Relationships among biological sex, body composition, and bone mineral density in young persons with and without diabetes

Abstract

Objectives

Bone mineral density (BMD) is influenced by factors including age, sex, body composition, and diabetes. However, data regarding these relationships in young individuals is limited. The objective of this study was to address this gap in the literature. 

Methods

We conducted a post-hoc analysis of participants from six cross-sectional cohort studies, encompassing individuals with type 1 diabetes (T1D) and type 2 diabetes (T2D), as well as controls of healthy weight (HWC) and with obesity (OC). Whole-body dual-energy X-ray absorptiometry (DXA) was employed to measure BMD and body composition. Multiple linear regression models assessed sexual dimorphism in BMD, adjusting for age and exploring effect modification by group and sex.

Results

A total of 325 participants were included (T1D [n=123, mean age 22.4 years, 50 % male], T2D [n=72, mean age 16.2 years, 33 % male], HWC [n=79, mean age 16.6 years, 41 % male], and OC [n=51, mean age 13.8 years, 53 % male]). Sexual dimorphism in BMD was evident only in T1D and HWC, with males having higher BMD than females (p=0.021; p<0.001, respectively). BMI was positively correlated with BMD in all groups (p<0.001 for HWC; p=0.001 for OC; p<0.001 for T1D; p=0.008 for T2D). Body fat percentage was inversely correlated with BMD in HWC and T1D (p<0.001; p=0.011, respectively), but not in OC or T2D. Additionally, lean mass percentage was significantly associated with higher BMD in HWC and OC (p<0.001; p=0.023, respectively), but not in groups with diabetes.

Conclusions

Our study documents sexual dimorphism in BMD in youth, with varied associations between body composition metrics and BMD across groups with diabetes and in controls without diabetes, underscoring the importance of understanding these relationships for optimizing bone health during adolescence.

Introduction

Bone mineral density (BMD) is a key indicator of bone health and predictor of osteoporosis. Numerous factors, including age, pubertal stage, sex, race/ethnicity, physical activity, dietary composition, BMI, and body composition, influence BMD.

Type 1 diabetes (T1D) is an autoimmune disease characterized by insulin deficiency, is linked to reduced BMD and elevated fracture risk [1], observed in both sexes and accentuated by aging [2]. This is potentially due to the absence of insulin’s anabolic effects during critical periods of bone accrual [2]. Dysregulation of the insulin-like growth factor (IGF-1) axis in T1D may further contribute to impaired bone health [3], [4], [5]. Studies investigating the association between glycemic control and BMD have yielded inconsistent findings [2], [6], [7], [8].

In contrast type 2 diabetes (T2D), now with increasing global prevalence in youth, is associated with normal to elevated BMD due in part to anabolic effects of hyperinsulinemia [1], [9], [10], [11], [12], [13]. Hyperinsulinemia may also contribute to increased BMD by suppressing hepatic sex-hormone binding globulin release, increasing free serum estradiol and testosterone [14], [15], [16]. However, despite normal to increased BMD, T2D is paradoxically associated with increased fracture risk [10]. This is partially attributed to hyperglycemia-driven production of advanced glycation end products, oxidative stress, and impaired vitamin D metabolism, which lead to alterations in bone competence [17], [18], [19], [20], [21], [22].

Obesity, a common comorbidity of T2D, adds complexity to the relationship between BMD and diabetes, as higher body mass may enhance mechanical loading on bone but also promotes inflammation, oxidative stress, and bone resorption [23], [24], [25], [26], [27], [28], [29]. Body fat percentage may impact BMD differently by sex, having more deleterious effects in males [25], [30], possibly because estradiol, which regulates bone metabolism, is higher in women. However, adipose tissue also contributes to conversion of androgens such as testosterone to estrogen via aromatase, creating a complex interplay [31], [32], [33].

Peak bone mass, achieved by the end of puberty, is a critical determinant of long-term skeletal health [34]. Puberty thus is an important time for bone acquisition, and it is during this time that differences in BMD based on sex and anthropometric characteristics become apparent [35], [36]. While the effects of diabetes and obesity on BMD in adults are well studied, significant gaps remain in understanding how these conditions impact bone health during youth and adolescence, particularly in regard to sex differences. This study aims to address these gaps by exploring the relationships among diabetes types, obesity, body composition, and BMD in youth and young adults.

