{
  "abstract": "Introduction Poorly controlled type 1 diabetes (T1D) has been associated with impaired bone health, but the mechanisms remain unclear. We aimed to investigate whether changes in glycemic control and glucose variability are associated with skeletal health and to evaluate the roles of insulin-like growth factor I (IGF-I) and advanced glycation end-products (AGEs) in bone mineral accrual.Research design and methods This longitudinal study included adolescents with poorly controlled T1D (HbA1c >9%/75 mmol/mol), who underwent dual-energy X-ray absorptiometry (DXA) at baseline and after 12 months. Glycemic control was assessed using glycohemoglobin (HbA1c), continuous glucose monitoring (CGM) parameters, and glycemic load. Serum IGF-I and AGEs, specifically methyl-glyoxal-hydro-imidazolone (MG-HI), were measured. Correlation analyses and linear regression models were used to evaluate the associations between glycemic markers, IGF-I, AGEs and bone parameters.Results Altogether, 37 adolescents (48.6 % female) with T1D, with mean HbA1c 9.9% (85 mmol/mol), were followed up from mean age of 14.3 for 12 months. DXA-derived bone mineral density (BMD) z-scores at lumbar spine, proximal femur, and total body less head were approximately 0.5 SDS lower than reference values (p=0.005–0.04). The only significant change in BMD z-scores during the 12-month follow-up was an increase in proximal femur in girls. In the whole group, an increase in IGF-1 was associated with BMD accrual, while changes in HbA1c, time in range, or MG-HI were not. No vertebral fractures were detected.Conclusions Despite lower BMD in adolescents with poorly controlled T1D, neither changes in glycemic control nor MG-HI levels correlated significantly with bone health measures, while increase of IGF-1 was associated with BMD accrual. Future studies should explore alternative AGEs and use advanced bone imaging techniques to better understand skeletal fragility in T1D.",
  "authors": [
    {
      "affiliations": [
        "Children’s Hospital, Pediatric Research Center, Helsinki University Hospital, Helsinki, Uusimaa, Finland",
        "Faculty of Medicine, University of Helsinki, Helsinki, Uusimaa, Finland"
      ],
      "name": "Mari-Anne Pulkkinen"
    },
    {
      "affiliations": [
        "Children’s Hospital, Pediatric Research Center, Helsinki University Hospital, Helsinki, Uusimaa, Finland",
        "Faculty of Medicine, University of Helsinki, Helsinki, Uusimaa, Finland"
      ],
      "name": "Tero Varimo"
    },
    {
      "affiliations": [
        "Children’s Hospital, Pediatric Research Center, Helsinki University Hospital, Helsinki, Uusimaa, Finland"
      ],
      "name": "Sanna Toiviainen-Salo"
    },
    {
      "affiliations": [
        "Research Program for Clinical and Molecular Metabolism, Faculty of Medicine, University of Helsinki, Helsinki, Uusimaa, Finland"
      ],
      "name": "Taina H Härkönen"
    },
    {
      "affiliations": [
        "Children’s Hospital, Pediatric Research Center, Helsinki University Hospital, Helsinki, Uusimaa, Finland",
        "Research Program for Clinical and Molecular Metabolism, Faculty of Medicine, University of Helsinki, Helsinki, Uusimaa, Finland"
      ],
      "name": "Saila Laakso"
    },
    {
      "affiliations": [
        "Children’s Hospital, Pediatric Research Center, Helsinki University Hospital, Helsinki, Uusimaa, Finland",
        "Faculty of Medicine, University of Helsinki, Helsinki, Uusimaa, Finland"
      ],
      "name": "Anna-Kaisa Tuomaala"
    },
    {
      "affiliations": [
        "Children’s Hospital, Pediatric Research Center, Helsinki University Hospital, Helsinki, Uusimaa, Finland"
      ],
      "name": "Matti Hero"
    }
  ],
