Relative contributions of lean and fat mass to bone strength in young Hispanic and non-Hispanic girls

Abstract

Background

With the high prevalence of childhood obesity, especially among Hispanic children, understanding how body weight and its components of lean and fat mass affect bone development is important, given that the amount of bone mineral accrued during childhood can determine osteoporosis risk later in life. The aim of this study was to assess the independent contributions of lean and fat mass on volumetric bone mineral density (vBMD), geometry, and strength in both weight-bearing and non-weight-bearing bones of Hispanic and non-Hispanic girls.

Methods

Bone vBMD, geometry, and strength were assessed at the 20% distal femur, the 4% and 66% distal tibia, and the 66% distal radius of the non-dominant limb of 326, 9- to 12-year-old girls using peripheral quantitative computed tomography (pQCT). Total body lean and fat mass were measured by dual-energy x-ray absorptiometry (DXA). Multiple linear regression was used to assess the independent relationships of fat and lean mass with pQCT bone measures while adjusting for relevant confounders. Potential interactions between ethnicity and both fat and lean mass were also tested.

Results

Lean mass was a significant positive contributor to all bone outcomes (p<0.05) with the exception of vBMD at diaphyseal sites. Fat mass was a significant contributor to bone strength at weight bearing sites, but did not significantly contribute to bone strength at the non-weight bearing radius and was negatively associated with radius cortical content and thickness. Bone measures did not significantly differ between Hispanic and non-Hispanic girls, although there was a significant interaction between ethnicity and fat mass with total bone area at the femur (p=0.02) and 66% tibia (p=0.005) as well as bone strength at the femur (p=0.03).

Conclusion

Lean mass is the main determinant of bone strength for appendicular skeletal sites. Fat mass contributes to bone strength in the weight-bearing skeleton but does not add to bone strength in non-weight-bearing locations and may potentially be detrimental. Bone vBMD, geometry, and strength did not differ between Hispanic and non-Hispanic girls; fat mass may be a stronger contributor to bone strength in weight-bearing bones of Hispanic girls compared to non-Hispanic.

1. Introduction

Osteoporosis is a major public health concern, especially in women, that has its origins in adolescence.[1, 2] Approximately 95% of adult bone mineral is achieved by the end of adolescence with almost half of this accrual occurring during the two years surrounding peak linear growth.[3] Although a large proportion of adult peak bone mass is genetically determined, up to 40% can be influenced by lifestyle factors.[3] This fact suggests that the period of rapid growth and bone accrual during adolescence is a window of both opportunity and vulnerability for optimizing bone strength and decreasing risk of osteoporosis later in life.[3]

Body weight, with its fat and lean components, is a major determinant of bone mineral content (BMC).[4] With the high prevalence of childhood obesity[5], the role of body weight on bone development and strength has become an increasingly important area of research. Bone adapts its strength to the strains placed upon it from gravitational impact and muscle forces.[3] It has been suggested that the bones of heavier subjects have greater mass and strength since they are subjected to greater mechanical loads compared to the bones of normal weight subjects.[4] However, the relative contributions of lean and fat mass to bone strength during childhood are unclear. While lean mass has been consistently found to have a strong positive association with bone strength[3], the effect of fat mass on bone remains controversial, with reports of augmented[6, 7], decreased[8], or no significant effect of fat mass on bone [9–11].

Many past studies have relied on the use of dual-energy x-ray absorptiometry (DXA), which provides 2-dimensional (areal) projections of the material properties of bone (i.e. BMC and areal bone mineral density (aBMD)).[12] However, bone strength, and consequently risk of fracture, is determined by both the material properties (i.e. mass and density) and the geometric properties of bone (i.e. size and shape).[13] Indeed, bone size has been shown to make a larger contribution to bone strength than bone density in human cadaver models.[14]

The development of peripheral quantitative computed tomography (pQCT) provides a low-radiation imaging alternative to DXA for measuring bone.[15] Unlike DXA, pQCT is a 3-dimensional technique that can provide an accurate estimate of volumetric bone mineral density (vBMD) in both cortical and trabecular bone compartments. In addition, pQCT provides site-specific geometric data for the appendicular skeleton which can be combined with vBMD to calculate estimates of bone strength and fracture risk.[15] Few studies have used pQCT to assess the independent effects of lean mass and fat mass on bone geometry and strength in children[13, 16–20], and even fewer studies have assessed the relationships among lean mass and fat mass with bone strength at both weight-bearing and non-weight-bearing skeletal sites.[13, 18]

In addition to the methodological limitations of past studies, the potential race and ethnicity-related differences in the relationships of lean and fat mass with bone geometry and strength has not been well characterized. Previous pQCT studies have enrolled predominantly non-Hispanic white children, with mixed samples of both males and females.[13, 16–20] No study to our knowledge has assessed the association of lean and fat mass with bone strength using pQCT in Hispanic girls. Given that women have a higher risk for developing osteoporosis in their lifetime[2] and Hispanic children have a higher prevalence of obesity compared to similarly aged non-Hispanic white children[21], the effect of a higher body weight and its composition is an important issue to address in this population.

