INTRODUCTION
Bone mineral content (BMC) and bone mineral density (BMD) increase as the skeleton expands during growth. [1,2] Therefore, the peak bone mass is a determining factor in bone health and osteoporosis development in adult life. [3,4] The prevalence of osteoporosis and fragility fracture risk increases with age and substantial differences between populations and sex have been shown due to changes in bone remodeling, regulated by biomechanical and hormonal factors.[5,6]
BMD is usually measured by dual energy X-ray absorptiometry (DXA), a non-invasive technique, which is currently the gold standard for osteoporosis diagnosis and monitoring.[7,8] The advantages of DXA include high precision, device availability, very low radiation exposure, and quick and easy BMD assessment. However, DXA indicates only BMC and BMD, but it does not entirely reflect the bone quality, considering that only 40% to 50% of patients with hip fractures have osteoporosis by DXA criteria.[9] Therefore, it is apparent that other parameters, such as bone geometry and microarchitecture and 3-dimensional (D) distribution of the mineralized material, are determining factors of bone quality and strength.[10,11] Quantitative computed tomography (QCT) provides accurate measurements of these structural parameters, and cortical and trabecular bone separately. Also, hip vBMD as measured by QCT it would be useful for monitoring age- and treatment-related changes and to predict hip fractures as well as hip BMD measured by DXA.[12] However, single-energy QCT image assessment is highly influenced both by the degree of marrow fat at baseline and changes in marrow fat over time while the dual energy QCT provides a more accurate estimation of both bone mass and mineral content.[13] In addition, due to higher cost and radiation dose, and lack of reference values, the use of QCT in clinical practice for osteoporosis assessment remains limited.[14] On the other hand, the proximal femur 3D reconstruction (3D-DXA) using standard hip DXA scans has demonstrated a good correlation with QCT for allowing the independent analysis of both trabecular and cortical compartments.[15-17]
The main purpose of this study was to evaluate the determinants influencing the volumetric density of trabecular and cortical bone within the proximal femur, concomitant with age- and sex-related differences, employing proximal femur 3D reconstruction (3D-DXA). Additionally, we aimed to determine the reference intervals for integral vBMD, recognizing the significance of regional data.
METHODS
1. Study population
A total of 1,001 women and 353 men between 20 and 90 years old were included as volunteers from eight centers in Argentina. We included clinically healthy adult subjects, without pathological conditions or treatments that could influence bone metabolism. Subjects with a previous history of fragility or low-energy fractures, a history of surgery or metal prostheses in the region of interest, primary ovarian insufficiency (before the age of 40), and pregnant women were excluded. Subjects were excluded if their body mass exceeded the DXA scan table weight limit or were unable to undergo BMD measurements. Data were collected from 2018 to 2020 in 1,354 individuals from Argentina.
The research was carried out according to the Declaration of Helsinki for Human Subjects and was approved by the Ethics Committee of Rosario Center for Perinatal Studies (Argentina, N°4/18). Informed consent was obtained from each individual participant to be included in the study. Each participant was identified by a number, to keep their identity confidential.
This study was endorsed by the Argentinian Society of Osteology and Mineral Metabolism (AAOMM).
2. BMD assessment and proximal femur 3D reconstruction (3D-DXA)
BMD (g/cm2) was measured by DXA on the left femoral neck (FN) and total hip (TH) in Lunar Prodigy DXA (GE Healthcare, Madison, WI, USA) and Hologic Discovery Wi (Hologic Inc., Bedford, MA, USA) devices according to recommendations provided by both the manufacturer and the International Society for Clinical Densitometry (ISCD). To ensure data quality, daily standard calibration was performed (coefficient of variation <1%).[18] The 3D analysis was performed with 3D-Shaper software (v2.10; Galgo Medical, Barcelona, Spain). To state it briefly, the software uses an algorithm based on a 3D statistical shape and density model of the proximal femur, built from a database of QCT scans to generate a patient-specific 3D model using a standard hip DXA scan.[15,16] Following this, the cortex is segmented, and the trabecular bone portion is extracted. More details about the algorithm can be found elsewhere. [17] The following parameters were considered: integral (cortical+trabecular) volumetric BMD (integral vBMD in mg/cm3), trabecular vBMD (trab vBMD in mg/cm3), and cortical surface BMD (sBMD in mg/cm2). sBMD was obtained with the following formula: cortical volumetric density (mg/cm3) * cortical thickness (cm). Coefficients of variation for cortical sBMD, trab vBMD, cortical vBMD were 1.5%, 4.5%, and 1.7%, respectively.
The primary outcome measure of this study was the age- and sex-related differences in vBMD of the proximal femur, specifically focusing on trabecular and cortical compartments using 3D-DXA. The secondary outcome measures included establishing reference intervals for integral vBMD, analyzing the annual rate of change (age slope) in bone density parameters (integral vBMD, trab vBMD, cortical sBMD) across different age groups, comparing bone density parameters between men and women and identifying the age thresholds where significant changes begin to occur, and evaluating the correlation between anthropometric variables (e.g., body mass index [BMI], body weight) and bone density parameters.
