Diabetes Risk Factors and Bone Microarchitecture as Assessed by High-Resolution Peripheral Quantitative Computed Tomography in Adults With Long-standing Type 1 Diabetes

Evidence continues to emerge supporting the deleterious effect of type 1 diabetes on skeletal health (1). Of particular concern, type 1 diabetes is associated with a nearly fivefold greater risk of hip fracture (2). Hip fractures are associated with significant morbidity and mortality that are magnified in the setting of diabetes. Fractures are of growing concern as life expectancy increases for those with type 1 diabetes (3).

As expected with higher fracture risk, bone mineral density (BMD) in people with type 1 diabetes is lower than in the general population. However, the modest decrease in BMD does not account for the substantial increase in hip fracture risk (4,5). This discrepancy suggests that other factors that impact bone strength, including bone geometry and microarchitecture, may be adversely affected by type 1 diabetes. However, data regarding bone microarchitecture in type 1 diabetes remain limited (2,6).

Bone microarchitecture as well as bone geometry are measured using high-resolution peripheral quantitative computed tomography (HR-pQCT). Outcomes include volumetric BMD, cross-sectional area, and microarchitecture of both trabecular and cortical compartments, which can be combined with finite element analysis (FEA) to provide an estimate of bone strength. Characterizing microarchitecture has important clinical relevance as HR-pQCT indices are known to be an important predictor of fracture risk beyond standard BMD measurements (7). We performed HR-pQCT in a well-phenotyped cohort of adults with type 1 diabetes from the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) study and a group of control participants without diabetes to determine whether type 1 diabetes and its complications are associated with bone geometry and microarchitecture. We hypothesized that type 1 diabetes and diabetes-related risk factors (suboptimal glycemic control, advanced glycation end product [AGE] accumulation, and microvascular complications) would be associated with deficits in trabecular and cortical bone.

The study design is a cross-sectional study embedded in DCCT/EDIC, a long-term study of participants with type 1 diabetes. HR-pQCT data were compared between EDIC participants with type 1 diabetes and control participants without diabetes. In addition, among EDIC participants, the associations between diabetes-related risk factors and HR-pQCT measurements, including bone geometry, BMD, microarchitecture, and failure load, were assessed. The study was approved by the institutional review boards of all participating centers, and all participants gave written informed consent.

The DCCT/EDIC study has been previously described in detail (8,9). Briefly, between 1983 and 1989, 1,441 participants with type 1 diabetes (aged 13–39 years) enrolled in the DCCT, a multicenter, randomized controlled clinical trial designed to compare the effects of intensive and conventional diabetes therapy on the onset and progression of microvascular complications. After an average of 6.5 years of follow-up, the DCCT ended, and all participants were encouraged to adopt intensive therapy. In 1994, 96% of the surviving DCCT cohort enrolled in the EDIC observational follow-up study. During EDIC years 24–26 (2017–2019), all surviving participants at the 27 EDIC clinics were invited to participate in the EDIC Skeletal Health ancillary study. At six of the clinics, HR-pQCT scanners were available, and HR-pQCT scans were acquired. Among the 247 active EDIC participants identified across these six clinics, 183 (73%) were enrolled, and HR-pQCT measurements using a second-generation scanner (XtremeCT II; Scanco Medical AG, Brüttisellen, Switzerland) were completed (Fig. 1).

The study aimed to recruit 100 control participants without type 1 or type 2 diabetes (exclusion criteria) from the same six clinics. For the control participants, the first priority was to recruit spouses of EDIC participants (defined as partners by opposite- or same-sex marriage, civil union, or domestic partnership). There were no age limits for spousal control participants. The second priority was to recruit other family members or friends of EDIC participants. Each nonspousal control participant was age-matched to an EDIC participant within ±5 years of the EDIC participant’s age but was not matched by sex. This resulted in 103 control participants recruited, 94 of whom underwent second-generation HR-pQCT measurements (Fig. 1). The range of current HbA_1c among control participants was 4.7–6.2%, below the threshold of 6.5% for diabetes.