The aim of this study was to utilize the advanced imaging capability of pQCT to assess the independent contributions of total body lean and fat mass on bone density, geometry, and strength in both weight bearing (tibia and femur) and non-weight bearing (radius) metaphyseal and diaphyseal skeletal sites of 9-to 12-year-old Hispanic and non-Hispanic girls.

2. Methods

2.1 Study population

Three hundred and forty-four girls aged 9–12 years were recruited from local schools, pediatric clinics, and wellness community events in Tucson, Arizona to participate in the Soft Tissue And Bone Development in Young GiRls (“STAR”) study, which was designed to assess the effects of adiposity and related metabolic risk factors on bone development in girls. The study protocol was approved by the University of Arizona Human Subjects Protection Committee. Written informed consent was obtained from all participants and their parents or legal guardians. Exclusion criteria included: diagnosis of diabetes, taking any medications that alter body composition, physical disability that limits physical activity, and learning disability that limited completion of questionnaires or otherwise made the participant unable to comply with assessment protocols. Following informed consent, participants’ guardians were asked to complete a health history questionnaire, which included questions on the participants’ ethnicity (Hispanic or non-Hispanic) and race (white/Caucasian, black/African American, American Indian/Alaska Native, Asian).

2.2 Anthropometry

Anthropometric measures were obtained according to standardized protocols, which have previously been described.[22, 23] In brief, body mass was measured to the nearest 0.1 kg using a calibrated scale (Seca, Model 881, Hamburg, Germany). Standing and sitting height were measured at full inhalation to the nearest mm using a stadiometer (Shorr Height Measuring Board, Olney, MD). Leg length was estimated by subtracting sitting height from standing height. Body mass index (BMI) was calculated as weight (kg) divided by height (m) squared. Based on CDC growth charts, BMI percentiles specific for age and gender were used to categorize girls as either underweight (<5^th percentile), healthy weight (≥5th and <85th percentiles), overweight (≥85th and <95th percentiles), or obese (≥95th percentile).[24]

2.3 Physical Maturation

Maturation was assessed several ways since there can be important differences in maturation in children of the same chronological age, especially during the pre-pubertal and pubertal years.[25] Using pictures illustrating the Tanner stages of pubertal maturation[26] girls rated their breast and pubic hair development as well as self-reported menarcheal status. Maturation was also assessed using maturity offset given the potential limitations of self-reported Tanner staging in children.[27, 28] Maturity offset is an estimate of years from peak height velocity that is determined from age and cross-sectional anthropometric measures (height, weight, sitting height, and leg length) using the Mirwald equation (Maturity Offset for females= −9.376 + 0.0001882 × Leg Length and Sitting Height interaction + 0.0022 × Age and Leg Length interaction + 0.005841 × Age and Sitting Height interaction − 0.002658 × Age and Weight interaction + 0.07693 × Weight by Height ratio).[25] The maturity offset has been shown to explain 89% of the variance in years from peak height velocity.[25] After peak height velocity is reached maturity offset is positive, while a negative maturity offset represents years before peak height velocity.

2.4 Dual energy X-ray absorptiometry (DXA)

Measures of whole-body fat mass and lean mass were obtained from dual energy x-ray absorptiometry (DXA) using GE/Lunar Radiation Corp (Madison, WI) following standard subject positioning and data acquisition protocols on the Prodigy model using software version 13.60.033 (n=287) and iDXA model with software version 16.20.059 (n=57) using enhanced analysis mode. One certified technician performed all DXA scan analyses. The DXA was calibrated daily according to manufacturer guidelines. The within-subject variation for soft tissue in our laboratory has been previously reported.[29, 30]

2.5 Peripheral quantitative computed tomography (pQCT)

Bone strength was assessed in cortical (diaphyseal) weight bearing (tibia and femur 66% and 20% sites, respectively, relative to the distal growth plates) and non-weight bearing sites (radius 66%) and a trabecular (metaphyseal) weight bearing site (tibia 4%) of the non-dominant limb using the STRATEC, XCT 3000 pQCT (Medizintechnik GmbH, Pforzheim, Germany, Division of Orthometrix; White Plains, NY). All pQCT scans were analyzed using Stratec XCT software, Version 6.0 and operators were trained for pQCT data acquisition and analyses following guidelines provided by Bone Diagnostic LLC. (Spring Branch, TX). A detailed description of the instrument and imaging processing and analysis protocols used in our laboratory has been previously published, as have coefficients of within-subject variation for pQCT bone measurements.[17]

