Obesity is associated with higher areal bone density (aBMD) but its protective effect on the risk of fracture is controversial. We aimed to analyze bone microarchitecture and biomechanical properties in obese (OB) postmenopausal French women compared with normal weight (NW) women. A matched case-control study from the Os des Femmes de Lyon (OFELY) cohort was conducted in 63 OB women (body mass index [BMI] > 30, mean age 69 ± 8 years) age-matched with 126 NW women (19 ≤ BMI ≤ 25). Bone architecture was measured with high-resolution pQCT at the distal radius and tibia and bone strength was assessed by micro–finite element analysis (µFEA). aBMD, total body fat mass (FM) and lean mass (LM) were measured by dual-energy X-ray absorptiometry (DXA). aBMD was 15% higher at the total hip in OB compared with NW women. At the radius, OB had 13% and 14% higher volumetric total and trabecular bone densities, 11% higher cortical thickness, 13% greater trabecular number, and 22% lower distribution of trabecular separation compared with NW (p adjusted for height, physical activity, and medication use, <0.01 for all). Differences of a similar magnitude were found at the distal tibia. At both sites, µFEA showed significant higher values of bone strength in OB compared to controls. After normalizing values for individual body weight, we observed that all the parameters were relatively lower in OB compared to NW women. The increase of FM was fourfold greater than the increase of LM in OB. The effect of FM on bone parameters was more pronounced at the tibia compared to the non–weight-bearing site. Nevertheless, the coefficients of correlation were about one-half of those of LM for the biomechanical parameters. We conclude that higher absolute values of bone densities, cortical and trabecular architecture, and strength indices were not in proportion to the excess of BMI and particularly of FM in obese postmenopausal French women.
Obesity is a growing worldwide health problem with an increasing prevalence and a high impact on both mortality and morbidity.[1-4] If a low body mass index (BMI) is a well-documented risk factor for future fracture, the protective effect of obesity on the risk of fragility fracture is controversial.[5, 6] Indeed, recent studies suggest that obese women could be more exposed to some nonvertebral[7-10] and vertebral[11, 12] fractures. A study from dual-energy X-ray absorptiometry (DXA) scans using hip structure analysis showed that bone strength did not remain in proportion to weight in obese patients. Nevertheless, extensive data have shown that higher body weight or BMI is correlated with higher areal bone mineral density (aBMD) or bone mineral content (BMC)[14-16] and that a decrease in body weight leads to bone loss.[17-20] However, a particular concern in studies of obesity is the overestimation of the aBMD values due to excess fat tissue, especially at axial sites.[21, 22] Volumetric densities measured with quantitative computerized tomography at peripheral sites (pQCT) should be less affected by the thickness of fat, which is smaller in peripheral sites compared to axial sites. Moreover, this method allows analyzing the cortical and trabecular compartment separately. Several recent studies were performed in obese children or adolescents with that technology, but few studies have been conducted in adult obese women.
Obesity is predominantly due to an excess of fat mass, and the positive effect of fat mass on bone is controversial. Using pQCT, Pollock and colleagues found that late adolescent females with a high percent of fat compared with normal-percent fat had significantly lower indices of bone strength at both the distal radius and tibia. The association between lean mass with bone strength is well documented, but the respective role of fat mass and lean mass on bone microarchitecture and bone strength in adult obese women has not been thoroughly evaluated. Recent developments of noninvasive high-resolution pQCT (HR-pQCT) permit the assessment of cortical and trabecular bone microarchitecture in addition to bone strength through micro–finite element analysis (µFEA).[26, 27]
Therefore, the aim of our study was to analyze bone microarchitecture, geometry, and biomechanical properties in obese postmenopausal French women compared with normal weight women from the Os des Femmes de Lyon (OFELY) cohort at both weight-bearing and non–weight-bearing skeletal sites. We hypothesized that obese women would have higher levels of volumetric densities and we explored whether the differences of bone geometry, microarchitecture, and strength according to the excess in weight would match the expected difference. We also examined their relationships with lean mass and fat mass.