  "full_text": "WHAT IS ALREADY KNOWN ON THIS TOPIC Type 1 diabetes (T1D), especially in patients with poor glycemic control, is associated with impaired bone health, both with elevated fracture risk and lower bone mineral density (BMD).WHAT THIS STUDY ADDS During pubertal development, BMD is compromised in youth with poorly controlled T1D, especially in boys.However, during 12 months of follow-up, no association was found between markers of long-term dysglycemia and measures of bone health, as assessed by dual-energy X-ray absorptiometry.Increase in circulating insulin-like growth factor I correlated with bone mineral accrual.Vertebral fractures were absent, arguing against increased risk of childhood osteoporosis.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Future studies with larger sample size should explore alternative advanced glycation end-products and use more sensitive bone imaging to better understand skeletal fragility in T1D.Background Patients with type 1 diabetes (T1D) with optimal glycemic control suffer markedly less from long-term diabetic complications than those with poor control. Increased glycohemoglobin (HbA1C) levels predict the complication risk. On one hand, the development of diabetic complications seems to accelerate during puberty, and poor metabolic control during adolescence or young adulthood markedly increases the incidence of microvascular or macrovascular complications during subsequent years. 1–3 On the other hand, intensive treatment during adolescence seems to reduce the risk of microvascular complications, even if the control later becomes poorer.4 5 Besides these well-known diabetic complications, T1D and especially poor glycemic control of the T1D are associated with impaired bone health. T1D has been shown to be associated with elevated fracture risk and modestly low bone mineral density (BMD) at the femoral neck,6 and meta-analysis of the literature indicates especially an increase in the hip-fracture risk of patients with T1D compared with matched controls.7Interventions targeting improved glycemic control during adolescence probably have sustained beneficial effects on the overall morbidity in patients with T1D. Treatment of the disease during puberty can be complicated, and particularly, treatment adherence often declines in youth.8 To this end, we saw a need to find an approach for clinicians to motivate adolescents with poor glycemic control toward better treatment adherence. Some studies had previously evaluated motivational interviewing (MI) in the treatment of adolescent patients with diabetes, with outcomes ranging from substantial benefit9 to neutral.10 11 In these studies, MI was employed by varying professionals within one study, and the study populations were quite small.9–11 In our randomized controlled study including 47 adolescents with poor glycemic control, we found that improved skills of the clinicians in applying the MI method may result in improved glycemic control of the patients.12 We also found that poor glycemic control affects the vascular health in adolescents with T1D; however, improvement in glycemia was not associated with improved vascular health parameters in 1 year’s follow-up time.13Prospective reports of bone health parameters in adolescents with poorly controlled T1D are scarce, and the underlying mechanisms remain unclear. The aim of the present study was to clarify if changes in glycemic control and glucose variability are associated with changes in parameters of skeletal health in a longitudinal setting. Further, we investigated whether changes in biomarkers that have previously been related to skeletal health in T1D, that is, changes in insulin-like growth factor I (IGF-I)14 and advanced glycation end-products (AGEs),15 explain changes in parameters of skeletal health in our cohort.Methods In the present study, MI was integrated into clinicians’ daily practice, as a part of normal clinical visit. In this randomized controlled trial, our primary hypothesis was that applying MI during regular clinical visits results in better acceptance and subsequently enhanced metabolic control in adolescents with poorly controlled T1D. As we saw improvements in glycemic control in some of the patients, 12 we hypothesized that bone health in these individuals had also improved during the follow-up. Bone health outcomes, as measured by dual-energy X-ray absorptiometry (DXA), were included in the study protocol as secondary endpoints and were part of the prespecified data collection and analysis plan.Patients Description of the study population, including inclusion and exclusion criteria, has been described previously. 