Subjects and methods

Study design and participants

This study was a post-hoc analysis of adolescent and young adult participants ages 8.2–37.3 from six different cross-sectional cohort studies. These cohorts consisted of 123 participants with T1D (ages 12.0–37.3 years) from the Copeptin in adolescent participants with type 1 diabetes and early renal hemodynamic function (CASPER, n=49), control of renal oxygenation, consumption, mitochondrial dysfunction, and insulin resistance (CROCODILE, n=28), and pathogenesis of kidney disease in type 1 diabetes (PANDA, n=46) studies. Additionally, 72 participants with T2D (ages 9.7–20.2 years) were included from the renal hemodynamics, energetics, and insulin resistance (RENAL-HEIR, n=46), impact of metabolic surgery on pancreatic, renal and cardiovascular health in youth with type 2 diabetes (IMPROVE-T2D, n=17), and puberty, diabetes, and the kidneys, when eustress becomes distress (PANTHER, n=9) studies. The healthy weight control (HWC) group (n=79) was drawn from participants (ages 8.2–30.0 years) from the CROCODILE (n=19), RENAL-HEIR (n=20), and PANTHER (n=40) studies. Furthermore, a control cohort of participants with obesity but not diabetes (OC) (n=51) was derived from participants (ages 8.8–20.7 years) from the RENAL-HEIR (n=21) and PANTHER (n=30) studies. Detailed inclusion and exclusion criteria for each study can be found in Supplemental Table 1. Analysis of missing data was performed, and data missing was minimal, thus a complete case analysis was performed by excluding those participants with any missing data.

Consent and IRB approval

All participants underwent assent plus written parental consent if age <18 years, or written participant consent if age ≥18 years. Study protocols were reviewed and approved by the Colorado Multiple Institutional Review Board (COMIRB), approval numbers 17–0820 (CASPER), 19–1,282 (CROCODILE), 16–1752 (RENAL-HEIR), 18–0704 (IMPROVE-T2D), 22–0250, and 21–3,019 (PANTHER).

Measurement of bone mineral density and body composition

Participants underwent whole-body dual-energy X-ray absorptiometry (DXA) on a Hologic device (Waltham, MA) to measure whole-body BMD, lean mass, and fat mass, and to calculate %body fat and %lean mass. BMI standard deviation score (SDS) was calculated for participants between ages 2–20 using the Centers for Disease Control and Prevention (CDC) growth charts as reference [37].

Statistical analysis

Participant characteristics were summarized as count and percentages, mean and standard deviation (SD), or median and interquartile range (IQR) based on visual inspection of histograms for distribution.

We used multiple linear regression models to investigate the association between BMD and sex. We defined four groups: T1D, T2D, HWC, and OC. Each of these groups was comprised of participants from one or more of the six cohort studies as outlined above in the “study design and participants” section. The main model included group, sex, and a group*sex interaction term to assess potential sex differences in BMD between groups. Additional models were run to examine whether the relationships between BMD and each body composition measures (BMI, body fat percentage, lean mass percentage) were modified by group or sex. These included models with group*body composition measure interaction terms, as well as models stratified by group with sex*body composition measure interaction terms to assess whether the sex differences in the association between BMD and body composition measures varied across groups. All models included age as a covariate to adjust for its potential confounding effect. Age was treated as a continuous variable in the model, allowing for a linear adjustment across the full age range in the study population.

A p-value of <0.05 was considered statistically significant. Analyses were performed using Python v.3.9.6 and R v.4.2.2.