Subjects and Methods
Obesity was defined as a BMI ≥ 30 kg/m2. Sixty-three obese (OB) postmenopausal women (mean age 69 ± 8 years) were randomly age-matched at the 14th annual follow-up of the OFELY study with 126 normal weight (NW) postmenopausal women from the same cohort. Briefly, OFELY is an ongoing prospective study of the determinants of bone loss in 1039 volunteer women, recruited between February 1992 and December 1993, 31 to 89 years of age, randomly selected from the affiliates of a large health insurance company (Mutuelle Générale de l'Education Nationale) from the Rhône district of France (ie, Lyon and its surroundings), with an annual follow-up.[28, 29] The 14th annual follow-up was the first visit of the study with an evaluation of bone architecture at the distal radius and tibia. In order to obtain a large range of age for the evaluation of bone architecture, 168 volunteer women aged 25 to 42 years—from the same insurance company—were recruited in addition to the original cohort between January 2005 and December 2006. The protocol was accepted by the local ethics committee and each of the 861 women—including 589 postmenopausal women—who came at this visit gave written informed consent. Among postmenopausal women, 63 (11%) women were obese and a matched case-control study was performed with 2 controls for each case. Controls were NW women with a BMI between 18.5 and 24.99 kg/m2.
Bone microarchitectural measurement
The nondominant forearm (or the nonfractured forearm in the case of prior fracture) and distal tibia were scanned using a 3D HR-pQCT device (XtremeCT; Scanco Medical AG, Brüttisellen, Switzerland). This system acquires a stack of 110 parallel CT slices with an isotropic voxel size of 82 µm. The details of the acquisition and analysis has been described. A new automatic segmentation script was used to differentiate cortical bone from trabecular bone,[31, 32] which permitted the measurement of cortical thickness using a direct transformation method. This new method also permitted the assessment of cortical porosity, which is not available on the standard microarchitecture evaluation proposed by the manufacturer.
The HR-pQCT parameters used in our analyses were the following: total, cortical, and trabecular bone cross-sectional area (Tt.Ar, Ct.Ar, and Tb.Ar, mm2); volumetric bone density for total, cortical, and trabecular bone (Tt.vBMD, Ct.vBMD, and Tb.vBMD, mg/cm3); cortical thickness (Ct.Th, µm) and porosity (Ct.Po, %); trabecular number (Tb.N, mm−1), thickness (Tb.Th, µm), and intraindividual distribution of separation (Tb.Sp.SD, µm).
Bone strength was assessed by µFEA, using the modeling described in the OFELY cohort.[26, 33] Briefly, FE models of the radius and the tibia were created using software delivered with the HR-pQCT device (IPL-FE version 1.13; Scanco Medical AG). A voxel-conversion procedure was used to convert each voxel of bone tissue into an equally-sized brick element. Material properties were chosen isotropic and elastic. Cortical bone elements were assigned a Young's modulus of 20 GPa, whereas trabecular tissue elements were assigned a Young's modulus of 17 GPa, and for all elements a Poisson's ratio of 0.3 was specified. A compression test of the radius was simulated in which a load in the longitudinal direction was applied at one end while the other end was fully constrained, to simulate a fall from standing height on the outstretched hand. The µFEA-derived variables used in our study included: the percentage of load carried by the trabecular bone at the distal and proximal surface of the volume of interest (Tb Dist Load [%], and Tb Prox Load [%]), the average and SD values of the Von Mises stresses (VMS) in the trabecular and cortical bone (Tb.VMS and Ct.VMS; Tb.VMS.SD and Ct.VMS.SD, [MPa]), stiffness (kN/mm), and the estimated failure load (N).
aBMD was measured at the same visit by DXA (QDR 4500, software version V8.26a; Hologic, Waltham, MA, USA) at the anteroposterior lumbar spine and at the total hip for all women and at the nondominant forearm for 56 OB (89%) and 101 (80%) NW women. Total body fat and lean masses were measured at the same visit with the same device.