12 In brief, patients recruited in 2015–2017 had been diagnosed with T1D for more than 2 years earlier, were 12.0–15.99 years old, and they had poor glycemic control (HbA1c >9 %/75 mmol/mol on two consecutive visits). Patients from the Helsinki University Hospital Pediatric Diabetes Units participated in DXA examination. Instead of skeletal parameters, power calculations of the study were based on the primary endpoint, that is, HbA1c change between the treatment groups. Clinically significant change was set at 1.0%, and SD 1.24 was used in the calculation based on our previous experience.12 13 With the power of 80% and alpha of 0.05, 50 patients were needed to complete the study; unfortunately, we were able to recruit only 47 patients, and of those, DXA results from both time points (0 and 12 months) were available for 37 patients. Randomization was performed with sealed containers in permuted blocks of six patients with balanced numbers of intervention and control subjects for each treating physician.Study protocol Study protocol has been described in detail previously 12 13 and is presented in figure 1. Every visit included usage of educational material, and in the MI group, usage of motivational interviewing during the patient visits.Figure 1Flowchart of study design. CGM, continuous glucose monitoring; DXA, dual-energy x-absorptiometry; HbA1C, glycohemoglobin.Laboratory assessments were performed in conjunction with standard hospital procedures and quality control. HbA1c levels were measured on every visit from fingertip samples (Afinion) and at baseline and 12 months’ visits from venous blood sample.16 Fasting venous blood samples were obtained at baseline and at 12 months and stored at −80°C. Markers known to be important for bone health (25-hydroxy vitamin D [25OHD], alkaline phosphatase (ALP), and IGF-1) and markers indicating levels of advanced glycation end-products (methyl-glyoxal-hydro-imidazolone, MG-HI) were evaluated from the stored sera collected at the start and at 12 months. IGF-1 was analyzed with the IMMULITE 2000 immunoassay system and hospital central laboratory (HUSLAB) age-matched reference values were used17 and were as follows: Boys: age: 12–13 years: 12–60 nmol/L, 14–15 years: 27–67 nmol/L, 16–18 years: 16–55 nmol/L. Girls: age 12–13 years: 19–71 nmol/L, 14–15 years: 27–58 nmol/L, 16–18 years: 23–56 nmol/L. 25OHD concentration was analyzed using electrochemiluminescence method (ECLIA, Roche Diagnostics), and ALP was analyzed photometrically using standard methods at the University Hospital’s core laboratory HUSLab. The serum levels of MG-HI at baseline and at 12-month time point were measured from stored (−80°C) sera, using ELISA (OxiSelect MG Competitive ELISA Kit #STA-811; Cell Biolabs, San Diego, California, USA) according to the manufacturer’s instructions. ELISA samples were run in duplicate. Coefficient of variation (CV) for concentrations was 4.2% from at least three independent tests for ELISA Kit.Finnish data for height, weight and body mass index (BMI) for children and adolescents were used as a reference data for weight, height and BMI.18Total exposure of the body to high blood glucose levels over time was calculated based on duration of diabetes and on mean HbA1c per each year after the diabetes diagnosis, for example, duration of diabetes of 5 years and mean