Results

Baseline characteristics

A total of 325 participants were included (T1D [n=123], T2D [n=72], HWC [n=79], and OC [n=51]). Participant characteristics were assessed across different groups and are presented in Table 1. The groups varied in mean age, with the OC group having the youngest mean age at 13 ± 2.7 years, and the T1D group being the oldest with a mean age of 22.4 ± 6.3 years. Additionally, differences in race and ethnicity distribution were observed, with the T1D group having the highest representation of non-Hispanic or Latino White participants at 84 %, while the T2D and OC groups had the highest proportions of Hispanic or Latino participants at 75 and 69 %, respectively. Expected differences were identified in several key parameters, including HbA_1c, BMI, %body fat, and %lean mass across the groups. The OC and T2D groups exhibited higher BMI (median [Q1, Q3] of 33.4 [29.5, 36.8] kg/m² and 36.5 [31.2, 42.1] kg/m², respectively), higher %body fat (means of 45.0 ± 5.9 % and 44.1 ± 6.8 %, respectively), and lower %lean mass (means of 53.9 ± 5.7 % and 55.1 ± 6.5 %, respectively) compared to the HWC and T1D. As anticipated, HbA_1c was higher in both the T1D (median of 7.7 [6.8, 8.9]%) and T2D (median of 6.3 [5.9, 7.3]%) groups in contrast to the OC and HWC groups (medians of 5.4 [5.3, 5.6]% and 5.2 [5.1, 5.4]%, respectively; p<0.001).

Table 1:

Participant characteristics for HWC, OC, T1D, and T2D groups.

	HWC (n=79)	OC (n=51)	T1D (n=123)	T2D (n=72)
Male, %	41 %	53 %	50 %	33 %
Age, years	16.6 ± 5.7, range 8.2–30.0	13.8 ± 2.7, range 8.8–20.7	22.4 ± 6.3, range 12.0–37.3	16.2 ± 2.4, range 9.7–20.2
Race and ethnicity, n (%)
Hispanic or latino	12 (15 %)	35 (69 %)	6 (5 %)	54 (75 %)
Non-hispanic or latino black	6 (8 %)	3 (6 %)	2 (2 %)	11 (15 %)
Non-hispanic or latino white	51 (65 %)	12 (24 %)	103 (84 %)	6 (8 %)
Non-hispanic or latino other	10 (13 %)	1 (2 %)	12 (10 %)	1 (1 %)
Diabetes duration, years	NA	NA	11.8 ± 7.4	2.2 ± 1.6
HbA1c, %	5.2 (5.1, 5.4)	5.4 (5.3, 5.6)	7.7 (6.8, 8.9)	6.3 (5.9, 7.3)
BMD, g/cm2	1.04 ± 0.18	1.03 ± 0.15	1.13 ± 0.15	1.09 ± 0.11
BMI, kg/m2	21.2 (17.5, 23.4)	33.4 (29.5, 36.8)	25.2 (21.5, 28.8)	36.5 (31.2, 42.1)
BMI SDSa	0.130 (0.924)	2.322 (0.402)	0.556 (1.025)	2.266 (0.496)
Height, cm	160.7 ± 13.3	162.8 ± 13.0	171.6 ± 9.6	167.1 ± 8.3
Weight, kg	55.0 ± 15.5	92.4 ± 26.5	76.1 ± 19.4	104.8 ± 26.5
Body fat mass, kg	15.5 ± 5.8	42.0 ± 15.4	23.6 ± 9.6	46.8 ± 17.1
Body fat mass, %	28.5 ± 6.8	45.0 ± 5.9	31.2 ± 8.4	44.1 ± 6.8
Body lean mass, kg	38.1 ± 11.7	50.4 ± 14.6	49.4 ± 12.0	56.8 ± 12.1
Body lean mass, %	69.0 ± 6.5	53.9 ± 5.7	66.5 ± 9.1	55.1 ± 6.5

Values reported as mean ± standard deviation, median (1st quartile, 3rd quartile), or range. BMD, bone mineral density; BMI, body mass index; HbA_1c, hemoglobin A_1c; HWC, healthy weight controls; MAP, mean arterial pressure; OC, obese controls, T1D, type 1 diabetes; T2D, type 2 diabetes. ^aBMI SDS, calculated based on CDC, reference for ages 2–20. Twenty-one from HWC, 1 from OC, 79 from T1D, and 2 from T2D missing due to age>20.

Sexual dimorphism

After adjusting for age, significant sexual dimorphism in BMD was only observed in the T1D (male: 1.09, 95 % CI: 1.06, 1.12; female: 1.04, 95 % CI: 1.01, 1.06 g/cm²; p<0.001) and HWC groups (male: 1.08, 95 % CI: 1.05, 1.11; female: 1.03, 95 % CI: 1.00, 1.06 g/cm²; p=0.021), with males having higher mean BMD than females in both groups (Figure 1).

