| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Endocrine Care |
Medical Department M (Endocrinology and Diabetes) and Medical Research Laboratories (C.H.G., J.S.C.), and Department of Clinical Chemistry (A.L.L.), Aarhus Kommunehospital, Aarhus University Hospital, DK-8000 Aarhus C, Denmark; Department of Endocrinology (K.B.), Odense University Hospital, DK-5000 Odense C, Denmark; and Departments of Endocrinology and Metabolism (K.B., L.M.) and Clinical Chemistry (L.H.), Aarhus Amtssygehus, Aarhus University Hospital, DK-8000 Aarhus C, Denmark
Address all correspondence and requests for reprints to: Claus Højbjerg Gravholt, M.D., Ph.D., Medical Department M (Endocrinology and Diabetes) and Medical Research Laboratories, Aarhus Kommunehospital, Aarhus University Hospital, DK-8000 Aarhus C, Denmark. E-mail: . ch.gravholt{at}dadlnet.dk
Abstract
Most women with Turner syndrome (TS) have no gonadal activity and thus lack estrogen. Bone mineral density (BMD) is often reduced, leading to an increased risk of osteoporosis and fractures. However, growth retardation with reduced final height and other endocrine disturbances may compromise interpretation of skeletal measurements. The aim of the present study was to explore skeletal findings, bone metabolism, and calcium homeostasis in TS. Sixty women with TS (age, 37 ± 9 yr) and 181 normal age-matched female controls were studied. Bone area (A; square centimeters), bone mineral content (BMC; grams), area-adjusted BMD (aBMD; grams/square centimeter), and volumetric BMD (vBMD; grams/cubic centimeter) were measured at lumbar spine, femoral neck, and forearm using dual energy x-ray absorptiometry. Twenty-eight percent had osteopenia, and 23% had osteoporosis, according to World Health Organization criteria. At the lumbar spine, A, BMC, aBMD, and vBMD were reduced by 18, 27, 11, and 6%, respectively; at the femoral neck, A, BMC, and aBMD were reduced by 2, 10, and 8%, respectively, whereas the 9% reduction in vBMD was insignificant (P = 0.07); and in the forearm, A, BMC, and aBMD were reduced by 53, 55, and 9%, respectively. Bone markers indicated an enhanced bone resorption (21 and 23% increase in C-terminal and N-terminal cross-linking telopeptides of type I collagen/creatinine, respectively) with unchanged (osteocalcin, procollagen I N-terminal propeptide) or reduced (54% reduction in bone alkaline phosphatase) bone formation. Plasma levels of calcium and 25-hydroxyvitamin D (26%) were reduced, and PTH levels increased (74%) in TS. IGF-I (30%), IGF binding protein 3 (18%), testosterone (50%), and SHBG (40%) were reduced in TS.
In summary, A, BMC, and aBMD were found to be universally reduced in TS, whereas vBMD was slightly reduced in the spine. Increased resorption of bone was present, with normal or blunted bone formation, suggesting uncoupling or imbalance in bone remodeling. Skeletal changes may be induced by chromosome abnormalities or by secondary endocrine or metabolic changes related to a relative estrogen deficiency, testosterone deficiency, reduced IGF-I, low vitamin D status, and secondary hyperparathyroidism.
MOST PATIENTS WITH Turner syndrome (TS) have no gonadal function and are thus, if untreated, deficient in female sex steroids from early childhood onward. Besides gonadal insufficiency, the cardinal stigmata of TS are growth retardation, with reduced final height and infertility. Furthermore, a number of congenital malformations and conditions are associated with TS.
Adolescents and young adults with TS have decreased bone mineral density (BMD; grams/square centimeter) (1, 2, 3, 4, 5), and osteopenia is also present in middle-aged women with TS (6, 7, 8). A register-based study has suggested that osteoporosis [relative risk, 10.12; 95% confidence interval (CI), 2.18–30.23] and fractures (relative risk, 2.16; 95% CI, 1.50–3.00) are frequent diagnoses in a TS population (9). This indicates that the decreased BMD seen in other studies does lead to clinical consequences. Treatment with estrogens is pivotal to induce maximal peak bone mass in adolescents and young adults (10, 11, 12, 13). This is supported by longitudinal studies of estrogen-deficient adolescents with TS. In these studies, patients with spontaneous menstruations had normal BMD, whereas patients without menstruations had reduced BMD (14, 15). Furthermore, GH seems to improve BMD (16, 17), although in these two small studies no untreated control group of girls with TS was included, and the patients were only followed for up to 2 yr. Vitamin D metabolism has been found to be abnormal, with a blunted response in serum 1,25-dihydroxy-vitamin D [1,25-(OH)2-D] levels to a low-calcium diet (18), whereas calcitonin metabolism seems to be normal (19). It has been suggested that the reduced BMD in TS is due to the syndrome per se in combination with estrogen deficiency (3), whereas others feel, that it is largely explained by estrogen deficiency alone (2, 5, 10). Recently, we observed that the increased risk of fractures in TS was present already in childhood and persisted throughout all age groups (9). This favors the view that the low BMD seen in TS is the result of both estrogen deficiency and other yet unknown mechanisms caused by the chromosome aberrations or other (endogenous and exogenous) factors.