The HR-pQCT scans were acquired once for EDIC and control participants during March 2018 to September 2019. The distal metaphysis of the radius and tibia and diaphysis of the tibia were imaged using second-generation HR-pQCT scanners (10). The nondominant limb was selected for scanning unless participants reported prior fracture of the wrist or ankle, metal shrapnel or implant, or recent nonweight bearing of the nondominant limb for >6 weeks. For these exceptions, the contralateral limb was scanned. The distal metaphyseal scan volumes were centered at offsets corresponding to 4.0% (radius) or 7.3% (tibia) of total limb length with respect to the distal articular surfaces (11), while the diaphyseal tibia was positioned at an offset corresponding to 30% of total limb length. All scans spanned 10.2 mm (168 slices) in height with a 140-mm field of view reconstructed into a 2,304 × 2,304 matrix with an isotropic nominal resolution of 60.7 μm. Operators received centralized training of the HR-pQCT protocol and completed a web-based training module to standardize scan positioning (12). All scanners were calibrated prior to being used in the current study, and a master density phantom was scanned on each scanner to cross calibrate density measures (13). Each local density quality control phantom was scanned daily to monitor for values that fell outside of the nominal precision range (8 mgHA/cm³). A custom data transfer pipeline automatically transferred image data from each scanner to the University of California, San Francisco (UCSF), HR-pQCT evaluation center for centralized analysis and quality assurance.

A central reader at UCSF read all images for motion artifacts and used an established semiquantitative five-point grading system (1 = superior, 5 = poor) to score image quality. Images scored <4 were included in the analytic data set (14). Scans satisfying this criterion were acquired for 96% (533 of 554) (distal and diaphyseal tibia) and 90% (249 of 277) (distal radius) of EDIC participants and control participants. A fully automated analysis pipeline was developed to segment the radius and tibia for quantification of bone density and structure (15). Segmentation accuracy was evaluated visually and manually corrected, as needed, by a single reader (16).

BMD and cross-sectional area of the total, cortical, and trabecular compartments were measured. Cortical porosity and trabecular thickness, separation, and number were calculated using direct three-dimensional morphometry (17). Linear elastic micro-FEA simulating a 1% axial compression was performed assuming a homogenous elastic modulus of 10 GPa and a Poisson ratio of 0.3. The failure load was estimated by calculation of reaction force where 5% of elements exceed a local effective strain of 1.0% (18). Models were computed with 8× parallelization using the Scanco FE Solver version 1.12 for Linux) (Scanco Medical AG) on UCSF’s Wynton high-performance computing cluster, a distributed Sun Grid Engine–managed CentOS cluster comprising 449 nodes (12,572 cores), each with a minimum of 48 GB of local memory.

Risk factors and coexisting diabetes-related complications were assessed in EDIC participants. Annual EDIC assessments included self-reported menopausal status, medication use, smoking, alcohol consumption, and history of diagnosed fracture and measurements of weight. HbA_1c was measured by high-performance liquid chromatography quarterly during DCCT and annually during EDIC. The DCCT/EDIC time-weighted mean HbA_1c was calculated by weighting each value by the time interval since the last measurement. Skin intrinsic fluorescence, a measure of skin AGEs, was obtained from the skin on the underside of the left forearm near the elbow using the SCOUT DS skin fluorescence spectrometer once during EDIC years 16–17 (2009–2010), 8 years prior to the HR-pQCT measurements.

Standardized seven-field fundus photographs were obtained every 6 months during DCCT and in one-quarter of the cohort annually during EDIC and graded centrally (19). Proliferative diabetic retinopathy (PDR) was defined by neovascularization observed on fundus photograph grading or self-report and/or confirmed by scatter photocoagulation at any time during DCCT/EDIC (20).

Estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation from serum creatinine measured annually. Albumin excretion rate (AER) was measured annually during DCCT and on alternate years during EDIC. AER was calculated from 4-h urine samples from DCCT baseline to EDIC year 18 and subsequently estimated from spot urine samples using a validated equation based on urine albumin and creatinine concentrations (21). Macroalbuminuria was defined as AER ≥300 mg/day.