Figure 1:

Sexual dimorphism in bone mineral density. The LSMeans and 95 % confidence intervals are shown stratified by group and sex which were generated from a multiple linear regression model, adjusting for the effects of age. The p-values displayed above each grouped pair of lines indicate the significance of the LSMean differences between sexes within each group.

Relationships between BMD and BMI

We found that the effect of BMI on BMD differed by group, where the BMI association with BMD was the strongest in HWC, followed by T1D, OC, and T2D. For every 1 kg/m² increase in BMI, BMD increased by 0.034 g/cm² in HWC (standard error; SE: 0.004, p<0.001), 0.015 g/cm² in T1D (SE: 0.002, p<0.001), 0.009 g/cm² in OC (SE: 0.003, p=0.001), and 0.005 g/cm² in T2D (SE: 0.002, p=0.008; Figure 2A).

Figure 2:

Associations between body mass index and bone mineral density by group and sex. (A) The relationship between body mass index (BMI) and bone mineral density (BMD) is depicted for each group, revealing a positive correlation across all groups. The HWC group exhibited the strongest association between BMI and BMD, while the T2D group demonstrated the weakest association. Effect modification by group*BMI was evaluated. Age is indicated by dot size, with smaller dots representing younger participants. While the legend uses age values of 10, 20, and 30 to demonstrate changes in dot size, the dots reflect a continuous range of participant ages across the cohort. (B) The relationship between BMI and BMD is depicted for each group and sex, revealing a positive correlation across all groups. Overall, males exhibited similar or stronger associations between BMI and BMD compared to females across all groups. Effect modification by sex*BMI was evaluated for each group. Age is indicated by dot size, with smaller dots representing younger participants. While the legend uses age values of 10, 20, and 30 to demonstrate changes in dot size, the dots reflect a continuous range of participant ages across the cohort.

Summary of associations between body mass index and bone mineral density by group and sex.

	HWC		OC		T1D		T2D
	β (95 % CI)	p-Value	β (95 % CI)	p-Value	β (95 % CI)	p-Value	β (95 % CI)	p-Value
Overall group	0.034 (0.026, 0.042)	<0.001	0.009 (0.004, 0.014)	0.001	0.015 (0.010, 0.019)	<0.001	0.005 (0.001, 0.008)	0.008
Female	0.029 (0.017, 0.041)	<0.001	0.009 (0.002, 0.016)	0.014	0.009 (0.002, 0.016)	0.008	0.003 (−0.000, 0.006)	0.065
Male	0.040 (0.271, 0.054)	<0.001	0.007 (−0.002, 0.017)	0.131	0.019 (0.126, 0.025)	<0.001	0.009 (0.005, 0.014)	0.002
Female vs. Male (Female as reference)	0.011 (−0.006, 0.029)	0.203	−0.001 (−0.013, 0.010)	0.817	0.010 (0.001, 0.019)	0.034	0.006 (0.001, 0.012)	0.031

The table summarizes beta coefficients with corresponding 95 % confidence intervals (CI) and p-values for the interaction effects of group*BMI and sex*BMI on BMD. The first row (“Overall group”) presents results from a model examining the interaction of group*BMI, while subsequent rows delineate results from a model with the interaction of sex*BMI stratified by group.

When stratified by group, we observed effect modification by sex and BMI on BMD in T1D (female as reference; 0.010, 95 % CI: 0.001, 0.019, p=0.034) and T2D (female as reference; 0.006, 95 % CI: 0.001, 0.012, p=0.031), but not in the HWC and OC (Figure 2B).

Relationships between BMD and %body fat

We found that the effect of %body fat on BMD differed by group, as the %body fat association with BMD was only significant in HWC and T1D. For every 1 % increase in body fat, BMD decreased by 0.009 g/cm² (SE: 0.002, p<0.001) and 0.004 g/cm² (SE: 0.002, p=0.01) in HWC and T1D, respectively (Figure 3A).

