Introduction

Diabetes mellitus is a common disease throughout the world [1]. Diabetes-related complications, such as cardiovascular and renal diseases, are becoming a huge health care and financial burden [1]. Osteoporosis is a common skeletal disorder characterized by reduced bone strength predisposing to an increased risk of fracture, which results in pain, impaired function, reduced quality of life, institutionalization, and death [2]. A number of previous studies have revealed the association between diabetes mellitus and osteoporosis, and it is now well established that both type 1 diabetes mellitus (T1DM) and type 2 diabetes mellitus (T2DM) are associated with an increased risk of osteoporotic fractures [3,4,5].

The mechanisms underlying increased bone fragility in diabetes mellitus are complex. Low bone turnover, accumulation of advanced glycation endproducts (AGEs), micro- and macro-architecture abnormalities, and tissue material damage lead to abnormal biomechanical properties may contribute to the impaired bone fragility [6]. Other factors associated with bone fragility in patients with diabetes mellitus include inflammation response, oxidative stress, adipokine alterations, WNT dysregulation, and increased marrow fat [6]. Bone extrinsic factors, such as increased frequency of falls, a factor that is closely related to diabetic complications and treatment-induced hypoglycemia, also contribute to the increased fracture risk [6,7,8]. However, after being adjusted for fall frequency, diabetes remains an independent risk factor of increased fracture risk [9,10,11]. Therefore, low bone strength is probably contributing to the increased bone fragility in diabetes. Both bone density and quality are important factors in the determination of bone strength. Multiple methods have been used to assess bone quantity and bone quality of diabetes mellitus patients, including bone mineral density (BMD), trabecular bone score (TBS), quantitative computed tomography (QCT), bone histomorphometry, microindentation, and bone turnover markers. This review will give a brief summary of the current development and clinical uses of these assessments.

We conducted a literature search for English language articles that was published before Mar 15th, 2018 or earlier in the PubMed and Embase online using the following keywords in various combinations: “diabetes”, “diabetes mellitus”, “type 1 diabetes mellitus”, “type 2 diabetes mellitus”, “bone”, “fracture”, “fracture risk”, “skeletal fragility”, “osteoporosis”, “bone mass”, “structure”, “microarchitecture”, “bone quality”, “strength”, “turnover”, “dual x-ray absorptiometry”, “DXA”, “bone mineral density”, “BMD”, “Fracture Risk Assessment Tool”, “FRAX”, “trabecular bone score”, “TBS”, “quantitative computed tomography”, “QCT”, “peripheral quantitative computed tomography”, “pQCT”, “cortical”, “trabecular”, “histomorphometry”, “microindentation”, “cyclic microindentation”, “impact microindentation”, “bone turn over”, “turnover”, “mechanism”, “advanced glycation end-products”, “AGEs”, “insulin”. Approximately 100 relevant articles were reviewed to discuss the current evidence for the use of BMD, FRAX, TBS, QCT, bone histomorphometry, and microindentation in assessment of bone quality in patients with diabetes mellitus. Only publications in English were included.

BMD and FRAX

BMD measured by dual x-ray absorptiometry (DXA) is currently the gold standard for both osteoporosis diagnosis and the monitoring of treatment efficacy.

Research indicates that BMD is lower in patients with T1DM [12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36]. A recent meta-analysis that includes 23 cross-sectional studies and 2 cohort studies shows a significant association between T1DM and decreased BMD values of total body with pooled mean differences of − 0.06 g/cm2 (95% CI − 0.11, − 0.01) [37].

However, patients with T2DM usually show BMD values above average [10, 33, 38,39,40,41,42,43,44,45,46,47,48,49,50]. A meta-analysis which is based on 15 observational studies (3437 T2DM patients and 19,139 control) shows that BMD in patients with T2DM is significantly higher with pooled mean differences of 0.04 g/cm2 (95% CI: 0.02, 0.05) at the femoral neck [51]. Therefore, T2DM may be associated with a reduction of bone quality, which cannot be reflected by BMD measurement.