Diabetic peripheral neuropathy (DPN) was assessed once in every participant during years 13–14 (2006–2007) and defined by the presence of confirmed clinical neuropathy, requiring at least two abnormal findings among symptoms, sensory signs, or reflex changes consistent with DPN as assessed by a qualified neurologist plus abnormal nerve conduction studies in at least two anatomically distinct nerves (22). In addition, the Michigan Neuropathy Screening Instrument (MNSI), comprising a structured foot examination and assessment of ankle reflexes and distal vibration perception, was obtained annually during EDIC. A clinical examination score ≥2.5 was considered evidence of peripheral neuropathy. The MNSI score obtained at the visit concurrent with HR-pQCT scans was used.

Serum 25-hydroxyvitamin D2 [25(OH)D2] and 25(OH)D3 were measured using liquid-chromatography tandem-mass spectrometry at the EDIC central biochemistry laboratory once at the time of the HR-pQCT visit. Interassay coefficients of variability were 7.5% for 25(OH)D2 and 5.4% for 25(OH)D3 at a mean concentration of 2 ng/mL. 25(OH)D was calculated as the sum of 25(OH)D2 and 25(OH)D3.

Control participants attended a clinic visit that included a self-reported medical history, assessment of current medications, and measurement of height and weight. Fasting blood was drawn for measurement of HbA_1c, serum creatinine, and 25(OH)D.

Adjusted differences, expressed as the absolute difference and the percent difference, between EDIC participants and control participants were assessed using generalized estimating equation models (23). The generalized estimating equation models used an identity link with compound symmetry covariance structure to account for the correlation between EDIC participants and control participants. Models were adjusted for known confounders of the relationship between diabetes and bone health: age, sex, menopausal status, weight, and corresponding limb length. We tested for interaction between diabetes status and sex. Of the 30 outcomes, only 1 was statistically significant, so we report results for the cohort as a whole. Serum vitamin D, bone-active medications (osteoporosis medications, oral glucocorticoids, thyroid medications, thiazide diuretics, and hormone replacement therapy [HRT]), current smoking, and alcohol consumption were also considered for inclusion in the models. Serum vitamin D, osteoporosis medications, and oral glucocorticoids, but not thyroid medications, thiazide diuretics, HRT, smoking, or alcohol consumption, were statistically significant in the models for several outcomes and were retained.

Among EDIC participants only, separate linear regression models, adjusted for age, sex, menopause status, weight, and limb length, evaluated the associations of 30 different HR-pQCT outcomes with diabetes-related risk factors and complications. More specifically, we estimated 1) the effect of time-weighted mean HbA_1c, 2) the independent effect of the time-weighted mean HbA_1c and skin AGEs by including both in the same models, and 3) the effect of each complication (e.g., PDR) in separate models adjusted for HbA_1c and skin AGEs. The potential pathways were prespecified using directed acyclic graphs. We also considered potential confounding by serum vitamin D, osteoporosis medications, and oral glucocorticoids identified as associated with the bone outcomes. Additional adjustment for these variables did not substantially alter the point estimates from these models. We also found no evidence that duration of diabetes was a confounder of models with time-weighted HbA_1c and skin AGEs. We therefore chose to report results for the more parsimonious models. The eGFR in the linear regression model was presented with a decrease of every 20 mL/min/1.73 m², approximately 1 SD of eGFR among EDIC participants.

All analyses were performed using SAS 15.2 software (SAS Institute, Cary, NC). Results nominally significant at P ≤ 0.05 (two-sided) are cited. There were few missing exposures or covariates (Table 1) or missing outcomes (Fig. 1), and we did not attempt to impute values. Participants with missing values were excluded from the models. Given the observational nature of our study, no adjustment for multiple testing was performed, and therefore, the findings should be interpretated with care.

Data collected for the DCCT/EDIC study through 30 June 2017 are available to the public through the National Institute of Diabetes and Digestive and Kidney Diseases Central Repository (https://repository.niddk.nih.gov/studies/edic/). Data collected in the current cycle (July 2017–June 2022) will be available within 2 years after the end of the funding cycle.