Figure 3:

Associations between body fat percentage and bone mineral density by group and sex. (A) The relationship between body fat percentage and bone mineral density (BMD) is depicted for each group, revealing a negative correlation across all groups. The HWC group exhibited the strongest association between body fat percentage and BMD, while the T2D group demonstrated the weakest association. Effect modification by group*body fat percentage was evaluated. Age is indicated by dot size, with smaller dots representing younger participants. While the legend uses age values of 10, 20, and 30 to demonstrate changes in dot size, the dots reflect a continuous range of participant ages across the cohort. (B) The relationship between body fat percentage and BMD is depicted for each group and sex, revealing mixed correlation directions across all groups. Overall, males exhibited inverse relationships between body fat percentage and BMD across all groups, although this was only statistically significant in HWC and OC groups, while females exhibited positive (but not statistically significant) relationships across all groups. Effect modification by sex*body fat percentage was evaluated for each group. Age is indicated by dot size, with smaller dots representing younger participants. While the legend uses age values of 10, 20, and 30 to demonstrate changes in dot size, the dots reflect a continuous range of participant ages across the cohort.

Summary of associations between body fat percentage and bone mineral density by group and sex.

	HWC		OC		T1D		T1D
	β (95 % CI)	p-Value	β (95 % CI)	p-Value	β (95 % CI)	p-Value	β (95 % CI)	p-Value
Overall group	−0.009 (−0.014, −0.004)	<0.001	−0.006 (−0.013, 0.001)	0.106	−0.004 (−0.007, −0.001)	0.011	−0.000 (−0.005, 0.005)	0.888
Female	0.000 (−0.009, 0.010)	0.948	0.009 (−0.003, 0.020)	0.141	0.000 (−0.005, 0.007)	0.776	0.003 (−0.002, 0.009)	0.230
Male	−0.019 (−0.030, −0.008)	<0.001	−0.012 (−0.021, −0.004)	0.005	−0.000 (−0.006, 0.005)	0.801	−0.001 (−0.001, 0.006)	0.845
Female vs. Male (Female as reference)	−0.019 (−0.033, −0.005)	0.010	−0.021 (−0.035, −0.007)	0.005	−0.0015 (−0.010, 0.007)	0.704	−0.004 (−0.012, 0.005)	0.360

The table summarizes beta coefficients with corresponding 95 % confidence intervals (CI) and p-values for the interaction effects of group*body fat percentage and sex* body fat percentage on BMD. The first row (“Overall group”) presents results from a model examining the interaction of group* body fat percentage, while subsequent rows delineate results from a model with the interaction of sex* body fat percentage stratified by group.

When stratified by group, we observed an effect modification by sex and %body fat on BMD in HWC (female as reference; −0.019, 95 % CI: −0.033, −0.005, p=0.010) and OC (female as reference; −0.021, 95 % CI: −0.035, −0.007, p=0.005), but not in T1D and T2D (Figure 3B).

Relationships between BMD and %lean mass

We found that the effect of %lean mass on BMD differed by group, where the %lean mass association with BMD was only significant in the HWC and OC groups. For every 1 % increase in lean mass, BMD increased by 0.010 g/cm² (SE: 0.003, p<0.001) and 0.008 g/cm² (SE: 0.004, p=0.02) in HWC and OC, respectively (Figure 4A).

Figure 4:

Associations between lean mass percentage and bone mineral density by group and sex. (A)The relationship between lean mass percentage and bone mineral density (BMD) is depicted for each group. There was a positive correlation between lean mass percentage and BMD in all groups, however this was only statistically significant in HWC and OC groups. The controls with no diabetes exhibited a stronger association between lean mass percentage and BMD compared to the groups with either T1D or T2D. Effect modification by group*lean mass percentage was evaluated. Age is indicated by dot size, with smaller dots representing younger participants. While the legend uses age values of 10, 20, and 30 to demonstrate changes in dot size, the dots reflect a continuous range of participant ages across the cohort. (B) The relationship between lean mass percentage and BMD is depicted for each group and sex, revealing mixed correlation directions across all groups. Overall, males exhibited positive relationships between lean mass percentage and BMD across all groups except in the T1D group, however this relationship was not significant for the male T2D group. Females exhibited inverse relationships across all groups except in the HWC, however none of these relationships were statistically significant. Effect modification by sex*lean mass percentage was evaluated for each group. Age is indicated by dot size, with smaller dots representing younger participants. While the legend uses age values of 10, 20, and 30 to demonstrate changes in dot size, the dots reflect a continuous range of participant ages across the cohort.