Fracture risk assessment tool (FRAX) is widely used in estimating individualized 10-year probability of hip and major osteoporotic fracture. The FRAX algorithm consists of femoral neck BMD T-score, age, sex, weight, height, previous history of hip fracture, current smoking, recent use of corticosteroids, presence of rheumatoid arthritis, and ≥ 3 alcoholic beverages per day. FRAX is currently widely used for the estimation of fracture risk in patients with diabetes mellitus. In older patients with T2DM, both femoral neck BMD T-score and FRAX score are significantly related to hip and non-spine fracture risk [52].

Although BMD and FRAX are important in assessment of fracture risk, they are limited in application in patients with diabetes mellitus. As parameters related to diabetes mellitus are not included in FRAX, it may underestimate the fracture risk in specific population. In 2011, using data from three prospective observational studies, the Study of Osteoporotic Fractures, the Osteoporotic Fractures in Men study, and the Health, Aging and Body Composition Study, Schwartz et al. demonstrated that among older adults with T2DM, fracture risk was higher than those without diabetes even with similar BMD T-score and age or FRAX score [52]. For a given hip fracture risk, patients with T2DM have a 0.59 (95% CI, 0.31–0.87) higher T-score for women and 0.38 (95% CI, 0.09–0.66) for men [52]. Therefore, it is suggested that fragility fractures in diabetes mellitus may also be associated with diabetes-related changes in bone quality, which are not reflected in BMD. Although BMD and FRAX are very useful tools for the assessment of fracture risk in older adults, interpretation of T-score of FRAX score in patients with diabetes mellitus should be careful enough to take into account the higher fracture risk associated with diabetes [52].

TBS

TBS is a new texture parameter that analyzes pixel gray-level variations in the DXA image and reflects bone microarchitecture. As TBS is based on DXA, it can be widely accessible without introducing new equipment. A meta-analysis which includes individual-level data from 17,809 subjects in 14 prospective population-based cohorts shows that TBS can serve as a significant predictor of fracture risk independent of FRAX [53]. The gradient of risk (GR; hazard ratio per 1 SD change in risk variable in direction of increased risk) of TBS for major osteoporotic fracture is 1.44 (95% CI 1.35–1.53). When adjusted for FRAX 10-year probability of major osteoporotic fracture, TBS is still an independent predictor of fracture risk (GR = 1.32, 95% CI 1.24–1.41). Moreover, when the FRAX probability is adjusted for TBS, there is a slight increase in the GR (1.76, 95% 1.65–1.87 vs 1.70, 95% CI 1.60–1.81) [53]. The Manitoba BMD Cohort also demonstrated a small but significant improvement in overall net reclassification improvement (NRI) for all individual FRAX interventional criteria (range 0.007 to 0.018) and all three national clinical practice guidelines (CPGs) (range 0.008 to 0.011) [54]. Almost all of the improvement was found in individuals close to the intervention threshold [54].

Diabetes mellitus is significantly associated with lower TBS in unadjusted and adjusted models [55]. Moreover, TBS is negatively related to levels of HbA1c, fasting plasma glucose, and fasting insulin [56]. Recent studies show the potential of TBS in predicting the fracture risk in diabetes mellitus patients [53,54,55,56,57]. Studies that assess bone quality in DM with TBS are summarized in Table 1. In a recent cross-sectional study in which 119 T1DM patients (mean age 43.4 ± 8.9 years) and 68 gender, age, and BMI matched controls are included, TBS values are significantly lower in T1DM patients with prevalent fractures (1.309 ± 0.125 versus 1.370 ± 0.127, p = 0.04) [57]. In this set-up, a TBS cutoff < 1.42 discriminates the existence of fractures with a sensitivity of 91.7% and a specificity of 43.2% [57]. Another retrospective study which includes 169 postmenopausal women with T2DM evaluates the distinguishing performance of TBS, BMD, original and TBS-adjusted FRAX scores [63]. The result shows that TBS (p = 0.008) are significantly lower and TBS-adjusted FRAX score (OR = 2.0, 95% CI 1.1–2.7) is significantly higher in the group with vertebral fractures compared with the group without vertebral fractures [63]. Moreover, the AUCs of TBS-adjusted major FRAX for overall vertebral fracture risk stratification are significantly higher than that of major FRAX (0.717 versus 0.687; p = 0.043) [63]. There is a lack of clinical study directly comparing differences in TBS between T1DM and T2DM.