The mean age of the 183 EDIC participants (50.8% women) was 59.5 ± 7.0 years, and the mean duration of type 1 diabetes was 38 ± 5 years. Among the 94 control participants (65.6% women), the mean age was 60.5 ± 8.2 years. History of any self-reported fracture was reported in 41% of EDIC participants and 31% of control participants (Table 1). The distribution of HR-pQCT outcomes among EDIC and control participants stratified by sex is presented in Supplementary Table 1.

Adjusted for age, sex, menopausal status, weight, corresponding limb length, serum vitamin D, osteoporosis medications, and oral glucocorticoids, EDIC participants had significantly lower BMD at the distal radius and the distal tibia compared with control participants, with a similar trend at the diaphyseal tibia (Table 2). Both trabecular and cortical BMD were lower at these distal sites in EDIC participants compared with control participants.

EDIC participants had significantly larger total area at all three sites compared with control participants (radius mean difference 15.5 mm² [P = 0.030]; distal tibia 48.1 mm² [P = 0.003]; diaphyseal tibia 13.3 mm² [P = 0.011]). EDIC participants had significantly larger trabecular area at both tibial sites (distal mean difference 55.2 mm² [P = 0.002]; diaphyseal 13.0 mm² [P = 0.001]) and significantly lower cortical thickness at the distal tibia (−0.13 mm; P = 0.004) compared with control participants.

In addition, EDIC participants had significant deterioration in microarchitecture at the radius with lower trabecular number (mean difference −0.07 mm; P = 0.010) and lower trabecular thickness (−0.004 mm; P = 0.035). The estimated failure loads in EDIC participants, derived by FEA, were nominally but not significantly lower at all sites compared with the control participants.

Among EDIC participants, HbA_1c was associated with lower trabecular BMD at the distal radius but not with microarchitecture or failure load (Table 3, model 1). When adjusted for skin AGEs, this association was attenuated and no longer statistically significant (model 2), suggesting a mediating effect of skin AGEs on the relationships between HbA_1c and trabecular BMD at the radius. Skin AGEs were associated with lower trabecular BMD and trabecular number at the radius, independent of HbA_1c (model 2). At the distal tibia, HbA_1c was associated with lower BMD, trabecular BMD, trabecular thickness, and failure load (model 1). When adjusted for skin AGEs, the significant associations with HbA_1c were attenuated and no longer statistically significant (model 2), suggesting a mediating effect of skin AGEs on the relationships between HbA_1c and bone parameters. Skin AGEs were associated with lower total BMD, trabecular BMD, trabecular thickness, and failure load, independent of HbA_1c (model 2). In addition, at the diaphyseal tibia, there were no significant associations with HbA_1c (model 1). Skin AGEs were associated with lower BMD, larger trabecular area, and lower cortical thickness, independent of HbA_1c (model 2).

Models for individual complications were adjusted for the two confounders: time-weighted HbA_1c and skin AGEs (Table 3, model 2 plus any PDR, eGFR, any AER ≥300 mg/day, DPN, or MNSI ≥2.5). At the distal radius, diabetes complications were significantly associated with deficits in trabecular microarchitecture. Decreased eGFR and any AER ≥300 mg/day were associated with lower BMD and lower cortical thickness. Decreased eGFR was also associated with lower trabecular number, and any AER ≥300 mg/day was associated with lower cortical BMD but with larger total and trabecular area. DPN, defined as confirmed clinical neuropathy, was associated with higher cortical BMD and smaller trabecular area. At the distal tibia, diabetes complications were significantly associated with deficits in trabecular microarchitecture. Both lower eGFR and DPN were associated with lower trabecular number, while PDR was associated with lower trabecular thickness and failure load. In addition, at the diaphyseal tibia, PDR was associated with smaller cortical area and lower failure load. MNSI was associated with a smaller total area.

The DCCT/EDIC study provides the largest type 1 diabetes cohort to date with HR-pQCT measurements. Bone density, geometry, and microarchitecture in these adults with a long duration of type 1 diabetes differed from control participants, with EDIC participants exhibiting lower BMD, larger total and trabecular area, and deficits in microarchitecture. Among EDIC participants, risk factors associated with lower BMD included higher HbA_1c, higher skin AGE levels, macroalbuminuria, and lower eGFR. Macroalbuminuria was associated with larger total area and lower cortical thickness at distal radius. Factors associated with deficits in trabecular microarchitecture included suboptimal glycemic control, higher skin AGE levels, retinopathy, lower eGFR, and peripheral neuropathy.