The in vivo reproducibility of DXA, expressed as the root mean square coefficient of variation (rms CV), calculated from three repeated measurements in 17 patients aged 32 to 80 years was 1.1% for the spine, 0.9% for total hip, and 1.2% for the ultradistal radius. The rms CV values for fat mass and lean mass, calculated from two repeated measurements in 26 women aged 48 to 83 years, were 1.4% and 0.8%, respectively.
At the same visit, all women completed a written health questionnaire as described, including current smoking, menopausal status, medication use, physical activity, falls, and occurrence of radiologically-confirmed low-trauma fractures. For the current analysis, physical activity was estimated by the time per week spent for sport or walk outside. Height and weight were measured with participants wearing indoor clothes and no shoes. BMI was calculated as body weight/height2 (kg/m2).
For each woman, fasting blood samples were collected before 9:00 a.m. at the same visit and kept frozen at –70°C until assayed. Serum osteocalcin (OC), intact N-terminal propeptide of type I collagen (P1NP), and β isomerized C-terminal crosslinking of type I collagen (CTX-I) were measured using an automatic test (Elecsys N-MID Osteocalcin, Elecsys P1NP, Elecsys β–Crosslaps; Roche Diagnostics, Meylan, France). Intraassay and interassay variations were <8% for all markers. Serum total 17 β-estradiol (E2) was measured by direct radioimmunoassay (CIS bio international SAS, Saclay, France). This assay shows a limit of detection of 11 pM/L and was validated by Agence Française de Sécurité Sanitaire des Produits de Santé (AFSSaPS) for the determination of very low E2 concentrations, especially in postmenopausal women. Intraassay and interassay coefficients of variation (CV) were <10%. Sex hormone–binding globulin (SHBG) was measured by an immunoradiometric assay (CIS bio international). Bioavailable E2 was calculated by multiplying total E2 concentration by the percentage of non-SHBG bound E2. Non-SHBG bound E2 was determined by differential precipitation of serum proteins after equilibration of the serum with tritiated tracer amounts of E2 (total radioactivity). At the end of the reaction, SHBG and E2 bound to it are precipitated using 50% ammonium sulfate solution. After centrifugation, radioactivity measured in the supernatant represents bioavailable E2 fraction of the total radioactivity. Serum intact parathyroid hormone (PTH) was measured by an immunoradiometric assay using two monoclonal antibodies (ELSA-PTH; CIS bio international) with a limit of detection of 0.7 pg/mL. Intraassay and interassay CV were <8%. Serum 25-hydroxyvitamin D [25(OH)D] was measured by an automatic competitive two-step chemiluminescence assay (LIAISON 25OH Vitamin D TOTAL; DiaSorin, Antony France) that recognizes both 25OH D2 and 25OH D3. Intraassay and interassay CV were <10% and the lower limit of detection was 12.5 nmol/L.
Chi-squared and Wilcoxon tests were used to compare characteristics between OB and NW women. Generalized linear models were used to compare bone parameters in both groups after adjusting for covariables. To evaluate how the bone parameters vary in proportion to the body weight, fat mass, and lean mass, those comparisons were repeated after dividing individual values successively by body weight, fat mass, and lean mass and the differences were expressed in percentage of the mean values of NW women.[13, 37] Correlation analysis between BMI, fat body mass, and lean body mass with the bone densities and microarchitecture parameters was performed by Spearman test because most of the variables were not normally distributed. Correlation analysis between levels of hormones with the bone parameters was also performed by Spearman test. All statistical analyses were performed using Stata 12 (StataCorp LP, College Station, TX, USA).