Table 1 Assessment of bone quality in diabetes mellitus with TBS

Multiple interventional studies have been done to investigate the response of TBS to osteoporosis treatment, though the study population is not limited to patients with DM (reviewed in [74, 75]). To summarize, current studies suggest that TBS tends to increase after treatment in parallel with the change of BMD, but the magnitude of increase of TBS is smaller than that of BMD [74, 75]. There is a lack of evidence in to what extent TBS increase after treatment predicts a reduction of fracture risk. The effect of diabetes treatment on TBS is still not fully known, though some studies demonstrate that HbA1c, fasting glucose, fasting insulin, and homeostasis model assessment for insulin resistance are negatively associated with TBS [56, 62, 66].

QCT

QCT is a noninvasive assessment of bone microarchitecture at the distal radius and tibia [76]. The high resolution of three-dimensional bone images generated by high-resolution peripheral quantitative computed tomography (HRpQCT) allows the measurement of volumetric BMD (vBMD) and other parameters. Mechanical properties of the bone can also be estimated from QCT images using microstructural finite element analysis (μFEA) [77].

QCT has been used in a number of studies to investigate the microarchitecture of bone in patients with diabetes mellitus [78,79,80,81,82]. Studies that assess bone quantity and quality in DM with QCT are summarized in Table 2. A cross-sectional study that includes 17 male T1DM patients aged from 18 to 49 years and 18 sex-matched healthy controls shows that T1DM patients have significant lower cortical vBMD in the femoral neck and significant lower vBMD, cortical thickness, and cortical area in the intertrochanter [78]. The bone strength estimated by pQCT was also lower in T1DM group [78]. Another cross-sectional study that includes 48 adolescents with T1DM also shows that T1DM is associated with reduced bone mineral content and small bone cross-sectional area and cortical strength [79]. In 2015, a cross-sectional study that includes 55 T1DM patients finds that diabetic microvascular disease is associated with deficits in cortical and trabecular vBMD and microarchitecture, while T1DM patients without diabetic microvascular disease show similar HR-pQCT parameters with controls [80]. Further investigation is needed to determine if this relationship between the presence of MVD and bone deficits is causal and impacts fracture risk, and whether diabetic bone disease is an extension of the spectrum of diabetic microvascular diseases.

Table 2 Assessment of bone quality in diabetes mellitus with QCT

QCT is also used to assess bone quality in patients with T2DM, but the study results are inconsistent among different studies. In the Framingham HR-pQCT Study which includes 1069 subjects (129 with T2DM, 940 without T2DM) with a mean age of 64 ± 8 years, researchers compare cortical and trabecular microarchitecture, bone density, bone area, and bone strength in T2DM and non-T2DM subjects [81]. After adjusting for age, sex, weight, and height, T2DM group has significantly lower cortical vBMD (p < 0.01), higher cortical porosity (p = 0.02), and smaller cross-sectional area (p = 0.04) at the tibia [81]. Moreover, lower cortical vBMD at the tibia and cortical thickness at the radius is seen in T2DM only among those with a prior fracture [81]. However, in the cross-sectional Gothenburg Study which includes 1053 women (99 with T2DM, 954 without T2DM) aged from 75 to 80 years, ultradistal tibial and radial trabecular volume fraction, distal cortical volumetric BMD, cortical area, and failure load are higher in patients with diabetes than in controls [82]. The different results between two studies may partly be due to the differences in characteristics of subjects and differences in study design [81]. It is also not fully known how well pQCT parameters could predict the fracture risk in T2DM patients.

Microindentation

Microindentation is a technique that can directly assess the mechanical characteristics of cortical bone in vivo. By inserting a probe assembly into a cortical bone’s surface at the anterior tibia and inducing microscopic fractures, microindentation measures bone mechanical strength at the tissue level. There are two approaches of microindentation, cyclic microindentation, and impact microindentation. In cyclic microindentation, relatively low forces (2–10 N) are applied over several seconds, and parameters such as first indentation distance (ID), total indentation distance (TID), indentation distance increase (IDI), creep ID, unloading slope (US) (ave of 3-last cycle), energy (ave of 3-last cycle) are output (reviewed in [104]). Impact microindentation is a single impulse indentation to a higher force within up to 0.25 ms (reviewed in [104]). Using the technique of impact microindentation, a ratio called bone material strength index (BMSi) can be derived, and a lower BMSi indicates lower fracture resistance [105]. The details of the two techniques are reviewed by Allen et al. [104].