In normal women, androgen independently predicts parts of peak bone mass (20), and androgen insufficiency may contribute to the development of postmenopausal osteoporosis (21). The GH-IGF axis may also play a role in determining peak bone mass, as well as maintaining bone mass in normal subjects (22). Women with TS, however, are also deficient in androgens (23, 24) and have reduced levels of free IGF-I (25), suggesting that the pathogenesis of osteoporosis in TS involves deficiency of several hormones in combination with congenital skeletal abnormalities.
Thus, we wanted to explore skeletal findings, bone metabolism, and calcium homeostasis in adult patients with TS. To this end, we examined a group of women with TS with dual energy x-ray absorptiometry (DEXA); assessment of bone markers, growth factors, and sex steroids; and history of estrogen supplementation. An age-matched group of healthy women was examined for comparison.
Subjects and Methods
Subjects
The study group consisted of 60 patients (age, 37 ± 9 yr; range, 22–67 yr) with TS, diagnosed by chromosome analysis. Karyotypes were distributed as follows: 45,X (n = 29); 45,X/46,XX (n = 5); karyotypes with isochromosomes (Xq) or deletions (n = 16); karyotypes with Y chromosome material (n = 5); and karyotypes with a marker or ring chromosome (n = 5).
All patients were recruited through the National Society of Turner Contact Groups in Denmark. Inclusion involved mailing of letters by the National Society to all 90 members above 20 yr of age. A total of 60 chose to participate. Fifty-five of the patients had menstruated spontaneously (n = 5) or received conventional sex hormone replacement therapy (HRT) (n = 50) consisting of 17ß-estradiol (2 mg) for the entire cycle and norethisterone (1 mg), medroxyprogesterone (10 mg), or levonorgestrel (0.25 mg) for 10 d every cycle. Two of the five women with spontaneous menstruations were still menstruating, whereas the other three experienced premature ovarian failure and now received HRT. All were interviewed concerning the age at menarche (if present), age at start of induction of puberty (by exogenous estrogen), age at premature menopause (if present), duration of HRT, and age at cessation with HRT (when relevant), enabling summation of total estrogen exposure (in years). These variables were used in subsequent statistical computations. The average duration of HRT was 16 ± 9 yr, and the age at start of HRT was 21 ± 10 (range, 9–60) yr. Five patients had chosen not to receive HRT. There were no differences in any DEXA or biochemical parameters between estrogen-exposed and unexposed patients, except a lower level of serum estradiol, as expected, and a higher level of IGF-II and IGF-binding protein (IGFBP)-3 in the untreated group (n = 5). We therefore chose to combine the two groups in all further computations.
A control group of 181 healthy women between 20 and 79 yr of age was also included in the study. They participated in an ongoing study on normal bone metabolism and were recruited by advertisement in factories, university, police and fire corps, and senior citizen clubs. DEXA scans of the lumbar spine were available for all 181 women, whereas scans of the hip and wrist were available for 179 and 135 women, respectively. None of the controls had been treated with or were currently receiving HRT. From this large cohort, 59 normal age-matched female controls (no karyotyping was performed) were selected for biochemical studies (details concerning age, body size, and VO2-max are shown in Table 1
). Data on DEXA were compared directly with this age-matched control group (n = 59) but were also reported as T- and Z-scores derived from the larger group of 181 normal women. There was no difference in Z- and T-scores of BMD between this smaller control group and the entire control group.
|
Methods
Participants were examined in the morning. Women were instructed not to eat or drink anything other than bottled mineral water and met fasting in the laboratory. All patients and controls delivered a second void spot urine sample that was collected in the fasting state in the morning and was stored at -20 C for later analysis. After blood was drawn, serum was immediately separated and stored at -20 C (-80 C for controls) in multiple vials for later analysis (more TS patients than controls had blood tests taken during the summer and autumn). Body weight was measured to the nearest 0.1 kg on an electronic scale, and body height was measured to the nearest 0.5 cm, with the subjects in underwear and bare feet. Body mass index (BMI) was calculated as weight (kilograms) divided by height (meters) squared, and the waist-to-hip ratio (only in TS) was determined in the supine position. Body surface area was calculated according to the DuBois equation: S = 0.007184 x weight0.425 x height0.725, where weight is expressed in kilograms, height in centimeters, and S in square meters. A detailed history of HRT and treatment with GH or androgens was obtained in women with TS.
BMC (grams) and area-adjusted BMD (aBMD) (grams/square centimeter) were measured at the lumbar spine (L2–L4), the hip (femoral neck and trochanteric region), and the nondominant forearm (ultradistal and proximal part of distal third) by DEXA on Hologic 1000/w or 2000/w osteodensitometers (Hologic, Inc., Waltham, MA). Cross-calibration was ensured through the use of double measurements and a phantom. Precision for BMD was 1.5% for the lumbar spine, 2.1% for the femoral neck, and 1.9% for the ultradistal forearm. These quantities included cross-over calibration, change of hardware, change of technicians, and long-term stability (<0.2%/yr).
aBMD values were reported in absolute values and as T- and Z-scores based on the distribution of BMD values in the larger control group across the age range. On the basis of mean and SD values derived from this analysis, T- and Z-scores corrected for age were computed as follows:
 |
 |
Volumetric BMD (vBMD) was estimated for the spine and the femoral neck in all TS patients (n = 60) and in the age-matched controls (n = 59). It was assumed that the geometry of each vertebral body could be aligned to a cylinder and the geometry of the scanned part of the femoral neck could be aligned to a truncated cone.