In brief, pQCT slice thicknesses were 2.3 mm and voxel sizes and scanner speed were set at 0.4 mm and 25mm/sec respectively. Contour, Peel, and Cort modes, as described in the Stratec XCT software manual,[31] were employed to obtain measures of bone geometry and material properties. At diaphyseal sites, total bone area (Tot A, mm²) was obtained using Contour mode 1 (710 mg/cm³) and measures of cortical vBMD (Cort vBMD, mg/cm³), cortical BMC (Cort BMC, mg/mm), cortical area (Cort area, mm²), endosteal circumference (EC, mm), periosteal circumference (PC, mm), and cortical thickness (Crt Thk, mm) were obtained using Cort mode 2 (710 mg/cm³). At the metaphyseal site, total bone area (Tot A, mm²) and total vBMD (Tot vBMD, mg/cm³) was determined using Contour mode 3 (169 mg/cm³) and trabecular vBMD (Trab vBMD, mg/cm³) and trabecular area (Trab area, mm²) were measured using Peel mode 4 (650 mg/cm³) with a 10% peel to ensure that only trabecular bone was included in the region of interest.

Using these pQCT measures of bone density and geometry, estimates of bone strength were calculated using previously described methods.[32, 33] To estimate the resistance of diaphyseal bone to bending and torsional loading, a strength-strain index (SSI, mm³) that assesses both the structural and material properties of bone was calculated for diaphyseal sites, as described by Shedd et al.[32], which has been shown to account for up to 80% of the variance in bending failure load in human tibias [33]:

$$\text{SSI}=\mathrm{\sum }(({{\mathrm{d}}_{\mathrm{z}}}^2\times \text{Av})\times (\text{Cort}\text{vBMD}/\text{ND}))/{\mathrm{d}}_{max,})$$

where, d_z is the distance of voxel from the center of gravity, Av is the area of the voxel (mm²), Cort vBMD is the cortical bone density (mg/cm³), ND is the estimated normal physiological bone density (1200 mg/cm³), and d_max is the maximum distance of a voxel from the center of gravity.[32]

The ability of weight bearing metaphyseal bone to withstand a compressive force was estimated by calculation of the bone strength index (BSI, mg/cm⁴) at the 4% tibia, which has been shown by Kontulainen et al.[33] to account for 85% of the variance in compressive failure load in human tibias:

$$\text{BSI}=\text{Tot}{\text{vBMD}}^2\times \text{Tot}\mathrm{A}$$

where, Tot vBMD is the total bone density (mg/cm³), and Tot A is the total bone area (mm²).

The pQCT XCT analysis software was additionally used to obtain muscle cross-sectional area at the 20% femur and 66% tibial and radial sites.

2.6 Physical Activity Assessment

Physical activity was assessed using Actigraph GT3X+ (Pensacola, FL) accelerometers. All girls were instructed to wear the accelerometer on their hip for seven consecutive days. The accelerometers were initialized for data collection at a 30hz frequency. Data were saved in 60-second epochs with the “low frequency extension” option selected. Daily moderate-to-vigorous physical activity performed during the 7-day wear period was estimated using algorithms and cut-points developed by Evenson et al..[34]

2.7 Dietary Assessment

Dietary energy and nutrient intakes (i.e. vitamin D (IU), calcium (mg/day)) were assessed using the semi-quantitative Harvard youth/adolescent questionnaire, a self-administered food-frequency questionnaire with questions on 131 food items that has been validated in children and adolescents.[35] Participants filled out the questionnaire with assistance from parent(s)/guardian(s). Trained study staff reviewed the questionnaires for completeness and coded them following standard coding procedures.[35] Questionnaires were then sent to Harvard T. H. Chan School of Public Health (Boston, MA) for nutrient analysis.

2.8 Statistical Analysis

Of the 344 girls recruited for the study, 18 girls had missing data for either DXA soft tissue measures (n=2), pQCT bone measures (n=5) or ethnic status (n=11), and they were excluded from the analyses. A total of 326 girls had complete data on all measures and were included in the statistical analysis described herein.

The mean and standard deviation for normally distributed variables and median and interquartile range for skewed variables, were used to characterize the sample; analyses were also stratified by ethnic group (Hispanic vs. non-Hispanic). Differences between Hispanic and non-Hispanic girls were tested using a two-sample t-test for continuous measures and a chi-squared test or Fisher’s exact test for categorical measures.

Multiple linear regression was used to assess the independent relationships of total body fat mass and lean mass with pQCT estimates of bone strength while adjusting for maturity offset, height, and ethnicity (Hispanic or non-Hispanic). Further regressions using pQCT measures of bone material and geometric properties as dependent variables were performed to further understand the independent effect of lean and fat mass on the components contributing to overall bone strength. It has been suggested that a “fat mass threshold” exists above which the relationship between fat and bone changes.[36] To test for the presence of a curvilinear relationship between fat and/or lean mass with bone measures, inclusion of fat mass and lean mass squared terms were added to each regression model. Both fat mass and lean mass were positively skewed and consequently log transformed for all regression analyses, with the exception of analyses testing for curvilinearity between fat mass and lean mass with the bone measures.

