar untreated patients had a reduction of BMD (Faber et al., 1998). Similar findings have been reported in postmenopausal patients treated with anti-thyroid medication (Mudde et al., 1994). A larger cross-sectional study found lower BMD at the femoral neck and midshaft radius in postmenopausal women but not in premenopausal women (Foldes et al., 1993). In a subsequent study, BMD was again found to be unchanged in 15 premenopausal women with subclinical hyperthyroidism compared with controls (Gurlek & Gedik, 1999). Endogenous subclinical hyperthyroidism has no effect on the BMD of premenopausal women. Untreated postmenopausal women have an increased rate of bone loss compared with women who receive treatment. Few data exist on the influence of subclinical hyperthyroidism on the skeleton in men. Drawing conclusions from the literature evidence on the effects of thyroid hormone replacement on bone is also fraught with difficulty. Many of the studies examining the effects of exogenous thyroid hormones on BMD incorporate a mixed group of patients, including those with primary hypothyroidism, benign euthyroid goitre, post-thyroiditis and previous treated hyperthyroidism. All of these conditions could be considered to have confounding factors influencing bone turnover; patients with euthyroid goitre are by definition initially euthyroid and are then rendered subclinically hyperthyroid by thyroxine treatment in order to shrink their goitres, patients post-thyroiditis and with previous hyperthyroidism have had exposure to high circulating levels of thyroid hormone for undefined periods of time prior to their thyroxine replacement therapy and may have a period of catch-up with regard to their bone mass once thyroxine replacement is started (Mosekilde & Melsen, 1978). Patients with primary hypothyroidism may actually have a higher than normal bone mass prior to thyroxine therapy, which then falls once treatment is started. Thyroid hormone replacement therapy is usually defined as a dose of thyroxine that will maintain TSH within the normal reference interval (0·5–5·0 mU/l) as measured by a sensitive TSH assay. Not all patients, particularly those treated before the advent of highly sensitive TSH assays, are treated to euthyroidism, some are in fact over-replaced with lower TSH or higher thyroid hormone levels than control subjects. In Table 2 we summarize the available data using those papers where results were broken down to allow analysis of only those patients treated with thyroxine for either primary hypothyroidism or hypothyroidism post-thyroiditis, in whom the duration of the preceding hyperthyroid phase is likely to have been relatively brief. In the five studies that included only postmenopausal women, there was no significant effect of thyroid hormone replacement on bone (Grant et al., 1993; Ross, 1993; Franklyn et al., 1994; Hawkins et al., 1994; Guo et al., 1997). Two studies included only premenopausal women and found that thyroid hormone replacement had an effect on bone, although the sites affected (lumbar spine, femoral neck and Ward's triangle) differed between the studies (Kung & Pun, 1991; Garton et al., 1994). One small longitudinal study included men as well as pre- and postmenopausal women and found significant reductions in BMD at all sites measured after 1 year of treatment (Ribot et al., 1990). This study also included a cross-sectional, case–control arm in which no effect on vertebral BMD was seen in women who had been treated with thyroid hormone for at least 2 years, suggesting that the bone loss noted over the first year of treatment in the longitudinal study might be a transient phenomenon, reflecting the shortening of the bone remodelling cycle from an average of 700 days to 200 days. Indeed, histomorphometric studies on adult bone specimens from patients receiving thyroid hormone replacement for primary hypothyroidism, have confirmed an increase in bone resorption with loss of both trabecular and cortical bone for up to 6 months after treatment is initiated (Coindre et al., 1986). A second cross-sectional study that also included both pre- and postmenopausal women similarly found no detrimental effect of thyroid hormone replacement on BMD in patients with primary hypothyroidism (Hanna et al., 1998). A number of other studies, not included in Table 2 for reasons discussed above, deserve a mention. One study of postmenopausal women and another, from the same authors, of men, found no significant effects of thyroid hormone replacement on bone (Schneider et al., 1994, 1995). Others have found reductions in BMD at the spine (but not the hip) or the hip (but not the spine) in premenopausal women on replacement therapy (Paul et al., 1988; Greenspan et al., 1991). One study showed no significant differences in T-scores in either pre- or postmenopausal women taking thyroxine replacement compared with controls but did find that these women had lower quantitative ultrasound variables, a finding which was more pronounced in postmenopausal women (Hadji et al., 2000). Appropriate thyroid hormone replacement has no effect on the BMD of postmenopausal women. Overzealous replacement therapy may have a detrimental effect on the BMD of premenopausal women, but whether this is more pronounced at the hip or the spine is unclear from present studies. The major weakness of many of the studies examining the relationship between thyroid disease and/or use of