In the vertebral body, the proportion of the actual height (h) and width (w) of the vertebral body equals the proportion of the height (y) and width (x) measured from the scans:
 |
 |
 |
 |
 |
 |
 |
 |
vBMD of the femoral neck was estimated as a truncated cone, with volume, V, where
 |
 |
Assays
Plasma intact PTH was measured by a chemiluminescence assay using an automated instrument (Immulite, DPC, Los Angeles, CA) (26). Urine N-terminal cross-linking telopeptide of type I collagen (NTX) was measured by an immunometric assay using a Vitros ECI analyser (Ortho Clinical Products, Amersham Pharmacia Biotech, Little Chalfont, UK). This assay uses monoclonal antibodies against human NTX (27). Plasma osteocalcin (total OC) was measured using the N-Mid-Osteocalcin assay on an automated analyser (Elecsys 2010 analyzer, Roche Diagnostics, Mannheim, Germany) with antibodies that determine both intact OC (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49) and the large N-Mid terminal fragment (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43) (28). The coefficient of variation (CV)total was 4–7% for the various automated assays.
Plasma bone-specific alkaline phosphatase (bone ALP) was measured after lectin precipitation using a Hitachi 917 automated analyzer (Roche Diagnostics) (29). This assay was performed with a CVtotal of 8%. Plasma concentrations of procollagen I N-terminal propeptide (PINP) (30), procollagen III N-terminal propeptide (PIIINP) (31), and C-terminal cross-linking telopeptide of type I collagen (ICTP) (32) were determined by commercial RIA kits (Orion Diagnostica, Espoo, Finland). Intra- and interassay CV of 5 and 7%, respectively, were observed. Plasma 25-hydroxyvitamin D (25-OHD) was measured by an equilibrium RIA procedure (DiaSorin, Inc., Stillwater, MN) (33). Intra- and interassay CV of 10 and 13%, respectively, were observed. Plasma 1,25-(OH)2-D was measured by a competitive RRA after chromatographic extraction (Nichols Institute Diagnostics, San Juan Capistrano, CA) (34). Intra- and interassay CV were 8 and 10%, respectively. Vitamin D binding protein (VDP) was measured by an immunonephelometric assay (35). Serum albumin, calcium, phosphate, and total ALP, and urinary creatinine, calcium, and phosphate were measured by standard laboratory methods.
As indices of the IGF axis, serum IGF-I and IGF-II were measured by noncompetitive time-resolved immunofluorometric assays (36), serum IGFBP-2 was measured by RIA, and IGFBP-3 was measured by an immunoradiometric assay (Diagnostics Systems Laboratories, Inc., Webster, TX); free IGF-I was calculated as the molar ratio between IGF-I and IGFBP-3. The following molecular masses were used: IGF-I, 7.5 kDa; and IGFBP-3, 42.0 kDa. Serum estradiol and testosterone were measured by a commercial time-resolved fluoroimmunoassay (DELFIA, Wallac, Inc., Turku, Finland).
Statistical analysis
All statistical calculations were performed with SPSS for Windows version 10.0 (SPSS, Inc., Chicago, IL). Data were examined by two-tailed unpaired t tests or the Mann-Whitney two-tailed test when appropriate. Results are expressed as mean ± SD. However, when not parametrically distributed, they were logarithmically transformed and given as geometric mean ·/÷ (i.e. multiplied/divided by) antilog SD and range. Significance levels less than 5% were considered significant. Pearson or Spearman correlation was used, where appropriate, to examine relations between bone markers, and between body size and BMD. When examining relations between markers of bone formation and resorption, a protected Bonferroni significance level of less than 0.01 was used.
Simple division of BMD by the square root of the surface (BMDsurface area) has been suggested as the best method of disregarding body size and thus the different size of the bones due to body size (37). However, other elements of body size may also contribute to the variance in BMD. In this study, vBMDs were derived. Furthermore, multiple backward stepwise linear regression was used to examine the principal determinants of BMD in the control group. Having identified different body size measures (i.e. height, weight, BMI, and surface area) as independent predictors of BMD, we used subsequent analysis of covariance to calculate adjusted levels of BMD (BMDadjusted) in the entire study group (including all controls and TS patients). The method allows for the removal of the linear effects of covariates on the dependent variable (38). This approach eliminates problems arising from not having a correlation of r = 1.0 between the dependent variable and independent variables and from the intercept between any two variables being different from zero.
Multiple backward stepwise linear regression was used to examine the principal biochemical determinants of skeletal size, aBMD and vBMD. As biochemical determinants, all bone markers, indices of the IGF axis, and testosterone were studied. Serum estradiol was not used in these computations, because it is dependent on whether or not women with TS received HRT.