The characteristics of the OB and age-matched NW postmenopausal women are shown in Table 1. OB women were 69 ± 8 years old, their BMI was 33.4 ± 3.4, 50% higher than NW women. Their height was significantly lower. Physical activity was less and falls were more frequent in OB compared with NW women. Medication use was less frequent in OB women (11%, n = 7, hormone replacement therapy [HRT], n = 6, bisphosphonates, n = 1) compared with NW women (27%, n = 34, HRT, n = 16, bisphosphonates, n = 10, selective estrogen receptor modulators, n = 6, tibolone, n = 2), p = 0.01. aBMD was 13% to 16% higher at the lumbar spine, total hip, and radius in OB compared with NW (p < 0.0001) women. The increase of total body fat mass was fourfold greater than the increase of lean mass in OB women. The proportion of lean mass to body weight was significantly lower in OB compared with NW women (p < 0.0001). The prevalence of fragility fracture was the same in both groups (16%). There were 15 nonvertebral and 5 vertebral fractures in NW women, nine nonvertebral and one vertebral fractures in OB women.
|OB women (n = 63)||NW women (n = 126)||Difference (%)||p|
|Age (years)||68.6 ± 7||68.2 ± 7.4||0.84|
|Height (cm)||156.9 ± 5.1||159.3 ± 5.7||−1.3||0.02|
|Weight (kg)||82.3 ± 10.1||56.4 ± 5.3||+45.9||<0.0001|
|BMI||33.4 ± 3.5||22.2 ± 1.7||+50.5||<0.0001|
|Age of menopause (years)||50.7 ± 4.0||50.8 ± 3.7||0.87|
|Current smoking, n (%)||3 (5)||12 (10)||0.25|
|Physical activity (>6 hours/week), n (%)a||24 (38)||68 (54)||−16||0.04|
|Fallers (≥2 falls in <1 year), n (%)||11 (17)||9 (7)||+10||0.03|
|Calcium–vitamin D supplements||8 (13)||25 (20)||0.31|
|Medication useb||7 (11)||34 (27)||+16||0.01|
|Prevalent fractures, n (%)||10 (16)||20 (16)||0.99|
|Total body fat mass (kg)||35.6 ± 6.3||18.2 ± 3.6||+95.6||<0.0001|
|Total body fat mass (%)||42.7 ± 3.3||31.7 ± 4.6||+34.7||<0.0001|
|Total body lean mass (kg)||45.4 ± 4.7||37.2 ± 3.6||+22||<0.0001|
|Total body lean mass (%)||54.9 ± 3.1||65.1 ± 4.4||−15.7||<0.0001|
|Lumbar spine aBMD (g/cm2)||0.979 ± 0.13||0.867 ± 0.14||+12.9||<0.0001|
|Total hip aBMD (g/cm2)||0.939 ± 0.11||0.814 ± 0.11||+15.4||<0.0001|
|UD radius aBMD (g/cm2)||0.398 ± 0.06||0.342 ± 0.05||+16.4||<0.0001|
At the distal radius, OB women had 13% and 14% higher volumetric total (Tt.vBMD) and trabecular (Tb.vBMD) bone densities (p adjusted for height, physical activity, and medication use, designated p*; <0.01), 3% higher volumetric cortical (Ct.vBMD) bone density (p* = 0.04), 11% higher cortical thickness (Ct.Th) (p* < 0.01), 13% greater trabecular number (Tb.N) (p* < 0.001), and 22% lower distribution of trabecular separation (Tb.Sp.SD) (p* = 0.01) compared with NW (Table 2). Similar results were found at the distal tibia except for Ct.Th, which was not significantly different between OB and NW women (+15% for Tt.vBMD, p* < 0.0001, +15% for Tb.vBMD, p* = 0.001, +7% for Ct.vBMD, p* = 0.003, +16% for Tb.N, p* < 0.0001 and –22% for Tb.Sp.SD, p* = 0.02). Moreover, Ct.Po was 21% lower in OB compared to NW women at the tibia (p* < 0.01). At both sites, total (Tt.Ar) and trabecular area (Tb.Ar) were not significantly different whereas cortical area (Ct.Ar) was higher in OB women (p* < 0.0001) (Table 2).