Summary of associations between lean mass percentage and bone mineral density by group and sex.

	HWC		OC		T1D		T2D
	β (95 % CI)	p-Value	β (95 % CI)	p-Value	β (95 % CI)	p-Value	β (95 % CI)	p-Value
Overall group	0.010 (0.005, 0.015)	<0.001	0.008 (0.001, 0.016)	0.023	0.001 (−0.002, 0.004)	0.517	0.001 (−0.004, 0.007)	0.623
Female	0.001 (−0.009, 0.011)	0.796	−0.009 (−0.022 0.004)	0.182	−0.002 (−0.008, 0.004)	0.428	−0.002 (−0.008, 0.004)	0.452
Male	0.020 (0.009, 0.030)	<0.001	0.014 (0.006, 0.023)	0.001	−0.003 (−0.007, 0.001)	0.086	0.001 (−0.006, 0.008)	0.750
Female vs. Male (Female as reference)	0.018 (0.004, 0.033)	0.015	0.023 (0.007, 0.039)	0.005	−0.001 (−0.008, 0.006)	0.750	0.003 (−0.006, 0.012)	0.462

The table summarizes beta coefficients with corresponding 95 % confidence intervals (CI) and p-values for the interaction effects of group* lean mass percentage and sex* lean mass percentage on BMD. The first row (“Overall group”) presents results from a model examining the interaction of group* lean mass percentage, while subsequent rows delineate results from a model with the interaction of sex* lean mass percentage stratified by group.

When stratified by group, we observed an effect modification by sex and %lean mass on BMD in HWC (female as reference; 0.018, 95 % CI: 0.004, 0.033, p=0.015) and OC (female as reference; 0.023, 95 % CI: 0.007, 0.039, p=0.005), but not in T1D and T2D (Figure 4B).

Discussion

This study investigated the relationship between BMD, body composition, and sex across young people with and without T1D and T2D. We found that males had higher BMD than females in the T1D and HWC groups, which was not seen in the T2D or OC groups. This aligns with previous research characterizing a relationship between sex hormones and bone health, where androgens and higher lean mass typically promote greater bone accrual in males [38], [39]. We hypothesize that sexual dimorphism in BMD may be lost in the T2D and OC groups due to higher androgen concentrations in females in these groups.

We found a significant positive association between BMI and BMD in all four groups, which was strongest in HWC, followed by T1D, OC, and T2D. Additionally, we observed an effect modification by sex on this association by sex in T1D and T2D groups, but not in OC or HWC groups. A positive association between BMI and BMD is well documented in the literature [10], [40]. However, it may be the case that this effect is lessened in the groups with obesity and/or diabetes due to the deleterious effects of impaired insulin and glucose metabolism and increased oxidative stress from adiposity on bone metabolism and bone strength. This may reflect a threshold effect, where additional increases in BMI do not further benefit BMD in OC and T2D groups, likely because these individuals already have higher BMI and BMD levels compared to HWC and T1D groups. This nuanced finding challenges the traditional perspective that higher BMI uniformly benefits bone health, showing instead that metabolic and hormonal disturbances in these populations alter the typical relationships observed in healthy youth. This underscores the importance of tailoring bone health interventions to specific subgroups, particularly during adolescence, a critical window for bone accrual.

The effect modification by sex in the association between BMI and BMD in T1D and T2D groups, but not in OC or HWC groups, may be attributed to the interplay between metabolic dysregulation inherent in diabetes and sex-specific hormonal effects on bone density. In diabetes, the negative impact of impaired glucose metabolism and increased adiposity might overshadow the protective effects of higher BMI on BMD [17], [18], [19], [20], [21], with these effects possibly being more pronounced in males due to the interaction between sex hormones and metabolic disturbances. This hypothesis aligns with the understanding that diabetes can alter bone metabolism through pathways that are sensitive to sex hormones [14], [15], [16].