Results
BMC, aBMD, and bone size (Table 2)
In TS patients, BMC (grams) and BMD (grams/square centimeter) were significantly reduced at all examined sites. According to the World Health Organization definition, 28% (n = 17) of the patients had osteopenia (-1 > T-score > -2.5) and 23% (n = 14) had osteoporosis (T-score < -2.5). Expressed in Z-scores, the reduction in BMD in the TS patients was approximately -1 SD. Z-scores showed no definite trend with age (spine, r = 0.046, P = 0.7; hip, r = 0.016, P = 0.9; arm, r = 0.158, P = 0.2) (Fig. 1). This was true for all regions studied. The reduction in Z-score was largest at the unloaded bones of the wrist, where the mean Z-score was -1.23, whereas the Z-scores at load-bearing sites of the spine and the femoral neck were -0.82 and -0.75, respectively (Table 2).
|
|
Bivariate correlations between BMD in the spine, hip, and arm, and height, weight, BMI, and the square root of the surface area revealed distinct differences between TS and controls. BMI was not correlated to BMD at any site in controls (P > 0.2), but it was the variable most consistently and significantly correlated to BMD in TS (BMDspine, r = 0.226, P = 0.09; BMDhip, r = 0.493, P < 0.0005; BMDarm, r = 0.399, P = 0.002). On the contrary, height and the square root of the surface area were consistently correlated to BMD at any site in controls (height: BMDspine, r = 0.442, P < 0.0005; BMDhip, r = 0.289, P < 0.0005; BMDarm, r = 0.274, P = 0.002; square root of surface area: BMDspine, r = 0.305, P < 0.0005; BMDhip, r = 0.302, P < 0.0005; BMDarm, r = 0.163, P = 0.07), whereas weight correlated significantly at the spine (r = 0.199, P = 0.009) and the hip (r = 0.249, P = 0.001), but not at the arm (P = 0.2). In women with TS, the picture was less clear. At the arm, height, weight, and the square root of the surface area were significantly correlated to BMD (height, r = 0.274, P = 0.04; weight, r = 0.414, P < 0.0005; square root of surface area, r = 0.412, P = 0.002), whereas no significant correlation was found at the spine (P > 0.09), and at the hip only weight (r = 0.449, P < 0.0005) and the square root of the surface area (r = 0.415, P = 0.001) correlated significantly, while height did not (P = 0.2). In controls, multiple regression models consistently showed that the square root of the surface area and age were the variables explaining most of the variation in BMD. Therefore, combined analysis of controls and TS women was performed, with age, the square root of the surface area, and status (being a TS patient or a control) as independent variables. At the level of the spine and the hip, the square root of the surface area and age were independent predictors of BMD (spine, multiple r = 0.574, P < 0.0005; hip, multiple r = 0.596, P < 0.0005), whereas status did not reach significance in the model. At the level of the arm, all three variables were significant explanatory variables (multiple r = 0.649, P < 0.0005).
As expected, the areas (A) of the bones in all three regions (forearm, spine, and hip) were reduced in TS compared with age-matched controls (Table 2
), indicating a smaller skeleton in TS patients. However, controlling for body size, no difference in the spine was found. ANCOVA-adjusted (mean ± SE) values for area of the bones at the level of the spine were comparable. Height, BMI, and age were used in the model and were significant explanatory variables, whereas status (i.e. being a woman with TS or a control) was not (Acorrected, TS patients vs. controls, 55.73 ± 0.84 cm2 vs. 56.17 ± 0.37 cm2, P = 0.68), whereas the area of the hip was larger in TS (Acorrected, 39.03 ± 0.61 cm2 vs. 35.21 ± 0.27 cm2, P < 0.0005; corrected for height, weight, BMI, age, and status), and the area of the distal forearm was very much smaller in TS than in controls (Acorrected, 13.65 ± 0.35 cm2 vs. 21.58 ± 0.19 cm2, P < 0.0005; corrected for height, weight, BMI, age, and status).
vBMD (Table 2)
To exclude the influence of variations in skeletal size on BMC and aBMD, we estimated vBMD at the spine and the femoral neck. vBMDspine was significantly reduced by 6% (TS vs. control, 0.29 ± 0.04 vs. 0.31 ± 0.03 g/cm3; P = 0.04), suggesting not only a reduction in bone size in TS but also a slight, but real reduction in vBMD. In comparison, bone area, BMC, and aBMD were reduced by 18, 27, and 11%, respectively. vBMD at the femoral neck was also reduced (9%), but the difference did not reach significance (TS vs. control, 0.31 ± 0.07 vs. 0.33 ± 0.05 g/cm3; P = 0.07). In comparison, bone area, BMC, and aBMD were reduced by 2, 10, and 8%, respectively. Using estimated vBMD, no correlation between measures of body size and BMD could be demonstrated in TS or controls (all r 
0.220; 0.1 < P 1, except height in controls, r = 0.229, P = 0.003), showing that the influence of skeletal size and body composition is all but abolished by computation of vBMD. Furthermore, although age and age at start of HRT exposure did not correlate with aBMD in TS, a strong correlation was found between these two measures and vBMD. This was true both at the level of the spine (age, r = -0.570, P < 0.0005; age at start of HRT, r = -0.510, P < 0.0005) and at the femoral neck (age, r = -0.415, P = 0.001; age at start of HRT, r = -0.419, P = 0.001). Duration of HRT did not correlate significantly with vBMD. In controls, age and vBMD also correlated significantly (vBMDspine, r = -0.358, P = 0.006; vBMDneck, r = -0.391, P = 0.002) (Fig. 2).
|
There was no difference between women with TS with a 45,X karyotype compared with all other karyotypes in skeletal area, aBMD, or vBMD in any of the sites examined (data not shown).