thyroxine replacement and its effects on bone is that, chiefly due to constraints on sample size and duration of follow-up, the important clinical endpoint of fracture has not been considered. Although reduction in BMD has been shown to correlate well with fracture risk it is nonetheless a crude measure that provides little information on bone quality (Marshall et al., 1996). It is also important to remember that although BMD measurement can predict fracture risk, it is only one of a number of risk factors for fracture and cannot identify those individuals who will have a fracture (Grisso et al., 1991; Kelsey et al., 1992; Cummings et al., 1995). Published studies are heterogeneous with regard to data provided on aetiology of hyperthyroidism, thyroxine replacement (dosage and duration of therapy) and the influence of treatment modalities (antithyroid medication, surgery or radio-iodine). To date, no published study has adequately addressed the issue of endogenous subclinical hyperthyroidism and fracture risk. However, a number of recently published studies have begun to address specifically the issue of the impact of thyroid disease on fracture risk (Table 3). To summarize these, with regard to the use of exogenous thyroid hormone, five studies have shown no difference in fracture risk in those individuals treated with thyroid hormone compared to those not treated (Leese et al., 1992; Solomon et al., 1993; Cummings et al., 1995; Wejda et al., 1995; Bauer et al., 2001; Van Den Eeden et al., 2003). All bar one of these studies examined white, postmenopausal women only. One study has shown that thyroid hormone use predicts fracture neck of femur risk in men [age-adjusted odds ratio (AOR) 1·69, 95% confidence interval (CI) 1·12–2·56], but not women (AOR 1·03, 95% CI 0·92–1·16; Sheppard et al., 2002). A single study in the Asian population found an increased risk of hip fracture with the use of thyroid drugs in both men [relative risk (RR) 7·1, 95% CI 2·0–25·9) and women (RR 11·8, 95% CI 1·3–106·0; Lau et al., 2001). There are two ways of approaching the question whether or not there is an association between hyperthyroidism per se and fracture risk. Firstly, by examining whether a diagnosis of hyperthyroidism is in itself associated with an increased fracture risk or, secondly, looking at the issue from another angle, whether a fracture is associated with a history of hyperthyroidism (Table 3). There is significant population overlap in the small number of published studies, the majority of studies being in postmenopausal women with few including either premenopausal women or men. An increased fracture risk in subjects with a history of hyperthyroidism has been found in several studies in varying anatomical sites as follows; femur (Vestergaard & Mosekilde, 2002), femoral neck/hip (Cummings et al., 1995; Wejda et al., 1995; Vestergaard et al., 2002), vertebrae (Vestergaard et al., 2000a), forearm (Vestergaard et al., 2000a) or foot (Seeley et al., 1996). Only one published study was negative, with no difference in fracture (hip, vertebral or forearm) occurrence between women with hyperthyroidism and those without (Solomon et al., 1993). In this study, women with hyperthyroidism were, however, found to have their first fracture earlier in life than those without. None of the studies used vertebral X-rays to ascertain asymptomatic spinal fractures. In those studies where it was examined, thyroxine replacement itself was not associated with an increased risk of fracture following adjustment for a history of hyperthyroidism (Solomon et al., 1993; Cummings et al., 1995; Wejda et al., 1995). A nationwide search of Danish records identified 11 776 patients with hyperthyroidism (6301 with diffuse toxic goitre and 5475 with nodular toxic goitre) diagnosed over a 14-year period (1983–1996) and followed up for 113 112 person-years prior to diagnosis and 75 469 person-years subsequent to diagnosis (Vestergaard et al., 2002). Each patient was compared with three age- and gender-matched controls and information on fracture incidence (all types) was obtained from the same computer-based national register as that used to identify the cases. A small, though significant, increase in overall fracture risk was seen in patients with a diffuse toxic goitre before diagnosis only [incidence rate ratio (IRR) 1·17, 95% CI 1·02–1·33]. This increase was accounted for by fractures of the femur, femoral neck and foot. In contrast, patients with nodular toxic goitre had no change in overall fracture risk, although there was an increase in femoral fractures both before and after diagnosis. Age, female gender and previous fractures were additional risk factors for fracture. Thyroid surgery as a treatment for both diffuse and toxic nodular goitre reduced the risk of fracture after diagnosis, (risk estimate 0·71, 95% CI 0·54–0·95 and 0·63, 95% CI 0·50–0·79 for diffuse and nodular toxic goitres, respectively). The mechanism behind this is not entirely clear, as only patients with diffuse nodular goitre had a higher BMD post-surgery, but it may relate to prompter control of the hyperthyroid state with reduced exposure to high circulating thyroid hormone levels. The same authors have recently published a meta-analysis in which a mathematical formula was used to translate BMD scores from 20 published studies (902 patients) into a derived fracture risk (Vestergaard & Mosekilde, 2003). Using this