|Distal radius||Distal tibia|
|OB women (n = 63)||NW women (n = 126)||Difference (%)||p*||OB women (n = 63)||NW women (n = 126)||Difference (%)||p*|
|Tt.Ar (mm2)||255 ± 44||254 ± 44||—||0.15||660 ± 95||661 ± 102||—||0.14|
|Ct.Ar (mm2)||45 ± 8||40 ± 8||+13||<0.0001||107 ± 16||96 ± 16||+11.5||<0.0001|
|Tb.Ar (mm2)||200 ± 45||205 ± 43||—||0.64||541 ± 96||552 ± 103||—||0.56|
|Tt.vBMD (mg/cm3)||299 ± 71||265 ± 64||+13||0.002||277 ± 52||241 ± 50||+14.9||0.001|
|Ct.vBMD (mg/cm3)||816 ± 68||791 ± 74||+3.2||0.04||769 ± 85||722 ± 94||+6.5||0.003|
|Tb.vBMD (mg/cm3)||155 ± 44||136 ± 38||+14||0.003||164 ± 33||143 ± 37||+14.7||<0.0001|
|Tb.N (mm−1)||1.69 ± 0.3||1.50 ± 0.3||+12.7||0.009||1.68 ± 0.3||1.45 ± 0.3||+15.9||<0.0001|
|Tb.Th (µm)||75 ± 12||75 ± 13||0.73||82 ± 12||83 ± 18||0.75|
|Tb.Sp.SD (mm)||0.28 ± 0.1||0.36 ± 0.2||−22.2||0.009||0.28 ± 0.2||0.36 ± 0.3||−22.2||0.02|
|Ct.Th (mm)||0.84 ± 0.2||0.76 ± 0.2||+10.5||0.0003||1.1 ± 0.2||1.1 ± 0.2||—||0.32|
|Ct.Po (%)||2.6 ± 1.4||3.0 ± 0.2||0.14||12.4 ± 5.8||15.7 ± 7.4||−21||0.002|
|Tb Dist load (%)||43.0 ± 8.0||43.7 ± 7.8||—||0.89||48.3 ± 7.7||48.6 ± 9.3||—||0.92|
|Tb Prox load (%)||14.0 ± 5.1||14.6 ± 5.3||—||0.70||27.8 ± 7.1||28.7 ± 8.3||—||0.97|
|Tb.VMS (MPa)||7.4 ± 1.1||8.2 ± 1.5||−9.8||<0.0001||3.2 ± 0.4||3.6 ± 0.5||–11.1||<0.0001|
|Tb.VMS.SD (MPa)||5.1 ± 1.1||5.8 ± 1.5||−12.1||0.0001||1.7 ± 0.3||2.0 ± 0.5||–15.0||<0.0001|
|Ct.VMS (MPa)||15.4 ± 2.9||17.4 ± 4.0||−11.5||0.0001||5.8 ± 0.9||6.6 ± 1.2||–12.1||<0.0001|
|Ct.VMS.SD (MPa)||4.1 ± 1.0||4.5 ± 1.3||−8.9||0.04||1.2 ± 0.2||1.4 ± 0.4||–14.3||<0.0001|
|Stiffness (kN/mm)||125.4 ± 23.6||113.4 ± 25.6||+10.6||0.0001||356.4 ± 53||310.1 ± 56||+15.0||<0.0001|
|Estimated failure load (N)||2958 ± 560||2677 ± 592||+10.5||0.0001||8441 ± 1207||7367 ± 1273||+14.6||<0.0001|
At both sites and after adjustment for height, physical activity, and medication use, µFEA showed no differences in load distribution between OB and NW women. The average and SD values of Von Mises Stresses (VMS) in the trabecular and cortical bone were significantly lower, and both stiffness and estimated failure load were significantly higher in OB compared to NW (Table 2).