Calcium homeostasis and biochemical markers of bone formation and degradation (Table 3)
These variables were measured in all women with TS (n = 60) and in a subgroup of controls (n = 59). The plasma levels of PIIINP, PINP, total OC, 1,25-(OH)2-D, VDP, 1,25-(OH)2-D/VDP ratio, and total ALP were comparable in TS patients and controls, whereas average plasma levels of ICTP were elevated by 21% (P < 0.0005) and PTH by 74% (P < 0.0005) in the TS group, while plasma levels of bone ALP and 25-OHD were reduced by 54% (P < 0.0005), and 26% (P = 0.001), respectively. The renal excretion of NTX (NTX/creatinine ratio) was increased by 24% (P = 0.035) in women with TS compared with controls. Plasma levels of calcium, albumin-corrected calcium, phosphate, and albumin were all reduced in women with TS. There was no significant correlation between PTH and 25-OHD or 1,25-(OH)2-D in women with TS or in controls (results not shown). Likewise, there was no significant correlation between PTH and calcium or albumin-corrected calcium in the two groups. On the contrary, PTH correlated significantly with total OC in both TS patients (r = 0.274; P = 0.03) and controls (r = 0.319; P = 0.02), and with the renal excretion of NTX/creatinine ratio (r = 0.294; P = 0.02) in TS patients, but not in controls. In controls, we found strong and positive correlations between all the examined markers of bone formation (PINP, total OC, and bone ALP) (all P 0.001), whereas in TS these correlations were weakened (0.1 < P < 0.0005). Likewise, a significant positive correlation was found between the markers of bone resorption (ICTP and NTX/creatinine ratio) in controls, (r = 0.442; P = 0.001) and in TS (r = 0.419; P = 0.001). There was evidence for tight coupling between bone formative and resorptive markers in controls, as seen from significant positive correlations between these markers (all P 0.001). In TS, NTX/creatinine ratio correlated positively with total OC, PINP, and bone ALP (all P 0.005), but ICTP did not correlate with total OC, PINP, and bone ALP.
|
)Serum IGF-I, IGFBP-3, and IGF-I/IGFBP-3 ratio, as a measure of free IGF-I, were reduced in TS compared with controls, whereas IGF-II and IGFBP-2 were comparable in the two groups. The level of plasma estradiol was comparable in the two groups, whereas plasma testosterone and SHBG levels were reduced by 50 and 40%, respectively, in TS women. Markers of bone resorption and formation correlated positively with indices of the IGF axis and with plasma testosterone and estradiol, however, none of these correlations reached significance at the 1% level.
Multivariate models to estimate variations in skeletal area
In the entire population of patients (n = 60) and the matched controls (n = 59), vertebrae size showed significant positive bivariate correlations with IGF-I (r = 0.429; P < 0.0005), IGFBP-3 (r = 0.442; P < 0.0005), IGF-I/IGFBP-3 (r = 0.278; P = 0.003), and bone ALP (r = 0.408; P < 0.0005), and negative correlations with PTH (r = -0.350; P < 0.0005), ICTP (r = -0.227; P = 0.014), and NTX/creatinine (r = -0.234; P = 0.014). The variables were entered into a stepwise multiple regression model including status (i.e. being a TS patient or a control), and with vertebral area as the dependent variable. In this model (r = 0.731; P < 0.0005), only status was a significant explanatory variable (P < 0.0005).
Likewise, total hip area showed significant bivariate positive correlations with IGF-I (r = 0.205; P = 0.031), IGF-II (r = 0.211; P = 0.026), and IGFBP-3 (r = 0.288; P = 0.002). In a multiple regression model with total hip area as the dependent variable (r = 0.353; P = 0.001), IGF-II (P = 0.071) and status (P = 0.002) were significant variables.
The total area of the distal forearm showed significant positive bivariate correlations with IGF-I (r = 0.310; P = 0.002), IGF-II (r = 0.259; P = 0.011), IGFBP-3 (r = 0.455; P < 0.0005), testosterone (r = 0.279; P = 0.006), and bone ALP (r = 0.404; P < 0.0005), and negative correlations with ICTP (r = -0.197; P = 0.049) and NTX/creatinine (r = -252; P = 0.014). In a multivariate model (r = 0.716; P < 0.0005) with area of the distal forearm as the dependent variable, IGF-I (P = 0.025), IGF-II (P = 0.033), and status (P < 0.0005) were all explanatory variables.
Multivariate models to estimate variations in areal BMD
In the entire population of patients (n = 60) and the matched controls (n = 59), aBMDspine showed significant positive bivariate correlations with IGF-I (r = 0.300; P = 0.001), IGF-II (r = 0.211; P = 0.026), IGFBP-3 (r = 0.358; P < 0.0005), testosterone (r = 0.189; P = 0.05), and bone ALP (r = 0.242; P = 0.009), and negative correlations with PTH (r = -0.236; P = 0.012) and NTX/creatinine ratio (r = -0.194; P = 0.045). These variables were included as independent variables in a stepwise linear regression model together with status, and with aBMDspine as the dependent variable. In this model (r = 0.443; P < 0.0005), IGFBP-3 (P = 0.075) was a significant contributory variable, together with status (P = 0.003).