estimate the relative risk for both spinal (RR 1·4, 95% CI 1·3–1·6) and hip (RR 1·4, 95% CI 1·2–1·6) fractures was increased in patients with untreated hyperthyroidism. However, this increased fracture risk did not persist, returning to normal a year after treatment. Prior to this study, a large population-based study mortality study (105 028 person-years of follow-up) had shown a 2·9-fold increased standardized mortality ratio (SMR) due to fractured femur in patients treated with radioactive iodine for hyperthyroidism (Franklyn et al., 1998). Two retrospective cohort studies, one in men (n = 136) and one in women (n = 630), both carried out in Rochester, Minnesota, USA, have examined the influence of a thyroidectomy on fracture risk (Nguyen et al., 1997; Melton et al., 2000). It is postulated that thyroidectomy may be associated with bone loss via a number of mechanisms including endogenous excess of thyroxine prior to surgery, overzealous replacement of thyroxine postsurgery, deregulation of bone resorption secondary to calcitonin deficiency postsurgery, or a combination of these. However, there was no increase in overall fracture risk seen for either men or women following thyroidectomy (RR 1·4, 95% CI 0·6–3·4 for men and SIR 0·9, 95% CI 0·8–1·0 for women). Both groups had small, though statistically significant, increases in hip fracture risk, while the women also had increases in rib, pelvic and vertebral fractures. These results contrast with the reduction in (Vestergaard et al., 2002) or lack of association (Wejda et al., 1995) with hip fracture risk seen after thyroid surgery in two studies mentioned above. Interestingly, neither study showed an association between a history of hyperthyroidism prior to surgery nor thyroxine replacement postsurgery and fracture risk. A particularly interesting study is that by Bauer et al. (1997, 2001) as it shows an increased risk of fracture associated with low levels of TSH in a large population of postmenopausal women who were previously shown to have similar BMD to control subjects. Women who reported a history of hyperthyroidism as well as women taking thyroid hormone had similar rates of bone loss to those who did not report hyperthyroidism or take thyroid hormone. Thyroid hormone use per se was not associated with an increased risk of fracture, while a history of previous hyperthyroidism was associated with a twofold higher risk of hip fracture but no increase in either vertebral or nonspine fractures. These data indicate that the increased fracture risk may be independent of changes in BMD and suggest that DXA scanning lacks sufficient sensitivity to predict fractures in thyrotoxicosis. The inference from this is that the high turnover bone loss seen in hyperthyroidism may result in bone microarchitectural changes and increased fragility that requires more sensitive imaging techniques for accurate quantitation. Future studies relating to fractures and bone structure will need to address this point specifically if fracture risk is to be accurately determined in thyroid disease. In population studies, suppressed TSH from whatever cause is associated with an increased risk of fracture. However, thyroid hormone use per se probably does not increase fracture risk in women, although it may have an effect in men. A past history of hyperthyroidism appears to be associated with an increased risk of fracture which may relate to the duration of exposure to excess thyroid hormones. Recent data suggest that an increase in fracture risk is not necessarily associated with a reduction in BMD. In view of the fact that the predominant thyroid hormone receptor in bone is TRα, it seems logical to suppose that substituting thyroxine with a selective agonist of TRβ might spare bone mass and reduce secondary osteoporosis. Such a synthetic analog (GC-1) has been synthesized and shown to have no effect on BMD in female adult rats treated with suppressive doses for 2 months compared with controls treated with suppressive does of T3 (Chiellini et al., 1998; Baxter et al., 2001; Freitas et al., 2003). In addition, GC-1 does not have the direct cardiac effects of thyroxine as heart, like bone, is primarily a TRα-expressing tissue (Trost et al., 2000). GC-1 has similar effects to T3 in reducing serum cholesterol and suppressing TSH but appears to have differing biological effects to T3 in some other tissues, including brain and adipose tissue (Ribeiro et al., 2001; Morte et al., 2002). Thus, much further work needs to be done in establishing the safety and efficacy of such selective thyromimetics in replacing thyroxine or T3 treatment. This represents an exciting area for ongoing research with the potential for improvements in long-term thyroid hormone therapy with regard to deleterious effects on heart and bone.
'%3e%3cpath fill='%23012169' d='M0 0v30h60V0z'/%3e%3cpath stroke='%23fff' stroke-width='6' d='m0 0 60 30m0-30L0 30'/%3e%3cpath stroke='%23C8102E' stroke-width='4' d='m0 0 60 30m0-30L0 30' clip-path='url(%23b)'/%3e%3cpath stroke='%23fff' stroke-width='10' d='M30 0v30M0 15h60'/%3e%3cpath stroke='%23C8102E' stroke-width='6' d='M30 0v30M0 15h60'/%3e%3c/g%3e%3c/svg%3e)
'/%3e%3cpath d='M200 752.362h200v300H200z' style='fill:%23fff;fill-opacity:1;stroke:none' transform='translate(0 -752.362)'/%3e%3cpath d='M400 752.362h200v300H400z' style='fill:%23ff883e;fill-opacity:1;stroke:none' transform='translate(0 -752.362)'/%3e%3c/svg%3e)