In contrast to the relationship between BMI and BMD, our study found a significant negative relationship between body fat percentage and BMD in HWC and T1D groups, but no relationship between these variables in OC or T2D groups. There was an effect modification on this association by sex in both the HWC and OC groups, but not in the T1D or T2D groups. Previous studies have reported significant negative relationships between body fat percentage and BMD in children and adolescents without diabetes [41], and our results corroborate this finding. However, to our knowledge there is no current literature on the relationship between body fat percentage and BMD in young cohorts with diabetes. Our findings are also consistent with previous reports that there is a stronger negative association between %body fat and BMD in males compared to females [25]. This may be due to the fact that estradiol, which is a regulator of bone metabolism, is found in higher concentrations in women than in men. Additionally, adipose tissue contributes to androgen conversion to estrogen as it serves as a site for sex steroid transformation and is a source for the synthesis of aromatase [31], [32], [33]. The absence of a significant relationship between %body fat and BMD in OC and T2D groups could, again, be explained by a threshold effect, whereby additional fat mass beyond a certain cutoff of body fat percentage does not further impact BMD. This plateau effect might suggest that the detrimental effects of excessive adiposity on bone health are capped, or that the metabolic and hormonal disturbances in these groups (such as insulin resistance and altered sex steroid metabolism) play a more significant role in bone health than adiposity per se. Additionally, the effects of estrogen and androgens, which are influenced by body fat, may have complex interactions with diabetes and obesity which may obscure the direct relationship between body fat percentage and BMD [42], [43], [44].

Finally, we found a positive association between %lean mass and BMD in HWC and OC, but not in the T1D or T2D groups. Similarly, there were effect modifications on this relationship by sex in the HWC and OC groups, but not in the T1D or T2D groups. These findings are not consistent with previous studies which have found positive relationships between lean mass and BMD in adults with T2D. The lack of a positive association between lean mass and BMD in the T1D and T2D groups, which is in direct contrast to findings in populations without diabetes, might be due to the negative impact of diabetes on bone health. For example, metabolic disturbances which are closely tied to diabetes, including hyperglycemia-induced oxidative stress and production of inflammatory cytokines may negate the benefits of having a great percentage of lean mass in youth. This could also reflect the more aggressive disease pattern seen in youth-onset T2D compared to adult-onset T2D.

Although we did not initially account for the duration of obesity or diabetes, we recognize that the chronicity of these conditions likely affects BMD. To address this, we performed a sensitivity analysis in participants with diabetes, adjusting for the duration of diabetes. The results indicated that the direction of the beta coefficients and statistical significance were consistent with our primary findings, suggesting that the duration of diabetes does not substantially alter the relationship between diabetes and BMD. These findings provide additional support for the robustness of our conclusions regarding BMD in individuals with diabetes.

Major strengths of our study include a large sample size encompassing four unique participant groups (i.e., T1D, T2D, HWC, and OC) which allowed for stratification by sex and age to more thoroughly characterize the relationships between diabetes, body composition, and BMD.

Our study had several notable limitations. A key limitation is the use of whole-body assessment of BMD by DXA vs. the gold-standard method of assessing lumbar hip or spine. Site-specific BMD measurements can be more sensitive to localized changes in bone health and may capture variations in BMD between different regions of the body. The use of whole-body BMD allows for a comprehensive assessment of overall bone health, however, it may not always correlate directly with site-specific BMD measurements and may not capture how the effect of BMI on BMD differs between body regions and specific bones. Additionally, our study was limited by the absence of bone mineral content (BMC) and Z-scores, which are important for evaluating bone health in children and adolescents. BMC provides information on total bone mass, while Z-scores adjust for age, sex, and body size, making them particularly relevant in pediatric populations. However, these measures were not available in our dataset, which restricted our analysis to Bone Mineral Density (BMD) and body composition. Furthermore, we did not obtain quantitative computed tomography (qCT) data, which would better characterize bone geometry and anatomy. While our study focused on BMD, there is growing evidence that suggests that fracture risk in T2D may be influenced by factors other than BMD, such as bone quality and microarchitecture. A recent study by Van Hulten, et al., which analyzed data from qCT scans, found that that T2D was associated with higher total BMD and cortical thickness, yet a smaller total and trabecular bone area [45]. Future studies including BMC and Z-scores, as well as data obtained from qCT, would allow for a more comprehensive evaluation of bone health.