In the same way, aBMDhip showed significant positive bivariate correlations with IGF-I (r = 0.294; P = 0.002), IGFBP-3 (r = 0.290; P = 0.002), IGF-I/IGFBP-3 (r = 0.206; P = 0.03), 25-OHD (r = 0.216; P = 0.02), and bone ALP (r = 0.198; P = 0.033), and negative correlations with 1,25-(OH)2-D/VDP ratio (r = -0.213; P = 0.024), IGFBP-2 (r = -0.251; P = 0.008), and NTX/creatinine ratio (r = -0.300; P = 0.002). Again, a stepwise linear regression model was constructed with aBMDhip as the dependent variable. In this model (r = 0.522; P < 0.0005), NTX/creatinine ratio (P = 0.049), 1,25-(OH)2-D/VDP ratio (P = 0.043), and IGFBP-2 (P = 0.006) were significant contributory variables together with status (P = 0.001).
At the level of the forearm, aBMDarm correlated significantly and positively with IGF-I (r = 0.381; P < 0.0005), IGF-II (r = 0.330; P = 0.001), IGFBP-3 (r = 0.345; P = 0.001), IGF-I/IGFBP-3 (r = 0.295; P = 0.004), testosterone (r = 0.225; P = 0.028), 25-OHD (r = 0.240; P = 0.016), and bone ALP (r = 0.254; P = 0.011), and negatively with PTH (r = -0.202; P = 0.045), total ALP (r = -0.225; P = 0.024), 1,25-(OH)2-D/VDP ratio (r = -0.235; P = 0.020), and NTX/creatinine ratio (r = -0.365, P < 0.0005). In a stepwise linear regression model (r = 0.548; P < 0.0005) constructed with aBMDarm as the dependent variable, NTX/creatinine ratio (P = 0.015) and IGF-II (P = 0.014) were significant contributory variables together with status (P < 0.0005).
Multivariate models to estimate variations in vBMD
vBMDspine showed significant bivariate correlations with IGF-I (r = 0.251; P = 0.008), IGFBP-3 (r = 0.218; P = 0.02), and IGFBP-2 (r = -0.206; P = 0.03) in the entire study population. In the subsequent stepwise linear regression model (r = 0.298; P = 0.007), IGF-I (P = 0.016) and IGFBP-2 (P = 0.099) remained significant contributory variables, whereas status did not.
In the entire study population, vBMDhip only correlated significantly with NTX/creatinine ratio (r = -0.214; P = 0.026). In the following stepwise linear regression model (r = 0.214; P = 0.026), NTX/creatinine ratio (P = 0.026) remained a significant contributory variable, whereas status did not.
Discussion
The main result of the present study is the reduction in skeletal size, BMC, aBMD, and to a lesser degree vBMD at all measured sites. The magnitude of this reduction was similar at all locations, when compared with a large age-matched control group. However, an interesting dichotomy emerged when adjusting for the pronounced differences in body composition between the women with TS and the controls. Indeed, at the level of the spine and the hip, there does not seem to be any reduction in BMD, adjusting for body composition. The same was not true for the arm. Likewise, when looking at vBMD, which removes most of the influence of body composition on BMD, only slight differences were found, although significant in the spine. Previous studies have also reported reduced BMD. However, none of these studies have taken the smaller size of women with TS into account (6, 7, 8), except one study of adolescents in which vBMDspine was comparable to that of a reference population (13). Small size and height may well confound interpretation of BMD, because of the two-dimensional nature of DEXA scanning (37). Because reduced height is the almost universal finding in TS, addressing this issue is pivotal. Indeed, very disparate results were evident when looking at skeletal bone size. Our study extends observations in previous anthropometric studies reporting disproportional growth in TS (39), because we found lumbar vertebrae to be of normal size adjusting for body composition, whereas the proximal hip was increased in size, and the distal forearm was markedly smaller compared with controls. Earlier, a study showed increased risk of fracture of the wrist in young girls with TS, despite normal BMD, when adjusting for relevant measures of body composition (40). The recent cloning of a novel gene from the pseudoautosomal region on the X and Y chromosome, named SHOX (41) or PHOG (42), and subsequent studies of expression are of considerable interest and may shed some light on the bone anomalies seen in TS. Haploinsufficiency of the SHOX gene is seen in TS and in Leri-Weill dyschondrosteosis, another syndrome showing some of the skeletal abnormalities seen in TS. SHOX is exclusively expressed in the developing limbs and in the first and second pharyngeal arches and may thus explain the mesomelic short stature and the other skeletal features of TS, i.e. short fourth metacarpal bones, cubitus valgus, and Madelung deformity. It is suggested that SHOX is a repressor of growth plate fusion and skeletal maturation in the distal limbs (43), meaning that haploinsufficiency of SHOX could lead to the premature growth plate fusion in the distal limbs in TS (44). This may also explain the current findings of distinct disproportionality of skeletal size in the different regions studied. Furthermore, Kosho et al. (45) have suggested that gonadal estrogens exert a maturational effect on distal skeletal tissue susceptible to premature growth plate fusion because of the SHOX haploinsufficiency, explaining why the skeletal features may not be apparent before spontaneous puberty or induced puberty.