3. Data analysis
The Shapiro-Wilk and Bartlett tests were used to assess normality and equal variances, respectively, and parametric or non-parametric tests were used, as appropriate. Continuous variables were expressed as mean±standard deviation. Differences were considered significant if P-value was less than 0.05. Pearson correlations were utilized to compare the femur variables with anthropometric measures. The percentage changes in the femur variables between ages 20 and 90 were calculated.
The locally weighted regression smooth scatter plot (LOESS) [19,20] was adopted to describe the relationship between age and integral vBMD, trab vBMD, and cortical sBMD, and to explore the potential age threshold. Subsequently, the exact cut-point values of age were determined by using segmented regression.[21] The age slopes were estimated and compared using a linear test model for women and men, considering the age cut-off points obtained for each femoral variable. In women, the adjustment was made according to menopausal status. A Z-test suggested by Harris and Boyd [22] was adopted to evaluate if the reference interval should be kept separate by age. To determine the reference interval for integral vBMD, the extreme percentiles (2.5 and 97.5) were calculated and adjusted by gender.
The LOESS model fitting and segmented regression were conducted by the R Statistical Software (version 4.2.3; The R Foundation for Statistical Computing, Vienna, Austria). GraphPad Prism 8 (GraphPad Software Inc., San Diego, CA, USA) was used for other statistical analyses. For all analyses, a P-value of less than 0.05 was considered statistically significant.
RESULTS
We evaluated 1,354 subjects: women (N=1,001; 73.9%) and men (N=353; 26.1%) between 20 and 90 years without conditions affecting bone health. The mean age, BMI, and 3D parameters of the whole group and according to sex are shown in Table 1. Men were younger and showed significantly higher BMI and 3D-DXA values compared to women (P<0.05).
Age negatively correlated with femoral 3D-DXA parameters in both women and men, while anthropometric variables showed a positive correlation (Table 2).
The age-related volumetric density differences in the trabecular and cortical bone of the proximal femur are shown in Table 3. Comparing men versus women, the latter showed higher absolute change and percentage of change indicating greater bone loss.
The LOESS analysis revealed a sharper decline in 3D-DXA parameters among postmenopausal women, whereas men exhibited a decrease starting around the ages of 40 to 50, except for cortical sBMD (Fig. 1).
Cut-off values for each 3D-DXA parameter were determined by segmented regression. Table 4 shows the age threshold for 3D-DXA parameters and the age slope before and after these thresholds indicating the mean annual decreases of bone loss.
In addition, we investigated whether the integral vBMD reference interval should be divided by age range. Specifically, a Z-test was performed using each decade of life as cut-off points (Table 5). The Z-test results indicated the necessity to stratify the reference interval by age range. Finally, we established the integral vBMD reference interval considering the results of Z-tests (Table 6).
DISCUSSION
In recent years, the proximal femur 3D reconstruction (3D-DXA) using standard hip DXA scans has been introduced in bone assessment including drug treatments, rheumatoid arthritis patients, and football players, swimmers, and sedentary controls among others.[16,23-28] As expected, age negatively correlated with femoral 3D-DXA parameters in both women and men, while anthropometric variables showed a positive correlation. Also, men had significantly higher values in 3D-DXA parameters. However, there are no reference values from healthy adult subjects reported according to age and sex. In addition, we considered it relevant to assess the behavior of these parameters in different populations because 3D-DXA parameters are obtained from DXA scans and heterogeneity between regions and populations has been reported for BMD values. Previous BMD reference values had been proposed in American countries for the United States, Mexico, and Chile populations.[6,29,30] Moreover, it has been shown that the reference data provided by manufacturers for a U.S. Hispanic population may lead to underestimations in the Mexican population.[29]
Therefore, we reported the age and sex-related volumetric density differences in the trabecular and cortical compartments of the proximal femur and reference curves from healthy adult subjects from Argentina using proximal femur 3D-DXA.
Previous studies indicated that after reaching its peak, bone mass remains relatively stable until about the 3er-4th decade of life and from that decade on it begins to decline according to age and sex.[6,31] In this study, we found a significant decrease between 20 and 90 years of age in integral vBMD in both sexes (women, −23.1%, men, −16.6%) due to higher annual bone loss. While in men the mean annual bone loss measure by integran vBMD was 1.156 mg/cm3 and 1.241 mg/cm3 before and after 41 years of age, women showed 1.071 mg/cm3 and 1.842 mg/cm3 of annual bone loss before and after 47 years of age being coincident with age of menopause. In addition, we demonstrated that integral vBMD should be analyzed by age (Table 5) and reported the reference intervals (Table 6). These findings are consistent with other reports showing BMD and BMC differences by DXA across the life span [6,29,30,32-35] including data from Argentina.[36] However, as far as we know, this is the first study showing age- and sex-related volumetric density differences in trabecular and cortical bone of the proximal femur using 3D-DXA.