All the differences in bone parameters were lower than the differences in body weight or BMI between OB and NW women (Tables 1, 2). After normalizing values for individual body weight and adjusting for physical activity and medication use, we observed that all the bone parameters were relatively lower in OB compared to NW women (Table 3). For comparison, the normalized difference was –22% for aBMD at the spine and –21% at the total hip (p < 0.0001 versus NW). All those bone parameters were even more lower in OB compared to NW women after normalizing values for body fat mass, whereas the differences were much lower after normalizing values for lean mass (Table 3).
|Distal radius||Distal tibia|
|Tt.Ar||−31.1 ± 12.0||−50.0 ± 11.2||−17.5 ± 13.5||−31.5 ± 9.2||−50.1 ± 9.4||−17.9 ± 10.6|
|Tt.vBMD||−21.9 ± 22.5||−43.2 ± 17.8||−7.3 ± 25.8a||−20.8 ± 19.3||−42.3 ± 15.7||−6.4 ± 20.9a|
|Ct.vBMD||−29.1 ± 11.0||−48.2 ± 9.8||−15.4 ± 11.5||−26.8 ± 12.8||−46.7 ± 11.6||−13.2 ± 13.1|
|Tb.vBMD||−21.4 ± 25.7||−42.7 ± 20.2||−6.6 ± 29.2a||−21.1 ± 19.1||−42.4 ± 15.7||−6.4 ± 21.3a|
|Tb.N||−22.2 ± 16.9||−43.4 ± 13.9||−7.2 ± 19.5a||−20.6 ± 15.6||−42.1 ± 13.2||−5.9 ± 17.6a|
|Ct.Th||−23.3 ± 19.3||−44.4 ± 15.0||−8.9 ± 21.1b||−31.3 ± 18.9||−49.8 ± 14.9||−18.2 ± 21.4|
|Stiffness||−23.8 ± 16.6||−44.7 ± 14.3||−9.1 ± 17.8c||−21.1 ± 13.3||−42.6 ± 12.8||−6.2 ± 13.4b|
|Estimated failure load||−23.8 ± 16.6||−44.7 ± 16.5||−9.1 ± 18.0c||−21.3 ± 12.6||−42.8 ± 12.4||−6.4 ± 12.6c|
In order to evaluate the role of lean and fat body masses in the increase of absolute values of bone densities and microarchitecture parameters in OB women, we performed correlation analysis in untreated women, showing fewer association with fat mass at the radius site compared with lean mass. At the distal tibia site, the effect of fat mass was more pronounced compared to the non–weight-bearing site. Nevertheless, the coefficients of correlation were about one-half those of lean mass for the biomechanical parameters. Total bone area was positively associated with lean mass with the same magnitude at both sites, but not with fat mass (Table 4).
|Distal radius||Distal tibia|
|Estimated failure load||0.18*||0.25**||0.13||0.32***||0.39***||0.47***||0.34***||0.54***|
Untreated OB women (n = 54) had significant lower levels of serum osteocalcin (p = 0.01) and SHBG (p < 0.0001) and higher levels of total and bioavailable E2 (p < 0.0001) compared with untreated NW women (n = 91) (Table 5). No significant correlation was found between estradiol levels with the bone microarchitecture and the biomechanical parameters in both groups.
|OB women (n = 54)||NW women (n = 91)||p*|
|CTX-I (ng/mL)||0.48 ± 0.2||0.54 ± 0.2||0.08|
|OC (ng/mL)||26.3 ± 9.7||31.3 ± 12.6||0.01|
|P1NP (ng/mL)||53.0 ± 21.3||55.6 ± 21.2||0.49|
|Total E2 (nmol/L)||31.0 ± 46.4||18.0 ± 12.6||<0.0001|
|Bioavailable E2 (nmol/L)||14.2 ± 15.8||6.9 ± 4.8||<0.0001|
|Bioavailable E2 (%)||44.3 ± 9.4||35.3 ± 9.1||<0.0001|
|SHBG (nmol/L)||38.3 ± 22.6||62.3 ± 27.1||<0.0001|
Serum PTH and 25OHD were measured 5 years earlier in 109 NW (87%) and 42 OB women (67%) belonging to the same BMI classification. The median (interquartile [IQ]) values in OB women compared with NW women were lower for 25OHD (19  versus 31  nmol/L, respectively, p < 0.0001) but not statistically higher for PTH (32  versus 31  pg/mL, respectively, p = 0.16). No correlation was found between 25OHD and PTH levels with the bone microarchitecture and biomechanical parameters in all women and in OB women.