Spontaneously menstruating girls with TS do deposit adequate amounts of bone mass during puberty (although being just as growth-stunted as their nonmenstruating peers) (15). Young women with TS with normal BMD at puberty fail to reach normal peak BMD, despite seemingly adequate HRT (46). A number of explanations may be given. First, biochemical bone markers indicate increased bone resorption (see below) in the present population of TS women, as noted previously (8). This could be a consequence of insufficient HRT, because most women with TS are given a dose of 2 mg 17ß-estradiol or equivalent. This dose may be sufficient in postmenopausal women but is probably too low for younger women with ovarian failure. Peak bone mass is dependent on a number of factors, such as genetic background, nutrition, physical activity, local growth factors, and a number of hormones. In particular, a normal secretion of estradiol during puberty is required for normal skeletal mineralization (47, 48, 49, 50), and recently, prepubertal estradiol secretion has been highlighted as playing a role for early accretion of bone (51). Often, adolescents with TS are introduced to estrogens late to avoid the stunting in growth conferred by the compounds, probably delaying and possibly reducing appropriate bone mineral acquisition. In our study, we did see an association between age at start of HRT in women with TS and vBMD, suggesting that induction of puberty at an appropriate age is important. However, it is important to stress that the cross-sectional nature of the present study makes it difficult to infer definite conclusions from this association.
Here, we confirmed a reduced level of testosterone (
50%) in TS (24). Previously, we have shown that free testosterone is also significantly reduced in TS, despite a similar reduction in SHBG as found in this study, which would otherwise tend to increase the free fraction of testosterone (24). Low levels of androgens have in peri- and postmenopausal women been associated with reduced BMD (21).
We have previously documented that despite comparable levels of total IGF-I, free IGF-I is reduced (25). Here, in a larger setting, total IGF-I is reduced in conjunction with IGFBP-3, the major IGFBP, as is an indirect measure of free IGF-I, the IGF-I/IGFBP-3 ratio. Thus, our findings suggest that a reduced GH-IGF-IGFBP axis contributes to the reduction in skeletal area and aBMD seen in TS. Serum IGF-I has been shown to be positively correlated with changes in BMD in the femoral neck and the ultradistal forearm in normal women (52). Recent studies of girls with TS receiving GH show that these girls have normal BMD through puberty, despite apparently rather late introduction to estrogens (13, 16, 17, 53). This suggests that the supraphysiological (acromegalic) level of GH (and thus IGF-I) is necessary to reach normal BMD (and possibly skeletal size) in females with TS.
We observed a lower level of 25-OHD in the TS group. The reason for this could be a reduced exposure to solar radiation in the TS group for psychological reasons, especially because more controls than TS patients had blood tests taken during the winter time, although dietary vitamin D deficiency and altered vitamin D metabolism cannot be excluded. The finding could explain the reduced levels of calcium and the increased plasma levels of PTH, also found by others (8), and the enhanced bone resorption (increased plasma ICTP and renal NTX/creatinine). The plasma levels of 1,25-(OH)2-D were normal despite the lower levels of plasma 25-OHD, which could be a consequence of the observed reduced calcium and increased PTH levels, which would both stimulate the renal production of 1,25-(OH)2-D. The findings do not support a previous report (18) of a blunted response of this hormone to a low-calcium diet in TS patients. VDP levels were also within the normal range. The VDP serves the role as a carrier protein for vitamin D, regulating its function thereof by delivering 1,25-(OH)2-D to the cells, and possibly partaking in inflammatory responses (54, 55). The results indicate that the concentration of free 1,25-(OH)2-D is normal in TS patients.
The markers of bone formation used in the present study reflect different stages of osteoblast cell function, i.e. the early phases of bone formation generate PINP, being involved in the deposition of collagen, whereas PIIINP reflects extraosseus collagen formation. They were both unchanged in women with TS compared with controls. Bone ALP and OC are generated during the process of bone mineralization, and are thus indicative of osteoblast function. Bone ALP mainly reflects matrix formation and OC matrix mineralization. However, although bone ALP was reduced in TS, plasma total OC was comparable to the levels in normal women. In contrast to the normal or decreased bone formation markers, plasma ICTP and renal NTX/creatinine, which are both sensitive markers of bone resorption (56), were significantly increased. In the present population, it is difficult to reveal whether inadequate sex HRT or the relative hyperparathyroidism is the cause of the enhanced resorption, or it is a consequence of other metabolic changes or the genetic defect per se. The observed positive correlation between plasma PTH and total OC and renal NTX/creatinine corroborates a biological effect of PTH on bone in TS patients.
The observed discrepancy between resorptive and formative bone markers is remarkable because it could indicate an impaired coupling between the two processes. In fact, relative estrogen and GH-IGF deficiency and IGF resistance (57) may act in concert to explain the absence of enhanced bone formation. In controls, these processes were tightly coupled, as indicated by close correlations between the two processes.