Bone loss in integral vBMD was mainly due to trabecular bone decrease in both sexes. The trab vBMD decreased −33.4% in women and −27.7% in men between 20 and 90 years. However, the behavior of trab vBMD bone loss was different according to sex. While in women is consistent with menopause and integral vBMD showing a greater slope after 49 years (annual bone loss 0.797 mg/cm3 and 1.159 mg/cm3 before and after 49 years), in men the trab vBMD bone loss does not show a significant change at a certain age, being more constant and more evident up to the 60s. These data are consistent with a previous study that found the highest value for lumbar spine BMD within 30 to 39 years old women, while in men was found earlier (D20).[37] A previous report in Italian premenopausal women, using peripheral QCT (pQCT), found that both trabecular and total BMD showed a linear decline with aging, decreasing by an overall slope of −1.28 and −0.55 mg/cm3 per year for total and trabecular BMD measurements, respectively.[38] Because of its high turnover rate, the trabecular bone is the preferred site for early detection of bone mineral changes.
Regarding cortical bone, we found a significant decrease in women (−9.9% between 20 and 90 years) with a higher increase in the slope (−0.437 mg/cm2) of cortical sBMD bone loss after 47 years, while in men the cortical bone loss is almost flat without significant changes throughout life. In another study using hip QCT, it reported an annual decrease in cortical sBMD ranging from −0.83 mg/cm2 to −1.35 mg/cm2 in a cohort of 243 postmenopausal women. [39]
In summary, for women, the age thresholds for integral vBMD, trab vBMD, and cortical sBMD are around 47 to 49 years, aligning closely with the typical age of menopause. This indicates that the bone loss accelerates significantly after menopause, highlighting the impact of hormonal changes on bone health.
QCT of vertebra trabecular bone has shown a good capability for assessment of age-related bone loss and discrimination of osteoporotic vertebral fractures.[40-43] However, QCT is not a standardized technique for osteoporosis diagnosis and requires additional radiation exposure while 3D-DXA has demonstrated the ability to serve as a good surrogate for QCT assessing cortical and trabecular bone compartments independently and the effects of osteoporosis drugs.[17,25,44]
The possibility of evaluating cortical and trabecular tissue separately is an advantage over BMD by DXA, while the low dose radiation, low cost, and higher device equipment availability are advantages over QCT. Despite the contribution of cortical and trabecular bone loss to fragility fracture risks are still under debate,[45] some studies suggested that cortical thinning contributes to the increase in fracture risk with aging,[46,47] whereas others, using QCT, reported that both cortical and trab vBMD are involved. [48,49]
It is important to consider how the age- and sex-related differences in volumetric BMD (vBMD) of the proximal femur, assessed using 3D-DXA, can influence the evaluation of osteoporosis risk and fractures in different populations. vBMD provides a 3D assessment of bone density, which allows for a more accurate representation of bone strength and quality. Age- and sex-related differences in vBMD could be used to develop more precise risk assessment models, improving the early detection of osteoporosis in postmenopausal women and older men. Establishing reference intervals specific to the Argentine population allows for greater diagnostic accuracy being particularly relevant in the context of demographic and genetic differences that can affect bone density. By better understanding the impact of menopausal status and other anthropometric factors on bone health, healthcare professionals can design more effective preventive interventions.
A limitation of this study is its cross-sectional design and other factors that affect bone mass, like ethnicity, physical activity, smoking or alcohol consumption, which were not considered. Also, we did not perform specific tests to evaluate intra- and inter-operator variability for the 3D-DXA measurements. This represents a limitation, as understanding operator-related variability is important for validating the reproducibility and reliability of the measurements. Future studies should include assessments of intra- and inter-operator variability to further validate the robustness of the 3D-DXA methodology. In addition, the 3D-Shaper software successfully reconstructed the geometry of the proximal femur from 2D-DXA scans, achieving average surface distances smaller than 2 mm for all femora, though it slightly underestimated the volume.[50]
Further there is no cross-calibration between different DXA scanners and particularly no cross-calibration between Hologic and Lunar. However, the use of both equipment should not be considered a limitation because similar least significant changes (LSC) from areal BMD measurements using 3D-DXA in total femur for both Hologic (Discovery W) and Lunar (iDXA) devices (3.1% and 3.7%, respectively) has been reported. This is according to ISCD recommendations regarding the minimum acceptable precision (LSC, 5.0% at TH and 6.9% at FN).[16] The TH BMD in repeated images was consistently below 0.5% when six to ten participants were considered, providing the minimal effect size that can be detected using the 3D-DXA procedure analyzed.[51]
In conclusion, we found integral vBMD age- and sex-related bone loss using proximal femur 3D-DXA with a decrease of 23.1% in women and 16.6% in men between 20 and 90 years. This bone loss was mainly due to trabecular bone decrease in both sexes. The age-related loss of cortical bone density is lower and limited to the female population.
Therefore, proximal femur 3D-DXA allowed us to show age and sex-related bone loss in our Argentina's cohort and propose reference intervals for integral vBMD.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