In this case-control study, we found that OB postmenopausal women had higher vBMD and higher values of cortical and trabecular architecture compared with NW postmenopausal women. Our results also suggest that BMI has a positive influence on strength indices assessed at the radius and the tibia using FEA. Nevertheless, the increase of all parameters in OB women was lower relative to the excess of weight or BMI, which might explain a relative bone fragility. Furthermore, our results suggest that lean tissue mass has a stronger effect on bone compared to fat mass.
In our study, the increase of total vBMDs at the distal radius and tibia had the same magnitude as the increase of aBMD assessed by DXA in OB women and was predominant for Tb.vBMD compared with Ct.vBMD. Differences in trabecular and cortical bone in obesity have been recently reported in a pQCT study. In that study of 211 women between 25 and 71 years of age measured at the distal tibia with pQCT, Sukumar and colleagues showed a positive association between BMI and Tb.vBMD, Ct.Th, and Ct.Ar, in agreement with our findings. In contrast, they reported a negative association between BMI and Ct.vBMD that we did not find. Nonetheless, that association was only observed in premenopausal women with both severe obesity and high levels of PTH, and no woman from our population matched those conditions. Moreover, the differences of the region of interest of the distal tibia between QCT and HRpQCT might explain some discrepancies.
We found that biomechanical parameters were also greater in OB compared with NW women, particularly at the weight-bearing tibia. The apparently better bone density, architecture, and biomechanical properties in OB women can be the result of an adaptive response to the higher loading of the skeleton. Nonetheless, all parameters were relatively lower in proportion to the excess of weight. Beck and colleagues analyzed indices of bone strength extracted from DXA scans using hip structure analysis in postmenopausal non-Hispanic women according to BMI. They showed that femur BMD, cross-sectional area, and bending strength declined relative to body weight in higher BMI categories. Thus, obesity may not confer greater protection against fracture in those individuals, given that the mechanical force during a fall is proportional to body weight. Nonetheless, most of the evidence suggests a protective effect of obesity on hip fractures. That may suggest that the protection is not entirely through bone density and structure. Also, the pattern of sites of fracture may differ in obese and non-obese women; eg, obese women may be more prone to humerus and leg fracture than hip fracture.
The effect of body weight on bone density is attributable to both lean tissue mass and fat mass. For some authors, fat mass is the stronger and more consistent predictor of bone density in postmenopausal women. Thus, fat mass is positively correlated with DXA-assessed bone measurements at weight-bearing skeletal sites, consistent with our study, showing positive associations between fat mass and bone parameters at the weight-bearing tibia. Fat mass has been shown to stimulate bone growth by direct mechanical actions from increased load and by increased production of hormones associated with bone formation, such as insulin from the pancreatic cells, and estrogen and leptin from adipocytes.[39-41] The increase of estrogens by adipose tissue is the result of several mechanisms, including the conversion of androgens to estrogen, the storage capacity of steroid hormones, and the alteration of the binding capacity of estrogen and SHBG. In agreement with this, we observed lower levels of SHBG and higher levels of total and bioavailable estradiol in OB women compared with NW women.
The positive effect of fat mass at non–weight-bearing sites is less evident. In a study of Chinese and white subjects, the authors showed a positive correlation between total body fat mass and bone mass, but when the mechanical loading effect of total body weight was removed, fat mass was negatively correlated with bone mass, suggesting a detrimental effect on bone. In our study, we observed no or slight association between parameters of bone geometry, volumetric densities, or strength evaluated at the distal radius with fat mass. These findings could be explained by the lack of direct mechanical effect of fat and the thinner soft tissue at this site. This detrimental effect of fat mass on bone may stem from decreased production of adiponectin in obesity. Adiponectin is a cytokine secreted by adipocytes that has been shown to be positively correlated with bone formation and negatively with BMD.