Impact microindentation has been used in several clinical studies to investigate the bone quality in DM patients [11, 82, 106]. The Gothenburg Study finds that BMSi is lower in T2DM (74.6 ± 7.6 versus 78.2 ± 7.5, p < 0.01) [82]. In a series of cases and control study, 60 postmenopausal women including 30 patients with a T2DM history for more than 10 years and 30 age-matched non-diabetic controls received impact microindentation. Compared to controls, T2DM patients had significantly lower BMSi while their BMD were similar to controls [11]. This study also finds that the level of glycated hemoglobin over the past 10 years is negatively associated with BMSi (r = − 0.41; p = 0.026) [11]. In another cross-sectional study, BMSi is also found to be negatively associated with T2DM status [106]. Moreover, advanced glycation end product accumulation is negatively related to BMSi [106]. Though the invasive procedure restricts its wide use as a clinical tool, microindentation helps to shed light upon bone quality change in diabetes mellitus population.

Histomorphometry

Bone histomorphometric analysis of bone biopsies provides a direct approach to assess bone remodeling rates at tissue level [107]. Moreover, bone microarchitecture can also be evaluated using micro-computed tomography (micro-CT) [107]. A number of studies in rodent models have shown reduced bone turnover rate, worse microstructure, and lower strength in rodent models of T1DM and T2DM [108,109,110,111]. However, as bone biopsy is an invasive test, only a few clinical studies have investigated the bone quality of diabetes mellitus patients using bone histomorphometry. A cross-sectional case-controlled study which includes 23 T1DM subjects (8 males and 15 females) and 23 age- and sex-matched controls does not find deterioration in bone histomorphometric or micro-CT variables in those patients without manifesting complications of T1DM [107]. However, T1DM patients with fracture history may have defects in bone microarchitecture [107]. A histomorphometric study published in 1964 finds an increase of cortical area of ribs in T2DM [112]. Another more recent study obtains iliac crest bone samples from 26 patients (13 male and 13 female) with an average age of 67.42 ± 2.74 years [113]. The control group consists of 20 non-diabetic subjects (11 male and 9 female) with a mean age of 57.95 ± 3.96 years that have suffered sudden or violent death [113]. As a result, significant lower bone volume (p < 0.0001), osteoid volume (p < 0.005), osteoid thickness (p < 0.0001), cortical thickness (p < 0.05), and osteoblast surface (p < 0.05) volume are found in the diabetic group [113]. The discrepancy of different studies may come from the relative low subject volume, the disease course, and complications. More high-quality clinical studies are needed to determine the histologic changes of diabetic bone.

Bone turnover markers

In patients with DM, the levels of multiple bone turnover markers are altered. A meta-analysis that includes 22 studies shows that osteocalcin (OC) and C-terminal telopeptide of type 1 collagen (CTX-1) are significantly lower in people with DM, while alkaline phosphatase (ALP), N-telopeptide of type I collagen (NTX), and hydroxyproline (HYP) do not differ [114]. The meta-analysis also reports decreased OC levels in T1DM compared to controls [114]. Some studies indicate that bone turnover markers are lower in children T1DM or at the onset of the disease [25, 115], but do not differ between T1DM adults and controls [116].

Levels of parathyroid hormone (PTH) tend to be 20–50% lower in T2D subjects than in controls [117,118,119] (reviewed in [120]). Most studies support that bone turnover markers, such as CTX, OC, P1NP, TRAP, and NTX, are reduced in patients with T2DM [11, 118, 119, 121,122,123]. Moreover, low P1NP and OC levels and high CTX, NTX, and sclerostin levels seem to be associated with increased fracture risks in T2DM, while are associated with [118, 121, 124, 125]. Further studies are needed to uncover whether bone turnover markers could predict alterations in bone quality and fracture risk.

Discussion

As is discussed in the introduction, multiple mechanisms contribute to the increased fracture risk in patients with DM. Some of these mechanisms have adverse effects on bone metabolism thus lead to declined bone quantity and quality.

