Diabetes-Induced Osteoporosis: Dual Energy X-Ray Absorptiometry Bone Quality Is Better than Bone Quantity

Abstract

Diabetes mellitus (DM) and osteoporosis are among the most common non-communicable diseases worldwide. Beyond their considerable socio-economic burden, both conditions significantly impair quality of life and reduce life expectancy, representing major causes of disability. DM-induced osteoporosis has recently emerged as a notable and frequent complication. Patients with type 2 DM have a twofold increased risk of fragility fractures, while those with longstanding type 1 DM exhibit a fivefold higher risk of hip, vertebral, and non-vertebral fractures. Bone mineral density (BMD) assessed by Dual Energy X-ray Absorptiometry (DXA) often fails to predict fracture risk in this population, as bone mass tends to be normal, slightly reduced, or even elevated. However, DXA-derived indices can offer additional clinical value. The Trabecular Bone Score (TBS), which reflects bone microarchitecture, is frequently reduced in patients with DM and is associated with increased fracture risk, particularly in postmenopausal women. TBS is also linked to glycemic control and microvascular complications and can improve with bone-active medications, thus aiding follow-up assessments. Another useful DXA-based tool is the Bone Strain Index (BSI), which evaluates load resistance and has been shown to be degraded in diabetic patients, offering further predictive value for fractures. Additionally, Hip Structural Analysis (HSA) provides information on the mechanical integrity of the proximal femur, which may be compromised in DM. Based on the available evidence, this review aims to highlight the clinical utility of DXA-derived tools in DM-induced osteoporosis, emphasizing their ability to provide quantitative and qualitative information on bone health and to predict the risk of fragility fractures.

1. Introduction

The term “diabetes mellitus” (DM) refers to a heterogeneous group of metabolic disorders of carbohydrate metabolism, all characterized by chronic hyperglycemia resulting from impaired glucose utilization as an energy source and/or excessive glucose production due to unregulated gluconeogenesis and glycogenolysis [1]. Although all forms of DM share the common feature of chronic hyperglycemia, their underlying etiology differs. Type 2 DM (T2DM), which accounts for more than 90% of all cases, arises from a complex interplay of genetic and environmental (including social, familial and educational) factors that contributes to insulin resistance and, ultimately, impaired insulin secretion [2]. In contrast, type 1 DM (T1DM) is an autoimmune disorder characterized by pancreatic β-cell destruction, resulting in absolute insulin deficiency and a sudden onset of symptomatic hyperglycemia. Classic manifestations include polyuria and polydipsia, with diabetic ketoacidosis (DKA) occurring in approximately half of the cases at diagnosis. T1DM diagnosis is based on specific autoantibody detection (i.e., glutamic acid decarboxylase (GAD), islet tyrosine phosphatase 2 (IA-2) and/or zinc transporter 8 (ZnT8) autoantibodies) [3].

Nowadays, DM represents one of the most frequent non-communicable disorders worldwide, since one in nine suffers from DM, and International Diabetes Federation (IDF) projections estimate that more than 850 million people (+46% vs. 2024) will be living with DM in 2050 [4]. T2DM frequently goes undiagnosed for many years, because hyperglycemia develops gradually and, at earlier stages, is often not severe enough for the individual to notice the typical symptoms caused by hyperglycemia. Several drugs have been developed, significantly improving DM management and control (conventionally evaluated with serum glycated hemoglobin (HbA1c)), to prevent and reduce complications, which are responsible for relevant costs for National Health Systems (NHS) and private insurances (USD 1 in 10 paid is related to DM) [1,2,4].

Beyond elevated blood glucose levels, DM-related metabolic changes, induced by insulin resistance, included increased levels of cholesterol and triglycerides in the circulation, ectopic fat distribution (such as in muscles, heart, and pancreas) and hepatic steatosis or metabolically associated fatty liver disease (MAFLD) [5,6], predisposing to arterial hypertension and overweight/obesity. Together, these factors significantly increase the risk of serious microvascular damage, such as retinopathy, nephropathy and neuropathy [7]. Furthermore, people with DM are at higher risk of cardiovascular disease [8,9,10], including a 72%, 84%, 52% and 56% higher risk of heart attack, heart failure, stroke, and dementia, respectively [4]; that is why DM has become a leading cause of disability and mortality [11,12].

In addition to these conventional complications, diabetic bone disease or DM-induced osteoporosis, has recently emerged as a further comorbidity, strongly compromising patients’ quality of life [13]. Patients with T2DM have a doubled higher risk for any fragility fracture, while adult patients with a long history of T1DM have a fivefold increased risk for hip, vertebral and non-vertebral fractures [14,15,16,17].