Previous studies based on DXA (5) and HR-pQCT measurements have found lower BMD in type 1 diabetes. Our finding of larger total area is also consistent with the study by Shanbhogue et al. (6) that identified larger total area at the radius in adults with type 1 diabetes. Reduced cortical thickness was reported at the distal radius by Shanbhogue et al., while we identified this deficit at the distal tibia. Consistent with our findings, Xu et al. (24) reported lower trabecular BMD and trabecular thickness in type 1 diabetes at the radius.

We hypothesize that the larger total area of peripheral bone in type 1 diabetes, accompanied by lower cortical thickness, may represent a compensatory mechanism to preserve bone strength for habitual loading, expanding bone size in response to accelerated endosteal expansion, lower BMD, and poorer microarchitecture. Indeed, this combination has been identified as the hallmark of aging bone (25). A similar pattern of peripheral bone geometry was reported in a study using pQCT among postmenopausal women with early onset of type 1 diabetes (26), though the opposite has been observed in the femoral neck by DXA-derived hip structural analysis (27). This may reflect differences in the effect of type 1 diabetes on axial versus peripheral bone, though in general populations, only moderate agreement has been observed between femoral neck cross-sectional area by DXA and distal extremity bone total area by HR-pQCT (28).

The estimated failure load in axial compression for habitual loading for type 1 diabetes tended to be lower but was not statistically different from control participants. Previous studies (6,29) have also reported lower bone strength in type 1 diabetes, but the only statistically significant difference was at the tibia (29). Our results suggest that the observed wide bone phenotype provides an adequate compensatory mechanism in EDIC. These findings seem surprising as estimated failure load at the radius and tibia predicts fracture in broader populations, and higher fracture risk is an established characteristic of type 1 diabetes. However, the average age of EDIC participants is only 59 years. Wider bones may leave them more vulnerable to a rapid increase in fracture risk as ability to mitigate further bone loss with aging may be more limited (30). In addition, the estimated failure load is specifically for axial compression, and larger total area may not be protective for other loading conditions. Finally, the FEA models assume equivalent bone tissue material properties across individuals; however, material property deficits, such as bone brittleness due to accumulation of AGEs, are not measured by HR-pQCT and not accounted for in the FEA but may contribute substantially to fracture risk in type 1 diabetes.

We found that higher HbA_1c was associated with lower BMD at the distal radius and distal tibia among those with type 1 diabetes. We previously reported a similar association between higher HbA_1c and lower BMD by DXA in EDIC (31). Higher HbA_1c was also associated with lower trabecular thickness and lower failure load at the distal tibia. We are not aware of other reports relating HbA_1c and HR-pQCT parameters in adults with type 1 diabetes. In a study using pQCT, glycemia was inversely associated with cortical BMD and cortical area at the radius in type 1 diabetes (32). Hyperglycemia has a negative impact on osteoblasts (33), and in this cohort, we have previously reported that hyperglycemia is associated with reduced bone formation (34).

We also found that skin AGEs were associated with lower BMD among those with type 1 diabetes independent of cumulative glycemic control, consistent with our previous finding in EDIC of a negative association between skin AGEs and BMD by DXA (31). Skin AGEs were also associated with lower failure load at the distal tibia. Our data also show, for the first time, that skin AGEs are a risk factor for trabecular microarchitectural deficits in type 1 diabetes, independent of HbA_1c.

AGE accumulation is increased in the setting of hyperglycemia, but other factors, notably oxidative stress and inflammation, contribute to AGE levels (35,36). In this study, to assess whether AGE levels were simply a marker of hyperglycemia, we adjusted our models using excellent long-term measures of glycemic control, capturing >30 years of exposure. The observed independent associations between skin AGEs and HR-pQCT outcomes, adjusted for long-term glycemic control, suggest that AGE levels represent more than cumulative effects of hyperglycemia (36).

The observed associations between skin AGEs and bone outcomes may represent a direct detrimental effect of AGEs on bone (37). It is also possible that the observed associations represent the effects of increased inflammation on bone outcomes, as AGEs are associated with greater inflammation (38,39).