Potential interactions between ethnicity and total body lean mass and fat mass were tested by inclusion of interaction terms in each regression model one at a time. There was a significant interaction between ethnicity and lean mass for total bone area of the femur and ethnicity and fat mass for total bone area of the femur and tibia (66% site) and femur strength. For all other bone outcomes there were no significant ethnicity-by-soft tissue-composition interactions, thus Hispanic and non-Hispanic girls were included together in all regression models.

All models were checked for linearity, normality, and homoscedasticity and some pQCT bone dependent variables were logarithmically transformed to meet linear regression assumptions. Since fat mass and lean mass are known to be significantly correlated, the variance inflation factor was calculated for all models to ensure collinearity was not a problem. Further adjustment of models for physical activity, energy intake, and essential bone micronutrients including calcium and Vitamin D, did not alter the relationship of total body lean and fat mass with the bone outcome measures; these covariates contributed <0.03 to the overall variance in bone strength explained by the models (data not shown). Thus, to maximize sample size, physical activity and diet were not included in the final models as covariates. Likewise, substitution of maturity offset with Tanner breast and pubic hair stage did not alter results, which is consistent with previous publications.[17] Thus we only report analyses that included maturity offset because its relation to bone parameters was consistently stronger.

A p-value of <0.05 was considered statistically significant. All analyses were performed using STATA (StataCorp LLC, College Station, TX, USA) version 13.1.

3. Results

The sample characteristics are shown in Table 1. Three percent of girls were underweight, 58% healthy weight, 15% overweight, and 24% were obese, based on age and gender-specific established cut-points for percentiles of body mass index (BMI, kg/m²).[37] The majority of girls were early pubertal (Tanner breast stage 2–3 and Tanner pubic hair stage 1–2) and on average had just reached their estimated peak height velocity (maturity offset=0.3 year). Self-reported calcium and vitamin D intakes were below the recommended dietary allowances for 9–12 year olds[38] and physical activity was less than the 60 minutes of moderate-to-vigorous physical activity per day recommended by the U.S. Department of Health and Human Services 2008 Physical Activity Guidelines for children and adolescents.[39] The majority of the sample (74%) identified as Hispanic. Compared to non-Hispanic girls, Hispanic girls were significantly younger and had greater intakes of dietary calcium as well as total calories (p<0.05). Hispanic girls had a significantly greater percent of their body weight as fat compared to the non-Hispanic girls (p<0.05).

Table 1.

Participant characteristics

Characteristic	Full sample (n=326)	Hispanic (n=241)	Non-Hispanic (n=85)	p-value
	Mean (SD)
Age (years)	10.8 ±1.1	10.7 ± 1.1	11.0 ± 1.1	0.024
Ethnicity [n(%)]
Hispanic	241 (73.9%)
Non-Hispanic	85 (26.1%)
Maturity Offset (years)	0.3 ± 1.2	0.2 ± 1.2	0.5 ± 1.3	0.10
Tanner breast stage [n(%)]				0.49
1	57 (17.5%)	39 (16.2%)	18 (21.2%)
2	117 (35.9%)	88 (36.5%)	29 (34.1%)
3	107 (32.8%)	80 (33.2%)	27 (31.8%)
4	36 (11.0%)	29 (12.0%)	7 (8.2%)
5	9 (2.8%)	5 (2.1%)	4 (4.7%)
Tanner pubic hair stage [n(%)]				0.19
1	182 (55.8%)	140 (58.1%)	42 (49.4%)
2	104 (31.9%	76 (31.5%)	28 (33.0%)
3	25 (7.7%)	14 (5.8%)	11 (12.9%)
4	11 (3.4%)	7 (2.9%)	4 (4.7%)
5	4 (1.2%)	4 (1.7%)	0 (0%)
Menarche [n(%)]	63 (19.3%)	47 (19.5%)	16 (18.8%)	0.89
Weight (kg)	41.2 (20.1)^	41.8 (20.1)^	40.7 (16.2)^	0.45
Height (cm)	145.8 ± 9.7	145.2 ± 9.4	147.4 ± 10.1	0.07
BMI (kg/m2)	19.2 (6.8)^	19.5 (6.8)^	18.3 (5.7)^	0.05
BMI Percentile Status [n(%)]1				0.06
Underweight (<5th)	10 (3.1%)	5 (2.1%)	5 (5.9%)
Healthy weight (≥5th and <85th)	190 (58.3%)	135 (56.0%)	55 (64.7%)
Overweight (≥85th<95th)	49 (15.0%)	42 (17.4%)	7 (8.2%)
Obese (≥95th)	77 (23.6%)	59 (24.5%)	18 (21.2%)
MVPA (min/day)2	20.4 (18.4)^	19.5 (16.0)^	23.0 (25.5)^	0.20
Total energy (kcals/day)3	2015.3 (1157.4)^	2130.3 (1218.9)^	1828.4 (995.8)^	0.0044
Calcium (mg/day)3	1093.3 (777.5)^	1132.9 (827.2)^	1042.7 (600.7)^	0.034
Vitamin D (IU/day)3	259.6 (264.9)^	258.6 (282.5)^	261.2 (220.4)^	0.92
Total fat mass (kg)	13.0 (13.2)^	13.8 (13.1)^	10.6 (8.8)^	0.06
Total body fat (%)	32.4 ± 9.8	33.4 ± 9.7	29.6 ± 9.8	0.0024
Total lean mass (kg)	26.2 (8.1)^	26.0 (7.4)^	26.6 (8.6)^	0.19
Total body lean (%)	64.0 ± 9.5	63.1 ± 9.3	66.8 ± 9.4	0.0024

