A Large-Scale Population-Based Analysis of Common Genetic Variation in the Thyroid Hormone Receptor Alpha Locus and Bone

Dear Editor:

Thyroid hormone (TH) is essential for normal bone development and the maintenance of adult bone mass. Childhood hypothyroidism leads to growth retardation and delayed bone age, whereas hyperthyroidism accelerates growth and advances bone age. In adults, hyperthyroidism leads to osteoporosis and increased fracture risk. TH receptor alpha (TRα), encoded by the THRA gene, is the predominant TR in bone. No patients with mutations in THRA have yet been described. Mice with inactivating mutations in THRA show a delay in bone development and osteosclerosis in adulthood (1). On the opposite chromosomal strand of THRA, the circadian clock gene NR1D1 (REV-ERBα) is located. THRA and NR1D1 partially overlap, and NR1D1 expression influences splicing and expression of THRA (2). To date, limited data exist on the role of the THRA/NR1D1 locus in human bone physiology. Previous candidate gene studies analyzing THRA in relation to bone mineral density (BMD) had limited sample sizes, and analyses were restricted to a subgroup of the population (i.e., older men; see Supplementary Data [available online at www.liebertonline.com/thy] for an overview of these studies). Therefore, we studied the effects of genetic variation in the THRA/NR1D1 locus on BMD, BMD change, fracture risk, and bone geometry.

A tagging set of 14 polymorphisms was selected to cover the genetic variation in the THRA/NR1D1 locus (see Supplementary Data). Serum TSH and FT4 levels were determined in 1350 subjects from the Rotterdam Study 1 (RS1). Femoral neck and lumbar spine BMD were measured in 19,195 subjects from the Genetic Factors for Osteoporosis (GEFOS) consortium (3). In RS1, femoral neck BMD was measured at baseline and at the second follow-up visit (follow-up [mean (SD)]: 6.51 (0.38) years) in 2366 subjects, and BMD loss rates were calculated. Four geometric outcomes measured at the femoral narrow-neck region in 4131 subjects were used: narrow-neck width, narrow-neck cortical thickness, buckling ratio (index of bone instability), and section modulus (index of bending strength). Thoracolumbar spine radiographs from 2994 subjects were scored for the presence of vertebral fractures (n=371). Information on incident osteoporotic fractures was available for 5974 RS1 subjects (follow-up: 7.79 (3.04) years), and 2157 RS2 subjects (follow-up: 3.95 (0.84) years). The associations of the selected THRA/NR1D1 polymorphisms with baseline characteristics, BMD, fracture risk, and bone geometry were studied using linear, logistic, and Cox regression analyses. See Supplementary Data for detailed information on materials and methods.

The studied THRA/NR1D1 polymorphisms were not associated with baseline characteristics, including serum TSH and FT4. None of the polymorphisms were associated with neither BMD, BMD change, vertebral or incident osteoporotic fractures (see Supplementary Table S1), nor narrow-neck width, narrow-neck cortical thickness, buckling ratio, or section modulus.

The lack of effects of THRA/NR1D1 polymorphisms on bone parameters was unexpected, considering the essential role of TH in bone physiology. THRA is expressed in both osteoblasts and chondrocytes. Mouse models with inactivating THRA mutations, as well as THRA ^0/0 mice (lacking all THRA transcripts), display delayed bone development and osteosclerosis in adulthood (1). Furthermore, patients with TH resistance due to TRβ mutations have increased levels of TH and an increased risk of osteoporosis, which is thought to result from overstimulation of TRα. It has been shown that core circadian clock transcription factors (including REV-ERBα) and multiple metabolic bone homeostasis pathways display a circadian expression profile in bone, and that mice lacking circadian clock components display abnormal bone remodelling (4). Previous genome-wide association studies for BMD and fracture risk have not identified as significant associations with the THRA/NR1D1 locus (3; see Supplementary Data for a review of these studies). However, these hypothesis-free approaches tested >300,000 variants, requiring stringent multiple-testing correction, and resulting in very low p-values to declare statistical significance (i.e., p<5×10⁻⁸). Since the THRA/NR1D1 locus is a plausible candidate locus for bone abnormalities, we performed a focused analysis of this locus on various bone parameters, thereby covering various aspects of bone (patho) physiology. Due to the large sample size, we were powered to detect at least small-to-moderate effects on the studied bone parameters (see Supplementary Data), but still did not find significant associations. However, it is important to note that the apparent absence of (common) functional variation in this locus does not negate its importance in bone.

Although TRα is the major TR in bone, mediating important effects of TH on bone development and turnover, our study excludes an important contribution of genetic variation in the THRA/NR1D1 locus to variations in BMD, fracture risk, and bone geometry in the elderly population.