yearly HbA1c level of 60 mmol/mol results in glycemic load 5×60 = 300 and used as a marker of prolonged exposure of tissues to elevated glucose levels. We chose this metric because, to the best of our knowledge, there are no validated alternatives that capture long-term cumulative exposure to hyperglycemia in pediatric patients with T1D.Continuous glucose monitoring (CGM) was performed at baseline, 6 months, and 12 months (iPro2 professional continuous glucose monitor; Medtronic Diabetes, Northridge, California, USA, or patients own CGM—Medtronic Enlite in Veo or 640G insulin pump: Medtronic Diabetes, or Freestyle Libre; Abbott Diabetes Care) for a 6-day period to measure interstitial glucose levels for analysis of glycemic variability. SD of sensor glucose values (SD), CV, and time in range (TIR) were used as parameters to define glycemic variability.19DXA was performed at baseline and at 12 months for 37 of the participants for analyses of BMD (total body less head (TBLH), proximal femur, lumbar spine (LS)), bone mineral apparent density (BMAD) and body composition, using the Hologic Discovery device (Hologic, Bedford, Massachusetts, USA). BMD z-scores represent age, sex, and equipment-specific normative data for Caucasian children.20 Vertebral morphology was analyzed by Genant criteria21 by experienced pediatric radiologist (ST-S) from lateral views of the DXA scan (instant vertebral assessment) at baseline and at 12 months.Statistics Statistical analyses were carried out with SPSS Statistic for Windows (V.25, Chicago, Illinois, USA). The results are shown as mean (SD). Comparisons between two groups were analyzed with independent samples t-test and within-group comparison with paired samples t-test. Correlations were analyzed with Spearman’s rank correlation. A linear regression model was used when multiple variables were evaluated. In the model, we first evaluated whether the changes in DXA parameters (LS BMD, TBLH BMD, and BMAD) were explained by the changes in markers of glycemia (HbA1c and TIR during the follow-up) and next whether the changes in DXA parameters were explained by the changes in MG-HI, IGF-1, and BMI. The analyses were adjusted for gender. The level of statistical significance was set at p value <0.05.Ethical issues Written informed consent has been obtained from the participants and their guardian. Collected patient data are stored in the university hospital’s patient information system. Approval for the protocol has been gained from the Helsinki University Hospital Ethical Committee (December 18, 2014). The study has been registered in ClinicalTrials.Gov (ID number NCT02637154). The principles of Good Clinical Practice and the Declaration of Helsinki are applied. Only study group personnel are allowed to access data.Results Skeletal findings and their correlates at baseline At the start, BMD z-scores of LS, proximal femur and TBLH were lowered (−0.5±1.2, p=0.041, –0.51±1.2, p=0.012, –0.48±0.94, p=0.005). All DXA variables are shown in table 1. At the same time, IGF-I levels were low compared with age-matched reference values and showed a negative correlation with total exposure of the body to high blood glucose levels over time (r=−0.43, p=0.011). Both IGF-1 and ALP concentrations declined in girls during follow-up, reflecting the attainment of mature pubertal development (Tanner stage 5 for breast development, table 1). The 25-D levels were low (54.9 (13.4) nmol/L) compared with levels in our whole patient population (74 (25) nmol/L, n=341).22 The levels of MG-HI showed no correlation with HbA1c, TIR or total exposure of the body to high blood glucose levels over time at the same time point (p=0.68–0.92).Table 1Bone mineral content (BMC), bone mineral density (BMD) and bone area at 0 and 12 months’ time pointsFemalesN=18P valueMalesN=19P value†0 month12 months0 month12 monthsAge (years)14.3 (1.0)14.3 (0.8)Diabetes duration (years)8.7 (3.7)8.6 (3.5)Height (cm)164.9 (5.7)165.9 (6.1)0.005170.4 (6.6)176.1 (6.1)<0.001Weight (kg)65.6 (11.6)67.2 (11.3)0.03661.8 (13.9)71.0 (14.9)<0.001Height SDS0.14 (0.9)0.09 (0.9)0.750.1 (0.8)0.2 (0.8)0.73BMI (kg/m2)23.5 (3.6)24.4 (4.1)<0.00121.1 (3.9)22.8 (4.3)<0.001BMI z-score0.90 (0.90)1.0 (0.9)0.960.2 (1.0)0.5 (0.9)0.15Tanner stage (n/%) M2/G21 (6)02 (10)0 M3/G301 (6)6 (30)1 (5) M4/G43 (17)1 (6)9 (50)8 (40) M5/G514 (77)16 (88)2 (10)11 (55)Proximal femur Area (cm2)32.0 (2.7)32.2 (3.3)0.3135.7 (3.3)38.4 (2.3)<0.001 BMC (g)29.1 (4.9)30.6 (5.0)0.00831.7 (6.4)36.1 (6.1)<0.001 BMD (g/cm2)0.91 (0.11)0.95 (0.10)<0.0010.87 (0.13)0.94 (0.15)<0.001 BMD z score−0.16 (1.03)0.08 (0.90)0.038−0.63 (1.30)−0.52 (1.28)0.30Lumbar spine Area (cm2)54.5 (6.0)56.3 (5.8)0.01257.0 (4.2)61.9 (5.4)<0.001 BMC (g)50.1 (10.5)54.5 (9.4)<0.00143.4 (7.5)51.2 (9.4)<0.001 BMD (g/cm2)0.92 (0.12)0.96 (0.09)<0.0010.75 (0.11)0.82 (0.12)<0.001 BMD z-score−0.26 (1.19)−0.08 (0.88)0.89−0.73 (1.14)−0.84 (1.16)0.13 BMAD0.24 (0.03)0.25 (0.03)0.0010.19 (0.03)0.21 (0.03)<0.001Total body less head BMC (g)1493.4 (242)1605.7 (209)<0.0011447.9 (274)1651.3 (272)<0.001 BMD (g/cm2)0.90 (0.07)0.93 (0.06)<0.0010.87 (0.08)0.91 (0.08)<0.001 BMD z-score−0.22 (0.97)−0.11 (0.82)0.55−0.75 (0.86)−0.83 (0.84)0.47HbA1c (%)10.0 (1.3)10.0 (1.6)0.979.9 (1.0)9.7 (1.0)0.56TIR (%)38.8 (16.7)32.2 (11.6)0.1133.7 (14.6)35.1 (15.7)0.1TEB*382.2 (238.5)406.8 (194.1)ALP (U/L)147.9 (57.8)121.6 (41.8)0.027316.3 (76.7)252.3 (86.1)0.085IGF-1 (nmol/L)30.8 (6.9)27.4 (6.3)0.03828.9 (8.4)28.8 (5.6)0.7225-OH vitamin D (nmol/L)52.4 (10.2)52.9 (12.9)0.9357.3 (15.8)58.5 (20.4)0.89Statistically significant P-values are shown in bold.*Total exposure of the body to high blood glucose levels over time (TEB) calculated as HbA1c (mmol/mol) × diabetes duration (years).†Statistical comparisons conducted using paired samples t-tests.BMAD, bone mineral apparent density; BMI, body mass index; HbA1c, glycohemoglobin; IGF-I, insulin-like growth factor I; 25OHD, 25-hydroxy vitamin D; TIR, time in range.HbA1c correlated positively with baseline proximal femur BMD (table 2), but the significant correlation disappeared when it was adjusted with baseline BMI z-score. The duration of diabetes (result not shown), markers of glycemia and glycemic burden showed no significant correlations with other baseline DXA variables (table 2). Further, 25OHD and IGF-I showed no correlation with baseline skeletal parameters. In particular, the levels of MG-HI did not correlate with DXA measures, that is, LS BMD, LS BMC, LS BMD z-score, TBLH z-score, and femoral BMD z-score (p=0.30–0.68). The only parameter that correlated with several DXA measures was BMI z-score with a positive correlation with LS BMD (r=0.48, p=0.002), LS BMD z-score (r=0.45, p=0.004), hip BMD (r=0.48, p=0.002), hip BMD z-score (r=0.51, p<0.001), TBLH BMD z-score (r=0.45, p=0.006), and lumbar spine BMAD (r=0.45, p=0.004).Table 2Correlations between baseline parameters of bone health by DXA and markers of glycemia and selected predictors of skeletal health in 37 adolescents with T1DHbA1cTIRTEBALPIGF-125OHDMG-HIBMI z-scoreTotal body less head BMD0.150.09−0.11−0.330.13−0.160.0020.44†BMC/lean mass−0.040.07−0.1−0.50†0.07−0.17−0.090.08Lumbar spine BMD0.140.13−0.13−0.53†0.18−0.310.060.48† Area−0.080.29−0.180.030.180.1−0.20.08 BMAD0.160.01−0.1−0.56‡0.14−0.36*0.110.45†Proximal femur BMD0.32*0.06−0.06−0.120.11−0.28−0.030.49†Statistically significant P-values are shown in bold.*P value <0.05.†P value <0.01.