MVPA, moderate to vigorous physical activity

^{^}median (interquartile range)

¹percentiles specific for age and gender[24]

²n=303 for full sample; n=222 for Hispanic; n=81 for non-Hispanic

³n=309 for full sample; n=228 for Hispanic; n=81 for non-Hispanic

⁴Significantly different from non-Hispanic using two-sample t-test (p<0.05)

There was no significant difference in any of the pQCT bone measures between Hispanic and non-Hispanic girls (Table 2).

Table 2.

Summary of bone measures by pQCT in Hispanic and non-Hispanic girls

	Full sample (n=326)	Hispanic (n=241)	Non-Hispanic (n=85)	p-value1
	Mean (SD)
Radius (66%)
Cort vBMD (mg/cm3)	1053.5 (58.2)^	1052.7 (59.6)^	1053.9 (56.6)^	0.77
Cort BMC (mg/mm)	55.4 ± 14.1	55.2 ± 13.5	55.8 ± 15.7	0.77
Cort A (mm2)	52.5 ± 11.9	52.4 ± 11.4	52.7 ± 13.3	0.83
Tot A (mm2)	171.9 (42.9)^	170.4 (42.2)^	175.8 (42.9)^	0.41
PC (mm)	33.0 ± 3.5	33.0 ± 3.3	32.9 ±3.9	0.69
EC (mm)	20.7 ± 3.6	20.8 ± 3.5	20.4 ± 3.8	0.44
Crt Thk (mm)	2.0 ± 0.4	2.0 ± 0.4	2.0 ± 0.4	0.56
SSI (mm3)	164.0 (68.9)^	160.0 (68.3)^	175.6 (66.0)^	0.56
Femur (20%)
Cort vBMD (mg/cm3)	1058.3 ± 27.7	1059.2 ± 27.0	1055.6 ± 29.6	0.30
Cort BMC (mg/mm)	192.0 (54.2)^	193.7 (51.7)^	189.2 (55.0)^	0.48
Cort A (mm2)	186.1 ± 34.7	186.9 ± 34.0	183.9 ± 36.7	0.51
Tot A (mm2)	413.9 (139.2)^	412.2 (137.6)^	415.7 (141.4)^	0.51
PC (mm)	79.8 ± 8.9	79.7 ± 8.9	79.9 ± 8.8	0.90
EC (mm)	63.5 ± 8.9	63.3 ± 9.0	63.8 ± 8.8	0.67
Crt Thk (mm)	2.6 ± 0.4	2.6 ± 0.4	2.6 ± 0.4	0.26
SSI (mm3)	1543.3 (646.9)^	1550.9 (651.5)^	1540.4 (602.3)^	0.87
Tibia (66%)
Cort vBMD (mg/cm3)	1034.2 ± 36.1	1036.1 ± 36.1	1028.8 ± 35.8	0.11
Cort BMC (mg/mm)	206.8 ± 43.6	207.9 ± 41.4	203.6 ± 49.4	0.44
Cort A (mm2)	199.5 ± 39.2	200.3 ± 37.5	197.1 ± 43.7	0.52
Tot A (mm2)	395.0 (120.8)^	396.8 (118.7)^	391.8 (120.5)^	0.87
PC (mm)	74.6 ± 7.2	74.6 ± 7.3	74.7 ± 7.2	0.93
EC (mm)	55.3 ± 7.2	55.1 ± 7.3	55.6 ± 6.9	0.60
Crt Thk (mm)	3.1 ± 0.5	3.1 ± 0.5	3.0 ± 0.5	0.31
SSI (mm3)	1413.0 (575.5)^	1413.0 (582.4)^	1413.0 (483.7)^	0.78
Tibia (4%)
Trab vBMD (mg/cm3)	226.7 ± 28.0	226.9 ± 27.8	226.2 ± 28.9	0.85
Trab A (mm2)	484.6 ± 102.5	481.0 ± 97.9	494.8 ± 114.6	0.32
Tot vBMD (mg/cm3)	302.6 ± 37.8	303.8 ± 36.2	299.2 ± 42.1	0.33
Tot A (mm2)	604.8 (151.5)^	605.2 (159.1)^	602.9 (134.1)^	0.36
BSI (mg/cm4)	54.1 (23.3)^	54.6 (22.1)^	53.1 (23.3)^	0.68