Conversely, we observed at both sites a positive association of most of the bone parameters with lean mass. Nevertheless, coefficients of correlations were greater at the weight-bearing site compared with the non–weight-bearing site for biomechanical parameters. The role of lean mass on bone strength is well documented. Obesity is the result of an excess of fat mass rather than lean mass and we observed that the proportion to total weight of lean mass was lower in OB compared with NW women. Because the positive effect of weight on bone appears to be mainly due to lean mass, this could explain why the changes in OB women are not in proportion to the excess of weight. Moreover, we observed that the poorer bone microarchitecture and strength—proportionally to weight—resulted much more from the excess of fat mass than lean mass. In a longitudinal study of overweight children measured with pQCT at the tibia, the authors showed that although overweight children seemed to be at an advantage in terms of absolute bone strength, bone strength did not adapt to the excess of body fat. Rather, bone strength was adapted to the greater muscle area in those children. In another study with pQCT, young girls with higher fat body mass relative to body mass had markedly diminished vBMD, geometry, and bone strength at metaphyseal and diaphyseal sites of the femur and tibia, contrasting with positive associations between muscle density and indices of bone strength. In the study of Sukumar and colleagues, lean mass explained greater variance than fat mass in geometric and strength indices of the distal tibia. Altogether, those findings suggest that the excess of fat mass induces an increase of the load on the skeleton with an adaptation of bone by increasing muscle mass and bone strength. However, because obesity is associated with a lower increase of lean mass compared to fat mass, the increase of strength is not adapted to the increase of fat mass.
As previously reported, we found that obesity was associated with lower physical activity, and more falls.[16-19] The risk of falling may be increased in obese subjects as a result of reduced muscular strength and mobility. Although greater soft tissue padding may reduce skeletal trauma following a fall, poorer protective responses to falling and the higher impact of the fall owing to high body weight may offset this potential benefit. Despite this, the prevalence of fractures was low in our study and not different in both groups.
We also found, as have others, lower levels of osteocalcin and circulating 25(OH)D.[23, 49, 50] In contrast, we did not find significant higher serum concentrations of PTH,[51, 52] probably because of the relatively low number of OB women with available measurements. The lower bone turnover could be the result of the higher levels of estrogens in OB women contributing to the better values of bone parameters. Nevertheless, we did not find significant negative association between osteocalcin and estradiol levels in OB women (data not shown). Several studies suggest that the lower serum 25(OH)D and higher PTH concentrations in obesity is the consequence of an increased storage of 25(OH)D in adipose tissue.[50, 52]
We acknowledge several limitations of the study. First, the cross-sectional design does not allow determination of a causal relationship between obesity and bone characteristics. In addition, women were French community-dwelling white volunteers, and our findings may not be generalized to other populations. In particular, the prevalence of obesity is lower in France compared with North America and we observed in our cohort a relatively low prevalence of morbid obesity (BMI > 35, n = 16). Bone microarchitecture was measured at the appendicular skeleton, and we cannot rule out that different outcomes might be observed at other skeletal sites. Finally, larger studies (thus with a lower effect of the variability of biochemical values) of OB women may explain the mechanisms leading to better bone microarchitecture and strength.
In conclusion, OB French women had higher volumetric bone densities, better cortical and trabecular architecture, and greater strength indices evaluated at weight-bearing and non–weight-bearing sites compared with NW women. Nevertheless, the increase of absolute values of all bone parameters was not in proportion to the excess of weight or BMI, leading to relative bone fragility. Our findings suggest that the role of lean mass has a stronger positive effect on bone compared to fat mass.
NV has become an employee of Scanco Medical AG since the end of this study. All other authors state that they have no conflicts of interest.
We thank Annick Bourgeaud, Sylviane Ailloud, and Yvonne Varillon for excellent technical assistance.
Authors' roles: Conception and design: ESR, RC. Acquisition of data: ESR, SB, NV, BC. Analysis and interpretation of data: ESR, SB, NV, BC. Participation in drafting manuscript or revising it critically for important intellectual content: ESR, SB, NV, BC, RC. Approving final version of manuscript: ESR, SB, NV, BC, RC. Approving final version of manuscript: ESR, SB, NV, BC, RC. ESR accepts responsibility for the integrity of the data analysis.