Notably, the negative associations between skin AGEs and HR-pQCT parameters were generally present in the trabecular, but not the cortical, compartment. This might be due to a greater amount of AGEs in trabecular bone (40). Since nonenzymatic glycation is a surface-based phenomenon, the greater surface-to-volume ratio in the trabecular compartment might predispose to a higher accumulation of AGEs. Indeed, trabecular bone has increased access for sugars to reach the bone surface and interact with amino acids to form AGEs (40), and trabecular bone accumulates microcracks in the presence of increased AGEs (41). Alternatively, we may have observed associations in the trabecular compartment because of greater precision of HR-pQCT in measuring trabecular microarchitecture (trabecular number and spacing) compared with cortical microarchitecture (cortical porosity).

Most of the significant associations that we observed between skin AGEs and microarchitectural deficits were at the tibia, a weight-bearing site. AGEs induce apoptosis and reduce functioning of osteocytes, which are responsible for sensing mechanical loading (42). A degraded response to loading by osteocytes may be particularly detrimental for bone microarchitecture in the context of weight bearing in aging and long-duration type 1 diabetes. The lack of an association with microarchitecture at the radius might also reflect greater variability at this site due to greater motion artifact.

In our analyses, reduced renal function was associated with several bone deficits, including lower BMD and reduced cortical thickness at the distal radius and lower trabecular number at the distal tibia. This is consistent with our prior report that kidney disease is an important clinical determinant of low BMD (31) and increased bone turnover (34) in EDIC. Others have also demonstrated cortical bone loss and increased cortical porosity in people with more advanced chronic kidney disease (43,44). Reduced renal function affects bone through a number of mechanisms, including changes in circulating regulatory hormones that promote net increased bone resorption, such as parathyroid hormone (45).

We found that peripheral neuropathy was associated with reduced trabecular microarchitecture at the distal tibia. A previous study reported an association of peripheral neuropathy with increased cortical porosity, but not with trabecular microarchitecture, in type 1 diabetes (46). Surprisingly, we also observed that peripheral neuropathy was associated with higher cortical BMD at the distal radius. However, given the large number of variables that were examined, this may be a chance finding. We also observed associations of retinopathy with decreased bone strength and decreased cortical area at the diaphyseal tibia. We suggest that in this relationship, retinopathy does not directly affect bone but, rather, is a marker of microvascular damage in kidney or bone that more directly contributes to bone disease in type 1 diabetes. Clinically, our results suggest that maintenance of adequate glycemic control combined with additional efforts to reduce AGE accumulation and prevent microvascular complications, particularly kidney disease, may prevent loss of bone density and microarchitecture, which may in turn reduce fracture risk.

Our study has several strengths. This is the largest study of HR-pQCT measurements in long-standing type 1 diabetes. In addition, the DCCT/EDIC study includes a comprehensive assessment of long-term diabetes-related risk factors and complications, including time-weighted HbA_1c as a long-term measure of glycemic exposure, measurement of AGE accumulation, and assessment of microvascular complications of the eyes, kidneys, and nerves. The use of a central HR-pQCT reading center with standardized acquisition and review of the images across the six EDIC sites along with the availability of state-of-the-art new-generation scanners is a major strength.

Limitations to our study include a single cross-sectional assessment of HR-pQCT, so we could not assess the temporal association between type 1 diabetes or diabetes risk factors and bone outcomes. Skin AGEs were only measured once ∼8 years before the HR-pQCT scans. However, any resulting misclassification of AGE levels was likely nondifferential with respect to the bone outcomes and would tend to attenuate actual associations. In addition, AGEs were measured by detection of autofluorescence of the skin, which may not fully reflect accumulation of AGEs in bone (47). Also, our control participants included spouses of EDIC participants, and it is possible that these spouses adopted a healthier lifestyle compared with the broader population without diabetes. While we aimed to be cautious in our interpretation of results, we acknowledge that our analyses considered multiple outcomes and that some results may be chance findings. The original DCCT cohort was selected based on willingness to participate in a randomized trial (48), and EDIC, the follow-up observational cohort, may not be fully reflective of the general type 1 diabetes community. In addition, the age range of the EDIC cohort is currently 44–74 years, not yet in the age range with the highest fracture risk. Thus, our results may not generalize to type 1 diabetes in general or to those outside the current age range of EDIC. Furthermore, the management of diabetes has advanced during the past 30 years. With improvements in glycemic control, differences between those with and without type 1 diabetes reported here may not be observed in future cohorts.