The post-hoc nature of this analysis also makes it subject to the inherent limitations of retrospective studies, as well as potential non-identification of confounding variables and an inability to determine causation. Additionally, our study does not account for the potential impact of ethnicity, demographic, or social differences on the outcomes that we collected. While we gathered this demographic data and reported it in Table 1, we did not adjust our models for differences in these variables. This would be an interesting future study. We also did not have data relating to sex steroids, habitual physical activity, or a history of prior fractures for the study participants. It is clear that while BMD is a key indicator of bone health, it does not fully explain fracture risk in young adults with diabetes. A recent meta-analysis by Thong, et al. found that fracture risk is significantly elevated in young adults with T1D compared to controls, even with only modest reductions in BMD [46]. Additionally, Rasmussen et al. found that risk factors for fractures in children and young adults with T1D included previous falls, previous fractures, hypoglycemia, and use of anxiolytic medication [47]. These results emphasize the importance of looking at factors beyond bone mass when considering overall fracture risk in this population.

Future studies may also build on our findings by leveraging novel computational tools to refine body composition assessments. Cichosz et al. recently showed the potential of machine learning algorithms to accurately predict fat and lean mass using basic anthropometric and demographic data. [48] Adopting such approaches in future studies could provide more nuanced insights into how variations in body composition influence BMD in youth with diabetes and obesity. Additionally, longitudinal studies integrating these advanced methods could help elucidate the mechanisms underlying the observed associations and inform interventions targeting bone health.

In addition, age differed to some degree in the study groups. While we adjusted for age as a continuous variable in our models, the youngest groups may not have completed puberty, especially in the male cohorts, which may have been a confounding factor which could impact our ability to make comparisons between groups. Future longitudinal studies are warranted to track changes in BMD and body composition over time, particularly during critical periods like puberty, where hormonal shifts significantly influence bone accrual. Such studies could clarify the long-term impacts of diabetes and obesity on bone health and fracture risk. Additionally, research should explore the role of sex hormones in mediating these effects, as our findings suggest that hormonal disturbances may underlie some of the observed differences in BMD between males and females.

Our study findings regarding changes in (BMD) underscore the complexity of interpreting significance in clinical contexts. While we observed statistically significant associations between BMD and various body composition metrics across different participant groups, it’s crucial to distinguish between statistical significance and clinical relevance. For instance, while the observed associations may be statistically significant, their clinical significance might be limited, especially considering the time required for notable changes in BMD to manifest. For instance, individuals who have been obese for a shorter duration of time might exhibit less pronounced changes in BMD compared to those with long-standing obesity.

Given the observed relationships between body composition and BMD, our findings suggest the need for early screening for body composition and bone health in young individuals with diabetes or obesity. Screening could focus on identifying individuals at higher risk of poor bone health due to elevated body fat percentage or inadequate lean mass. Clinicians may also consider integrating assessments of metabolic health and sex hormone levels into bone health evaluations, as these factors may influence BMD in these populations. Interventions promoting lean mass development and mitigating excess adiposity may help optimize bone health during adolescence.

In summary, our study provides insights into the nuanced relationships among BMD, biological sex, body composition, and diabetes type in young individuals. Through an analysis of data from six cross-sectional cohort studies involving 325 participants, we observed sexual dimorphism in BMD, with males exhibiting higher BMD than females in individuals with T1D and HWC, but not in those with T2D or OC. The associations between BMD and body composition metrics varied across the groups, highlighting the complex interplay between these factors. While BMI was positively correlated with BMD in all groups, body fat percentage was inversely associated with BMD in HWC and the T1D group but not in the T2D or OC groups. Additionally, higher %lean mass was significantly associated with higher BMD in HWC and OC groups, but not in the groups with diabetes. These findings underscore the importance of understanding these relationships for bone health during adolescence, particularly in the context of diabetes and obesity. Further research integrating additional variables and longitudinal studies is warranted to elucidate the mechanisms underlying these associations and inform strategies for optimizing bone health in young individuals with obesity and diabetes.

Supplementary Material

Supplementary Material

Supplementary Material

This article contains supplementary material (https://doi.org/10.1515/jpem-2024-0254).