‡P value <0.001.ALP, alkaline phosphatase; BMAD, bone mineral apparent density; BMC, bone mineral content; BMD, bone mineral density; HbA1c, glycohemoglobin; IGF-1, insulin-like growth factor type 1; MG-HI, methyl-glyoxal-hydro-imidazolone; 25OHD, 25-hydroxy vitamin D; TEB, total exposure of the body to high blood glucose levels over time; TIR, time in range.None of the patients had significant vertebral compressions according to Genant criteria. Slight anterior wedging or compression in the lower thoracic region (Th8 to Th12) ranging from 10% to 19% was noted in seven patients (18.4%): six male and one female participants. Their glycemic parameters (HBA1c, TIR, diabetes duration and glycemic burden) were similar to those with no such findings (p=0.44–0.85). Further, their BMD z-scores did not significantly differ from those without vertebral changes (p=0.33–0.41).Changes in parameters of skeletal health during the study During the 12 months of follow-up, only a significant increase in BMD z-scores was found at proximal femur in girls ( table 1). There was no significant difference in the changes of DXA measures between the intervention groups (MI or control group; results not shown). We next examined whether markers of glycemia (change in HbA1c and change in TIR), or mean MG-HI or IGF-I, were associated with changes in DXA measures (LS BMD, TBLH BMD, and BMAD). The linear regression model showed significant associations between increase of IGF-1 and BMD accrual at lumbar spine and whole body during the 12 months’ follow-up (table 3). Changes in HbA1c, TIR, or MG-HI were not associated with BMD accrual (table 3).Table 3Results of linear regression analyses examining the associations between changes in DXA parameters (LS BMD, TBLH BMD, and BMAD) and changes in glycemic markers (HbA1c and TIR) in model 1, and between changes in DXA parameters and changes in MG-HI, IGF-1, and BMI in model 295% CI for BBSEβLower boundUpper boundP valueBMAD ∆0–12mo Model 1  HbA1c (∆0–12mo)0.0000.0000.1480.0000.0000.540  TIR (∆0–12mo)0.0000.000−0.0020.0000.0000.993  Sex−0.0050.003−0.324−0.0110.0020.162 Model 2  IGF-1 (∆0–12mo)0.0000.0000.3410.0000.0010.108  MG-HI (∆0–12mo)−0.0010.001−0.170−0.0030.0010.384  BMI (∆0–12mo)0.0000.0000.096−0.0010.0010.645  Sex0.0000.003−0.002−0.0060.0050.992LS BMD ∆0–12mo Model 1  HbA1c (∆0–12mo)0.0000.0000.131−0,0010.0010.397  TIR (∆0–12mo)0.0000.000−0.073−0.0010.0010.632  Sex−0.0450.008−0.796−0.063−0.0280.000 Model 2  IGF-1 (∆0–12mo)0.0040.0010.5670.0010.0060.002  MG-HI (∆0–12mo)−0,0020.005−0,082−0,0120.0070.606  BMI (∆0–12mo)−0.0030.001−0.355−0,0060.0000.044  Sex−0.0270.011−0.392−0.049−0.0040.021TBLH BMD ∆0–12mo Model 1 HbA1c (∆0–12mo)0.0000.001−0,142−0,0020.0010.499 TIR (∆0–12mo)−0.0010.001−0.173−0.0020.0010.405 Sex−0.0300.012−0.516−0.055−0.0060.017 Model 2  IGF-1 (∆0–12mo)0.0030.0010.5220.0010.0060.007  MG-HI (∆0–12mo)−0.0040.005−0.137−0.0150.0070.427  BMI (∆0–12mo)−0.0010.002−0.085−0.0040.0020.643  Sex−0.0260.012−0.376−0.051−0.0010.039Both models were adjusted with gender.BMAD, bone mineral apparent density; BMI, body mass index; DXA, dual-energy X-ray absorptiometry; HbA1c, glycohemoglobin; IGF-I, insulin-like growth factor I; LS BMD, lumbar spine bone mineral density; MG-HI, methyl-glyoxal-hydro-imidazolone; TBLH BMD, total body less head bone mineral density; TIR, time in range.Discussion In line with previous meta-analysis, 23 we found that DXA-derived z-scores of BMD were approximately 0.5 SDS lower across different measurement sites in adolescents with poorly controlled T1D than in the reference population. In contrast to most previous studies,23 in our cohort of adolescents with T1D, this finding was apparent particularly in boys. The reason for this difference between the sexes in our cohort is unclear. However, it may reflect a complex interplay of biological and developmental factors in boys with poor glycemic control or behavioral factors that have not been captured by our assessments. Despite lowered BMD in our cohort, we did not find any statistically significant associations between markers of long-term dysglycemia (total exposure of the body to high blood glucose levels over time), HbA1c or CGM parameters and measures of bone health at baseline or during follow-up. This finding may seem unexpected as diabetes duration and dysglycemia have been reported