Cort vBMD, cortical volumetric bone mineral density; Cort BMC, cortical bone mineral content; Cort A, cortical area; Tot A, total bone cross-sectional area; PC, periosteal circumference; EC, endosteal circumference; Crt Thk, cortical thickness; SSI, strength strain index; Trab vBMD, trabecular volumetric bone mineral density; Trab A, trabecular area; Tot vBMD, total volumetric bone mineral density; BSI, bone strength index

^{^}median (interquartile range)

¹Test of significance between groups based on two-sample t-test

The independent contributions of lean vs. fat mass to bone vBMD, structure, and strength (SSI, BSI) after adjusting for maturity offset, height, and ethnicity, are presented in Table 3.

Table 3.

Multiple linear regression of total body lean and fat mass as independent predictors of bone density, geometry, and strength in weight-bearing and non-weight-bearing skeletal sites

	Log lean mass (β)	p-value	Log fat mass (β)	p-value	Adjusted R2
Non-weight bearing diaphyseal
Radius (66%)
Log Cort vBMD	−0.12	0.33	−0.10	0.16	0.10
Cort BMC	0.52	0.0001	−0.11	0.04	0.52
Log Cort A	0.55	0.0001	−0.10	0.07	0.53
Log Tot A	0.83	0.0001	0.03	0.50	0.61
PC	0.57	0.0001	0.09	0.14	0.44
EC	0.29	0.02	0.24	0.001	0.11
Crt Thk	0.39	0.0001	−0.23	0.0001	0.35
Log SSI	0.67	0.0001	0.002	0.96	0.58
Weight bearing, diaphyseal
Femur (20%)
Cort vBMD	−0.35	0.01	−0.04	0.56	0.13
Log Cort BMC	0.51	0.0001	0.04	0.35	0.73
Log Cort A	0.58	0.0001	0.05	0.25	0.74
Log Tot A	0.74	0.0001	0.16	0.0001	0.73
Log PC	0.46	0.0001	0.15	0.001	0.70
Log EC	0.36	0.0001	0.16	0.003	0.51
Log Crt Thk	0.40	0.001	−0.07	0.30	0.21
Log SSI	0.53	0.0001	0.07	0.05	0.82
Tibia (66%)
Cort vBMD	−0.36	0.003	−0.03	0.63	0.22
Log Cort BMC	0.60	0.0001	0.08	0.06	0.71
Log Cort A	0.71	0.0001	0.09	0.03	0.72
Log Tot A	0.84	0.0001	0.17	0.0001	0.74
Log PC	0.59	0.0001	0.15	0.002	0.63
Log EC	0.36	0.001	0.14	0.03	0.30
Crt Thk	0.54	0.0001	0.02	0.71	0.33
Log SSI	0.68	0.0001	0.13	0.0001	0.78
Weight bearing, metaphyseal
Tibia (4%)
Trab vBMD	0.73	0.0001	0.12	0.09	0.18
Trab A	0.36	0.001	0.11	0.07	0.38
Tot vBMD	0.63	0.0001	0.06	0.38	0.16
Log Tot A	0.47	0.0001	0.13	0.02	0.51
Log BSI	0.86	0.0001	0.14	0.007	0.58

n= 326

β=standardized beta-coefficient adjusted for maturity offset, height, ethnicity and fat mass or lean mass

Cort vBMD, cortical volumetric bone mineral density (mg/cm³); Cort BMC, cortical bone mineral content (mg/mm); Cort A, cortical area (mm²); Tot A, total bone area (mm²); PC, periosteal circumference (mm); EC, endosteal circumference (mm); Crt Thk, cortical thickness (mm); SSI, strength-strain index (mm³); BSI, bone-strength index (mg/cm⁴)

In diaphyseal bone (66% radius and tibia, 20% femur), with the exception of cortical bone density (Cort vBMD), lean mass was a significant positive contributor to all bone outcomes, including estimated bone torsional and bending strength (SSI), in both the weight bearing and non-weight bearing bones. When substituting muscle cross sectional area, a surrogate for the local dynamic loads on bone, for lean mass in our regression models, the relationships of muscle cross-sectional area with bone strength at the radius, tibia, and femur were similar to those of total body lean mass (data not shown). Similar to lean mass, fat mass had a positive relationship with most bone outcomes including bone strength (SSI). However, in contrast to lean mass, these relationships were much weaker, with fat mass significantly contributing only to bone strength (SSI) at the weight bearing bones. Moreover, in the non-weight bearing diaphyseal radius, fat mass had a significant negative association with cortical bone content (Cort BMC) and thickness (Crt Thk).

