Clinical bone assessment in diabetes
The ideal investigations to identify individuals with diabetes at elevated fracture risk remain unclear [
52]. Both type 1 and type 2 diabetes are associated with changes in areal BMD (aBMD) when measured by dual-energy x-ray absorptiometry (DXA) (Fig.
1). In type 1 diabetes, BMD is low, probably because of inadequate accrual of peak bone mass due to hypoinsulinaemia and lower levels of IGF-1 [
3,
53]. In a cross-sectional study of the long-term Epidemiology of Diabetes Interventions and Complications (EDIC) study, higher HbA
1c and nephropathy were independently associated with lower aBMD in older (59.2±6.7 years) participants with type 1 diabetes [
54]. In contrast, aBMD is relatively preserved and even elevated in type 2 diabetes and is related to increased body size [
38]. However, in both type 1 and type 2 diabetes, fracture risk is higher than predicted based on aBMD levels [
3]. Nevertheless, low aBMD remains a predictor for fractures in type 2 diabetes [
15].
Rather than deficits in BMD, diabetes may increase skeletal fragility through altered microarchitecture, including increased cortical porosity, and low bone turnover (Fig.
1). High-resolution peripheral quantitative computed tomography (HR-pQCT) provides in vivo assessment of volumetric BMD (vBMD) and trabecular/cortical compartments of the distal radius and tibia. As with most diabetes-related studies, individual studies are confounded by significant clinical heterogeneity in diverse cohorts. A recent meta-analysis found site-specific differences in bone structure between people with type 1 and type 2 diabetes and those without diabetes [
55]. Compared with control participants without diabetes, type 1 diabetes was associated with impaired trabecular parameters (vBMD, number, and heterogeneity) at the radius but not the tibia. Cortical parameters were preserved. In contrast, type 2 diabetes was associated with preserved trabecular features and enhanced cortical thickness but increased cortical porosity (particularly at the radius). Conversely, in a recently published cohort of 59 older individuals with long-standing type 1 diabetes (duration 37.7±9.0 years, age 59.9±9.9 years), type 1 diabetes was associated with poorer cortical measurements (thickness, vBMD) at the ultradistal tibia but not the radius [
56]. However, cortical changes (and decreased bone strength and stiffness) were dependent on the presence of diabetic neuropathy, suggesting that changes may have been driven by vascular complications. Similarly, in three studies of type 2 diabetes, cortical changes were not observed in all those with type 2 diabetes but only in those with previous fracture [
57], microvascular complications [
58] or clinically significant peripheral vascular disease [
59].
Together, HQ-pQCT data provide several insights. First, differences between the radius and the tibia suggest that mechanical load, and therefore obesity, may affect bone microarchitecture. Second, differences between type 1 and type 2 diabetes, and the association of changes in HQ-pQCT with vascular complications, underscore the complex interplay of metabolic factors, vascular complications and age, especially as the phenotype of older type 1 diabetes appears to resemble that of type 2 diabetes. Further studies in well-characterised individuals with type 1 and type 2 diabetes and examining the role of HR-pQCT parameters in fracture risk prediction are warranted.
Bone turnover is best assessed by tetracycline-labelled iliac bone biopsy, although the invasiveness of this technique limits its widespread use in clinical practice and research studies. Histomorphometry studies have shown older bone with reduced bone turnover and abnormal collagen structure in insulin-requiring women with type 2 diabetes [
60]. Changes were not associated with type 2 diabetes duration or HbA
1c levels. Type 2 diabetes has also been associated with stiffer and harder cortical indices and relatively preserved trabecular mechanical properties [
35].
Serum bone turnover markers (BTMs) can be used to non-invasively assess bone turnover clinically. Although there are some conflicting studies, meta-analyses suggest BTMs reflecting bone formation and resorption are reduced in both type 1 diabetes [
61] and type 2 diabetes [
62]. Separating out the metabolic contributors to lower BTMs has been challenging, with inconsistent associations with HbA
1c [
63], adiposity [
64] and microvascular complications [
65]. In our detailed cross-sectional analysis of the DOES cohort, type 2 diabetes was independently associated with lower BTMs (25–50% lower than in those without type 2 diabetes) [
38]. Insulin resistance, but not obesity or visceral adiposity, was also associated with lower BTMs, suggesting that hyperinsulinaemia may be a key pathophysiological contributor. However, the utility of BTMs in fracture prediction in type 2 diabetes remains unclear, as one case–control study found that BTMs were directly associated with fracture risk in participants without type 2 diabetes, but not in those with type 2 diabetes [
66], and prospective studies are required.
Given the limitations of conventionally derived aBMD using DXA in diabetes, other clinically available modalities such as trabecular bone score (TBS) and advanced hip analysis (AHA) are being investigated. The TBS indirectly measures lumbar spine trabecular microarchitecture by evaluating grey-level variations in pixels from a spine DXA image. The TBS is probably lower in individuals with type 1 diabetes than in those without type 1 diabetes [
67], although there may be no differences in younger people (aged 19–50 years) with type 1 diabetes without diabetic complications [
68]. Similarly, type 2 diabetes is associated with a lower TBS [
69] and this has been shown to partially explain the fracture risk in type 2 diabetes [
70]. The lower TBS in type 2 diabetes appears to be associated with BMI and fat mass [
71], and therefore abdominal adiposity rather than type 2 diabetes per se may drive the apparent discrepancy of poorer trabecular bone on TBS compared with the preserved trabecular parameters seen in the HR-pQCT data.
AHA uses hip DXA geometry and structural parameters to estimate hip strength. In one study, type 1 diabetes was associated with poorer cortical measurements and femoral neck instability, although the participants in this study also had end-stage kidney disease [
72]. Both type 2 diabetes and impaired glucose tolerance have been associated with worse strength parameters in some but not all studies, especially when adjusted for lean/total body mass [
73,
74]. AHA parameters in type 2 diabetes appear to be associated with BMI and body size, rather than type 2 diabetes, although visceral adiposity is inversely associated with some measures of skeletal load strength [
38]. Further characterisation of AHA changes across various dysglycaemic states would clarify its clinical utility.