to associate with fracture rate in adults.24 However, previous studies have provided mixed results on the effect of glycemic burden on skeletal health and in a recent meta-analysis of skeletal health in youth with T1D, no association between diabetes duration or HbA1c and DXA-measured BMD was found.23 While DXA is a widely used tool for assessing BMD, it provides only areal measurements and lacks sensitivity to important aspects of bone quality, such as microarchitecture and direct measures of cortical and trabecular bone compartments. The use of DXA may have limited the ability to detect subtle or compartment-specific skeletal changes related to poor glycemic control. As metabolically active trabecular bone may be particularly sensitive to the effects of chronic hyperglycemia,14 and vertebral bodies are particularly rich in trabecular bone, we also investigated vertebral morphology in our cohort, which has not been studied before in youth with T1D, to the best of our knowledge. We found no compression fractures, which are well-established markers of osteoporosis and advanced skeletal fragility. This finding is in line with previous reports on a large national cohort from the UK showing no increase in fracture rate in children or young adults with T1D.25The mechanism by which T1D impairs bone mineral accrual and in the longer term results in skeletal fragility is somewhat poorly defined. Standard glycemic metrics, such as HbA1c or glucose variability, appear to be insufficient surrogate markers of bone mineral accrual in youth with T1D. There is an evident need for broader approaches to skeletal risk assessment in this population. The accumulation of AGEs due to hyperglycemia changes bone microarchitecture and has been proposed to make bone less resistant to fractures.26 Both increased level of AGEs and impaired IGF-I production have been suggested as important contributing factors.14 15 Circulating IGF-I is low in T1D, and according to some reports, the levels of IGF-I associate with measures of trabecular bone and bone strength estimates by pQCT (peripheral quantitative computed tomography).14 27 Lowered IGF-I levels in T1D likely result from relative insulin deficiency in the portal circulation and may contribute to observed slow bone turnover in patients with T1D.28–30 In our cohort, IGF-I was expectedly low and correlated negatively with markers of glycemia. In longitudinal analysis, individual increases in IGF-1 were associated with BMD accrual at the lumbar spine and total body less head. While group-level mean IGF-1 concentrations declined during follow-up, particularly in girls, this association reflects within-subject variation rather than overall group trends. Therefore, our findings support the role for IGF-1 in bone mineral accrual during adolescence in patients with T1D, despite average declines across the cohort. Meanwhile, AGEs as measured by circulating MG-H1 showed no association with DXA-measured bone parameters in our cohort at baseline or during the follow-up. To this end, AGEs have been shown to affect bone structure,31 and diabetes, especially hyperglycemia induces protein glycation leading to higher levels of AGEs.26 In diabetes, hyperglycemia leads to an increase in the levels of AGEs, which then accumulate in bone, changing both bone matrix composition and bone strength.32 To our knowledge, the present study is the first to address AGEs and skeletal health in youth with T1D. Several uncertainties need to be considered prior to making any firm conclusions. First, circulating AGEs at certain time point may not adequately reflect AGE burden at the level of bone tissue. Second, while MG-HI is a well-documented AGE related to metabolic dysfunction, other AGEs may have a stronger direct impact on bone health in diabetes.31 To this end, serum levels of pentosidine have been associated with prevalent fractures in T1D independently of BMD.33 