In the weight bearing metaphyseal bone (4% tibia), despite the lack of association of fat mass with total vBMD, which is an important element in estimating resistance to compression (BSI=Tot vBMD² × Tot A), both lean and fat mass were predictive of BSI as well as all other bone outcomes, albeit with weaker associations for fat (vs. lean) mass.

There was a significant curvilinear relationship of lean mass with bone strength at both the weight and non-weight bearing skeletal sites (p<0.05), indicating that the relationship of lean mass with bone strength was not constant across different levels of lean mass. No significant curvilinear relationship between fat mass and bone strength was observed (p<0.05).

There was a significant interaction between ethnicity and log fat mass with total bone area (Tot A) at the 20% femur (p=0.02) and 66% tibia (p=0.005) as well as bone strength (SSI) at the 20% femur (p=0.03), such that, in Hispanic girls, fat mass was a significant predictor of total bone area (Tot A) and strength of the femur (SSI) and total bone area of the tibia (Tot A). In non-Hispanic girls fat mass was not a significant contributor of bone outcomes (data not shown). Likewise, there was a significant interaction between ethnicity and log lean mass with total bone area (Tot A) at the 20% femur (p=0.04). Lean mass in Hispanic girls was not as strong of a predictor of total bone area (Tot A) and strength (SSI) at the 20% femur and total bone area (Tot A) at the 66% tibia compared to non-Hispanic girls (standardized β for log lean mass in Hispanic girls= 0.67, 0.45, and 0.75 vs. non-Hispanic girls= 0.95, 0.76, 1.06 for femur Tot A, SSI, and tibia Tot A respectively).

4. Discussion

The aim of this study was to assess the independent contributions of total body fat and lean mass on bone density, geometry, and strength at weight-bearing and non-weight-bearing skeletal sites in young Hispanic and non-Hispanic girls. Our results showed that lean mass significantly contributed to bone structure and overall bone strength at both weight-bearing and non-weight-bearing skeletal sites. This is consistent with Frost’s mechanostat theory that bone adapts primarily to the dynamic loads imposed by muscle contractions.[40, 41] Thus, since overweight and obese children have higher amounts of lean mass in addition to higher levels of fat mass, it has been postulated that the positive associations seen between higher body weight and bone outcomes is explained by their higher lean mass and not the extra weight from fat mass.[36] However, we found that fat mass was a significant contributor to bone strength at the femur and tibia, independent of lean mass, through its positive effects on bone area and periosteal expansion. Thus, despite lean mass playing the major role in bone acquisition during childhood, this finding indicates that fat mass has an additional mechanical loading effect on bone strength at weight bearing skeletal sites beyond that of the dynamic loading imposed by muscle forces. Contrary to our findings, Cole et al., found that total body fat mass, although positively related, was not a significant predictor of pQCT estimated bone strength at the distal tibia in 6-year-old children after adjustment for total body lean mass.[16] However, Farr et al., using high-resolution pQCT, found that total body fat mass had a significant positive relationship with bone strength, independent of appendicular lean mass at the distal tibia in 8–15 year old girls.[42] Unlike for the tibia and femur, total body fat mass had no effect on bone strength at the radius in our sample of girls. Likewise, Farr et al., found no relationship between fat mass and bone strength at the radius after controlling for appendicular lean mass.[42] In contrast, Wey et al., found total body fat mass had a significant negative association with bone strength at the radius measured by pQCT in girls aged 13 to 17.[20] Our findings together with those of Cole, Farr and Wey suggest that fat mass may have differential relations to bone strength at weight-bearing and non-weight-bearing skeletal sites, such that for the weight-bearing skeleton of the lower limbs, fat mass may positively add to bone strength whereas fat mass provides little or no benefit to bone strength in the upper limbs.

Although we found that total body fat mass had a positive relationship with bone strength of the weight-bearing skeleton, there have been reports of fat mass having a negative effect on bone outcomes. For example, in a study of older adolescent females, Pollock et al., found that females with ≥32% total body fat had lower bone strength at the tibia and radius compared to females with <32% total body fat.[13] It is possible that a threshold exists at which an increase in fat mass no longer has a beneficial effect on the growing skeleton and may actually be detrimental to bone development in children.[36] We did not observe a significant curvilinear relationship between fat mass and bone strength in our sample. However, we did find a significant curvilinear relationship between lean mass and bone strength at both the weight-bearing and non-weight-bearing skeletal sites. This suggests that at higher levels of lean mass, there was less of an increase in bone strength compared to that at lower levels of lean mass and it could be that higher levels of fatness attenuate the gains in bone strength with increasing lean mass, given the unfavorable metabolic changes that accompany excess fat accumulation, especially in ectopic depots.