Of interest, correlation analyses and subsequent stepwise linear regression analyses showed the importance of the IGF axis in predicting skeletal size and BMD. In the present populations, we found moderate correlations between the IGF axis and skeletal area, suggesting that the activity of this axis has a potential influence on skeletal size. Also, NTX/creatinine was almost universally a strong predictor of skeletal size and BMD in these two populations, as were different measures of vitamin D, as well as ALP, especially the bone fraction. The importance of the measures of the IGF system in predicting BMD in postmenopausal women has been shown before (52).
In conclusion, bone area, BMC, and BMD are universally reduced in TS, being partly attributable to differences in body composition and possibly to a relative estrogen deficiency and low IGF-I. However, vBMD was only slightly reduced, showing that the main deficiency involves skeletal size. Numerous bone-specific biochemical abnormalities are present in TS, some indicating that the conventional HRT regimen may be inadequate. Furthermore, levels of testosterone and total and free IGF-I are reduced, possibly contributing to the reductions seen in BMD. In the clinical practice of women with TS, surveillance of BMD should be an integral part.
Acknowledgments
We thank Lone Korsgaard, Lone Svendsen, Iben Christensen, Kirsten Hald, Charlotte Gylling, and Donna Arbuckle Lund for expert technical help.
Footnotes
This work was supported by Grant 9600822 from the Danish Health Research Council (Aarhus University–Novo Nordisk Center for Research in Growth and Regeneration). C.H.G. is supported with a research fellowship by the University of Aarhus.
Abbreviations: A, Area; aBMD, area-adjusted BMD; ALP, alkaline phosphatase; BMC, bone mineral content; BMD, bone mineral density; BMI, body mass index; CI, confidence interval; CV, coefficient(s) of variation; DEXA, dual energy x-ray absorptiometry; HRT, hormone replacement therapy; IGFBP, IGF-binding protein; NTX, N-terminal cross-linking telopeptide of type I collagen; OC, osteocalcin; 1,25-(OH)2-D, 1,25-dihydroxy-vitamin D; 25-OHD, 25-hydroxy-vitamin D; PIIINP, procollagen III N-terminal propeptide; PINP, procollagen I N-terminal propeptide; TS, Turner syndrome; vBMD, volumetric BMD; VDP, vitamin D binding protein.
Received October 2, 2001.
Accepted March 5, 2002.
References
,25-dihydroxyvitamin D3 function as ligands for the D-binding protein, nuclear receptor and membrane receptor: a status report. J Steroid Biochem Mol Biol 56:13–22[CrossRef][Medline]
This article has been cited by other articles:
 |
|
 |
|
  C. H. Gravholt, H. E. Poulsen, P. Ott, J. S. Christiansen, and H. Vilstrup Quantitative liver functions in Turner syndrome with and without hormone replacement therapy Eur. J. Endocrinol., June 1, 2007; 156(6): 679 - 686. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. A. Bondy and for The Turner Syndrome Consensus Study Group Care of Girls and Women with Turner Syndrome: A Guideline of the Turner Syndrome Study Group J. Clin. Endocrinol. Metab., January 1, 2007; 92(1): 10 - 25. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. H. Gravholt, B. E. Hjerrild, L. Mosekilde, T. K. Hansen, L. M. Rasmussen, J. Frystyk, A. Flyvbjerg, and J. S. Christiansen Body composition is distinctly altered in Turner syndrome: relations to glucose metabolism, circulating adipokines, and endothelial adhesion molecules. Eur. J. Endocrinol., October 1, 2006; 155(4): 583 - 592. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Stochholm, S. Juul, K. Juel, R. W. Naeraa, and C. Hojbjerg Gravholt Prevalence, Incidence, Diagnostic Delay, and Mortality in Turner Syndrome J. Clin. Endocrinol. Metab., October 1, 2006; 91(10): 3897 - 3902. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. H. Doswell, J. Visootsak, A. N. Brady, and J. M. Graham Jr Turner Syndrome: An Update and Review for the Primary Pediatrician Clinical Pediatrics, May 1, 2006; 45(4): 301 - 313. [Abstract] [PDF]
|
|
|
|
|
|
W. Hogler, J. Briody, B. Moore, S. Garnett, P. W. Lu, and C. T. Cowell Importance of Estrogen on Bone Health in Turner Syndrome: A Cross-Sectional and Longitudinal Study Using Dual-Energy X-Ray Absorptiometry J. Clin. Endocrinol. Metab., January 1, 2004; 89(1): 193 - 199. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. K. Bakalov, L. Axelrod, J. Baron, L. Hanton, L. M. Nelson, J. C. Reynolds, S. Hill, J. Troendle, and C. A. Bondy Selective Reduction in Cortical Bone Mineral Density in Turner Syndrome Independent of Ovarian Hormone Deficiency J. Clin. Endocrinol. Metab., December 1, 2003; 88(12): 5717 - 5722. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. H. Gravholt, R. W. Naeraa, K. Brixen, K. W. Kastrup, L. Mosekilde, J. O. L. Jorgensen, and J. S. Christiansen Short-Term Growth Hormone Treatment in Girls With Turner Syndrome Decreases Fat Mass and Insulin Sensitivity: A Randomized, Double-Blind, Placebo-Controlled, Crossover Study Pediatrics, November 1, 2002; 110(5): 889 - 896. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||
) and control women (
). The solid line represents the linear regression function of the TS group across the age range, and the dashed line represents the control group. Middle, vBMDneck vs. age in women with TS (