Thus, accumulation of AGEs in bone collagen, resulting in alterations in the material properties and quality of the bone, may not be adequately captured by DXA measurements.The current study has several strengths. First, studies addressing skeletal health in youth with T1D in a longitudinal design are scarce and provide better sensitivity to detect associations between BMD and its predictors. Second, we used up-to-date markers of glycemia, including TIR, which has not previously been used in studies addressing skeletal health in children and adolescents with diabetes. Third, we included past glycemic load as a predictor in our analyses. Fourth, we addressed skeletal health in a more comprehensive fashion compared with previous studies as we included vertebral morphology evaluation in our study design.Limitations of our study include the relatively small number of patients, which was based on power calculation related to other than bone health parameters. Additionally, DXA measurements at both time points were available only for 37 out of 47 patients. We therefore acknowledge that the study may be underpowered to detect modest but still clinically relevant associations. Second, 12-month follow-up may be too short to detect significant changes in measures of bone health in patients with T1D. Third, DXA provides a two-dimensional analysis of targeted bone regions and does not provide information on bone geometry or density in different bone compartments. Thus, the use of DXA may have limited our ability to detect subtle or compartment-specific skeletal changes related to poor glycemic control. The use of pQCT in addition to DXA would have brought more detailed information on skeletal health. Fourth, as mentioned above, AGE measurement from blood samples may be inadequate. Further, it is unclear which AGE measure best correlates with bone health in humans.34 We used MG-H1, which has previously been shown to correlate with osteocalcin levels, but more specifically in obese individuals.35 A previous study in adults with T1D indicated an association of pentosidine with bone fractures,33 and another study indicated non-carboxymethyllysine (CML) to associate with microvascular complications in T1D,36 thus both pentosidine and CML could have been better AGEs to be used in the present study as well. Additionally, each patient wore the CGM device for only 6 days, as the study was conducted between 2015 and 2018, when the 2-week wear period had not yet been established as the golden standard. Finally, physical activity, a potential determinant of bone accrual particularly at weight-bearing sites, was not assessed. This limits interpretation of the observed site-specific and sex-specific differences in BMD changes.In conclusion, adolescents with poorly controlled T1D had lower DXA-derived BMD z-scores compared with the reference population, but no association was found between bone health and markers of long-term dysglycemia, HbA1c, CGM parameters, or glycemic load. Vertebral fractures were absent, arguing against increased risk of childhood osteoporosis. In longitudinal analyses, circulating IGF-I was positively associated with BMD accrual. Contrary to previous hypotheses, circulating AGE marker MG-HI was not correlated with bone mineral accrual, although serum AGE levels may not fully reflect bone tissue accumulation, and other AGEs may have a stronger impact. Study strengths include its longitudinal design, comprehensive skeletal assessment, and advanced glycemic markers, while limitations include a small sample size, short follow-up, and reliance on DXA alone. Future research should explore alternative AGEs and use more sensitive bone imaging to better understand skeletal fragility in T1D.",
  "title": "Skeletal health in adolescents with poorly controlled type 1 diabetes: results from a randomized controlled trial",
  "uid": "9c332630-ed7f-5152-b879-7b61a73d8c8f"
}