A strength of our study was a large sample, including a sizable sample of Hispanic girls, which is an underserved population at risk for obesity and adverse cardiometabolic outcomes.[43] Previous studies have reported the presence of racial and ethnic differences in DXA measured bone mass of children.[44–47] However, few studies have examined the effect of race and ethnicity on measures of bone geometry and strength by pQCT. In our study, ethnicity accounted for very little variance in the bone parameters we measured, and after accounting for maturation, height, and lean mass, there were no significant differences in bone strength between Hispanic and non-Hispanic girls. This is contrary to the findings of Weztsteon et al., who reported both black and Hispanic 9-to-12-year-old children have greater bone strength than white children after adjusting for age, sex, limb length, and muscle cross-sectional area.[48] Although the Hispanic girls in our sample did not differ in bone outcomes compared to non-Hispanics, there was a significant effect of ethnicity on the relationship of fat mass with total bone area at the femur and diaphyseal tibia, and with bone strength at the femur, such that for a given amount of fat mass, bone area and strength were greater in Hispanic compared to non-Hispanic girls. Even though Hispanic females are more likely to have a higher percentage of body fat compared to non-Hispanic whites [49], it is possible that higher body fat up to a certain level, is not deleterious in Hispanics and this fat mass may be advantageous to bone. There was also a significant effect of ethnicity on the relationship of lean mass with total bone area at the femur such that compared to non-Hispanic girls, Hispanic girls had a smaller bone area for a given amount of lean mass. No study to our knowledge has assessed differences in the relationship of fat and lean mass with bone measures of geometry and strength in Hispanic and non-Hispanic children. Further studies that include a larger sample of non-Hispanic children are needed to validate our findings of an ethnicity interaction with fat and lean mass on bone area and strength.

A major strength of this study was the use of pQCT to measure bone outcomes at both the weight-bearing and non-weight-bearing skeletal sites. Unlike DXA, pQCT is able to measure vBMD and bone structural properties, as well as provide estimates of bone strength. Many previous studies assessing the effect of soft tissue composition on the developing bones of children have relied on DXA, which has inherent limitations when it comes to assessing bone density in the growing skeleton and lacks an analysis of bone geometry and therefore estimates of bone resistance to fracture in children.[12]

This study was not without limitations. Due to the studies cross-sectional design, we cannot determine whether the relationship of fat and lean mass with bone strength persists as girls progress through puberty and into adulthood. A second limitation was our use of lean mass as a surrogate measure of muscle strength although lean mass is highly correlated with muscle size and muscle size is highly related to muscle strength.[13] In addition, localized fat within muscle and around bone may have different associations with bone geometry and strength than that of total body fat mass. Further studies are needed to assess whether the relationships of site-specific measures of soft tissue (i.e. muscle density, muscle cross-sectional area, subcutaneous fat, skeletal muscle fat, etc.) with bone strength are similar to those of whole body lean and fat mass with bone strength. Lastly, we did not include measures indicative of metabolic dysfunction, which may modify the relationships we found between fat and lean masses and bone outcomes. Metabolic dysfunction, such as insulin resistance, chronic inflammation, and dyslipidemia, often present with excessive body fat levels, has been shown to impair bone.[50, 51]

5. Conclusion

Lean mass was consistently a stronger determinant of measures of bone structure and strength in the upper and lower limbs of both Hispanic and non-Hispanic girls compared to total body fat mass. These findings emphasize the importance of muscle mass for optimizing bone geometry and strength in children. While lean mass, a surrogate for the local dynamic loads on bone, is the main driver of bone strength for all skeletal sites, in the weight-bearing skeleton, there appears to be an independent effect of the gravitational loads imposed by fat mass. Although fat mass was positively related to bone strength in the weight-bearing skeleton, it is possible that accumulation of fat within the skeletal muscle and metabolic dysfunction, commonly occurring with excessive levels of body fat, may attenuate these positive effects. Further studies in children are needed to assess whether the relationships of total body fat and lean mass with bone strength are altered by the presence of pathogenic fat depots and metabolic risk factors.

Highlights.

• Lean mass is the main determinant of bone strength of appendicular bones of girls

• Fat mass contributes to bone strength of weight bearing bones

• Fat mass provides no benefit to bone strength of the non-weight bearing upper limb

• Relationships between soft tissue and bone similar for Hispanic and non-Hispanics

Abbreviations

BMC

bone mineral content

DXA

dual-energy x-ray absorptiometry

aBMD

areal bone mineral density

pQCT

peripheral quantitative computed tomography

vBMD

volumetric bone mineral density

Tot A

total bone area

Cort vBMD

cortical vBMD

Cort BMC

coritical BMC

Cort area

cortical area

EC

endosteal circumference

PC

periosteal circumference

Crt Thk

cortical thickness

Trab vBMD

trabecular vBMD

Trab area

trabecular area

SSI

strength-strain index

BSI

bone strength index