In conclusion, type 1 diabetes is associated with lower BMD, larger bone area, and reduced trabecular microarchitecture, hallmarks of the aging process in bone. Among patients with type 1 diabetes, suboptimal glycemic control, AGE accumulation, and microvascular complications, particularly reduced renal function, are associated with deficits in trabecular and cortical bone. Maintenance of adequate glycemic control combined with additional efforts to reduce AGE accumulation and prevent microvascular complications, particularly kidney disease, may preserve bone density and microarchitecture.

Acknowledgments. The authors acknowledge EDIC Skeletal Health Working Group members Valerie Arends (University of Minnesota, Minneapolis, MN), Kaleigh Farrell (Case Western Reserve University/Rainbow Babies and Children's Hospital, Cleveland, OH), Ming-Hui Lin and Victoria R. Trapani (The George Washington University, Rockville, MD), and Amisha Wallia (Northwestern University Feinberg School of Medicine, Chicago, IL). The DCCT/EDIC Research Group owes its scientific success and public health contributions to the dedication and commitment of the DCCT/EDIC participants.

Funding. The DCCT/EDIC has been supported by cooperative agreement grants (1982–1993, 2012–2017, 2017–2022) and contracts (1982–2012) with the Division of Diabetes, Endocrinology, and Metabolic Diseases of the National Institute of Diabetes and Digestive and Kidney Disease (NIDDK) (current grants U01 DK094176 and U01 DK094157) and through support by the National Eye Institute, National Institute of Neurologic Disorders and Stroke, General Clinical Research Centers Program (1993–2007), and Clinical Translational Science Center Program (2006–present).

The sponsor of this study is represented by the NIDDK Project Scientist, who serves as part of the DCCT/EDIC Research Group and plays a part in the study design and conduct as well as the review and approval of manuscripts. The NIDDK Project Scientist was not a member of the writing group of this article.

The opinions expressed are those of the investigators and do not necessarily reflect the views of the funding agencies.

Duality of Interest. The following industry contributors provided free or discounted supplies or equipment to support the DCCT/EDIC participants’ adherence to the study: Abbott Diabetes Care (Alameda, CA), Animas (Westchester, PA), Bayer Diabetes Care (North America Headquarters, Tarrytown, NY), Becton Dickinson and Company (Franklin Lakes, NJ), Eli Lilly (Indianapolis, IN), Extend Nutrition (St. Louis, MO), Insulet Corporation (Bedford, MA), Lifescan (Milpitas, CA), Medtronic Diabetes (Minneapolis, MN), Nipro Home Diagnostics (Ft. Lauderdale, FL), Nova Diabetes Care (Billerica, MA), Omron (Shelton, CT), Perrigo Diabetes Care (Allegan, MI), Roche Diabetes Care (Indianapolis, IN), and Sanofi (Bridgewater, NJ). No other potential conflicts of interest relevant to this article were reported.

The industry contributors have had no role in the DCCT/EDIC study.

Author Contributions. N.S.G., A.J.B., J.-Y.C.B., M.R.R., I.B., B.H.B., D.J.K., T.M.L., G.J.K., A.B., J.M.L., R.G.-K., I.H.d.B., and A.V.S. revised the manuscript for critical content and approved the final version. N.S.G., A.J.B., J.-Y.C.B., and A.V.S. drafted the manuscript. J.-Y.C.B. and I.B. performed the analyses. B.H.B., R.G.-K., and A.V.S. designed the study. B.H.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at The American Society for Bone and Mineral Research Annual Meeting, Vancouver, British Columbia, Canada, 13–16 October 2023.

Handling Editors. The journal editors responsible for overseeing the review of the manuscript were Steven E. Kahn and Jessica Castle.