High glucose levels in DM can lead to the accumulation of AGE collagen crosslinks in bone, which leads to biomechanically fragile bone [6]. AGEs can also be specifically recognized by AGE receptors (RAGE), which express in multiple bone-derived cells [126]. The AGE-RAGE interaction induces activation of nuclear factor kB (NF-kB) in RAGE-expressing cells [126]. As a result, the production of proinflammatory cytokines and reactive oxygen species (ROS) are increased in these cells which activate osteoclastogenesis and suppress osteoblast differentiation [127, 128]. Pentosidine is one of the best studied AGEs. Pentosidine levels were significantly higher in DM patients with fractures than those without fractures [129,130,131]. In patients with T2DM, urinary pentosidine levels negatively associate with trabecular bone scores [129]. This indicates that AGEs may play a role in the deterioration in bone microstructure, and this microstructure change can be estimated by methods like TBS.

Although T1DM and T2DM share many mechanisms in inducing osteoporosis and osteoporotic fractures, these mechanisms may affect bone metabolism differently in two diseases, and the two diseases may each have its special mechanisms. As is reviewed, compared those with T1DM, patients with T2DM have higher BMD [132, 133]. Young patients with T1DM, even only a few years after the onset of diabetes mellitus, have a lower BMD score [134,135,136], though decreased BMD are more frequent in patients with longer disease courses [16, 34]. Decreased BMD observed in T1DM is found to be associated with the presence of microvascular complications such as diabetic nephropathy, retinopathy, neuropathy [133]. Also as is reviewed in the main text, the presence of microvascular disease is associated with deficits in bone microstructure. These results indicate that microvascular disease, which may lead to inadequate blood flow in bone tissue, adversely affects bone formation in young T1DM patients and causes the deterioration in both bone quantity and quality. BMI has been found to be significantly greater in patients with T2DM compared to T1DM. BMI is positively associated with BMD and negatively associated with fracture risk [137, 138]. Therefore, high BMI may partly explain the relatively high BMD and low fracture risk in T2DM patients compared with T1DM patients.

Recent studies indicate that T1DM and T2DM differ in levels of bone turnover markers. Recently, Starup-Linde et al. find that compared to T2DM, T1DM has lower levels of P1NP, osteocalcin (OC), and s-receptor activator of nuclear factor kappa beta ligand (RANKL) [139]. The differences between T1DM and T2DM in biomarkers may be interpreted with insulin resistance status in T2DM [139]. How these differences affect bone health has not been fully known.

Patients with T1DM suffer from low levels of insulin and IGF1 since the stage of onset, while T2DM is characterized by insulin resistance, though in the advanced stage of T2DM, there is a relative insulin deficiency. Low levels of insulin and IGF1 have an adverse effect on osteoblasts during growth and can result in low peak bone mass at an early age [124, 140,141,142]. Though insulin mainly plays an anabolic role in the bone formation in physiological concentration, the relationship between hyperinsulinaemia and insulin resistance and bone metabolism is complex, and current clinical studies show conflicting results [45, 143,144,145,146,147]. Moreover, studies also suggest that hyperinsulinaemia and insulin resistance may contribute to reduced bone turnover even in the absence of hyperglycaemia [148].

Conclusion

In summary, diabetes mellitus is closely associated with increased risk of fracture. Therefore, assessment of bone quantity and bone quality may benefit DM patients, especially those with other factors of fractures. Although BMD and FRAX are very useful clinical tools in assessing bone quantity, they may not accurately account for extra fracture risk in diabetes mellitus. Through advances in new technologies, physicians can better assess the bone quality of patients with diabetes mellitus. Apart from BMD and FRAX, recent clinical evidence demonstrates that incorporating TBS brought a moderate improvement to currently used FRAX tool. Therefore, TBS can serve as a useful method in assessing bone quality in patients with DM. As TBS is based on DXA, it can be widely accessible without introducing new equipment. HR-pQCT is also a useful noninvasive method to discriminate bone quality change in DM patients but its association with fracture risk is far from being illustrated. Although the invasive procedures restrict the clinical use of microindentation and bone histomorphometry, they are useful in the research of bone quality in diabetes mellitus. Bone turnover markers have been found to be altered in diabetes mellitus and may be associated with fractures, but more studies are needed. In addition, fracture risk of DM patients is also associated with complications of DM. Further large-scale prospective studies about risk factors of fractures in DM may be needed to bring a special tool for fracture risk assessment in DM patients and identify high-risk population.