Osteoporosis is a systemic skeletal disorder characterized by decreased bone mass and qualitative alterations (macro- and micro-architecture and bone material properties) associated with increased fracture risk. Primary osteoporosis occurs after menopause (postmenopausal osteoporosis) or with advancing age (senile osteoporosis), whereas secondary osteoporosis is caused by a number of disorders and drugs, including DM [18]. It is noteworthy that osteoporosis remains asymptomatic until fragility fractures occur, with a relevant socio-economic and health burden. In Western countries, one in three women [19] and one in five men [20] over 50 face osteoporosis: mortality within the first year following a hip fracture may rise to 25%, whereas disability affects up to 60% of all fractured subjects, with increasing numbers in the projections for the next decades [21]. Appallingly, patients with diabetic bone diseases have an even poorer quality of life and lower life expectancy [22]. Only 3 out of 10 subjects with an osteoporotic fracture who deserve specific anti-osteoporotic drugs are indeed treated to reduce their high/very-high fracture risk, while less than half of patients starting a medication will have an adequate (>80%) compliance after the first year of treatment [23,24].

In this review, we present a narrative overview of the tools available for assessing bone quality and fracture risk in DM-induced osteoporosis. Relevant literature was identified through comprehensive searches of PubMed and Scopus databases, using keywords: diabetes, osteoporosis, fractures, fracture prediction, bone mineral density (BMD), Trabecular Bone Score (TBS), Bone Strain Index (BSI), and Hip Structural Analysis (HSA). The search covered studies published from the early adoption of these indices (from 2008/2009 for TBS and 2018 for BSI) through March 2025. Original articles were selected based on clinical relevance, novelty and the specific focus on DXA-based bone assessment. Only articles in English were included. Conference abstracts and case reports were excluded. To this purpose,

Table 1. Summary of the main DXA-derived techniques used to assess bone quality and fracture risk in DM-induced osteoporosis. The table outlines the principle of each method, along with its clinical advantages and limitations.

Technique	Principle	Advantages	Limitations
DXA-BMD	Measurements of areal bone mineral density (g/cm2)	Reference standard for osteoporosis diagnosis Excellent precision and good accuracy	Affected by degenerative changes BMD is mainly a bone quantity measurement
TBS	Texture analysis of lumbar DXA to indirectly estimate trabecular microarchitecture	Surrogate of bone quality independent predictor of fracture risk less affected by degenerative changes	Applied only to lumbar spine May be influenced by soft tissue thickness
BSI	Finite Element Analysis applied to DXA images to estimate bone strength	Estimates local strain distribution Integrates loading mechanics Applied also at proximal femur	Less widely available Caucasian normative data only (Italian population)
HSA	Geometric analysis of the proximal femur	Assesses structural strength	Bidimensional Fracture risk prediction only from selected parameters

provides a comparative summary outlining the main advantages and disadvantages of each tool.

2. Pathophysiology of Diabetes-Induced Osteoporosis

From a pathophysiological point of view, different cellular and molecular mechanisms are involved in the impairment in the bone material properties. First of all, DM-induced osteoporosis is characterized by low bone turnover, as evidenced by several human studies in which both markers of bone formation, such as osteocalcin (a peptide produced by osteoblasts) and procollagen type I N-terminal propeptide (P1NP), as well as bone resorption markers like serum C-terminal telopeptide of type I collagen (CTX) and urinary N-terminal telopeptide of type I collagen (NTX), were detected as reduced and negatively correlated with HbA1c levels [25,26]. This evidence has been documented also by bone biopsies and histomorphometry, where a decreased number of osteoblasts and reduced osteoids with poor mineralization and, thus, impaired trabecular and intra- and endocortical bone formation were reported [27,28].

Advanced glycation end products (AGEs) are increased in patients with DM as a result of hyperglycemia and increased oxidative stress, playing a crucial role in bone health derangement in this specific context. Activation of AGEs receptor promotes inflammation, producing reactive oxygen species (ROS) and stimulating the secretion of inflammatory cytokines (i.e., tumor necrosis factor-α (TNFα) and interleukin-6 (IL-6)) that activate osteoclastogenesis and inhibit osteoblast differentiation [29,30]. Moreover, dysregulation in adipokine secretion has been reported: adiponectin, which has an anabolic effect stimulating osteoblasts and suppressing osteoclast activities, is reduced in patients with DM as compared to healthy controls [31].

Accumulation of AGEs in bone is also associated with poor biomechanical and bone material properties, as demonstrated by an increase in serum and urinary pentosidine (the most studied among AGEs) in patients with T2DM and fragility fractures [32]. Moreover, AGEs may affect bone formation by inhibiting synthesis of type 1 collagen and osteocalcin, osteoblast differentiation, function and attachment to the collagen matrix [27]. Intriguingly, both hyperglycemic acidosis and ROS may promote mesenchymal stem cell differentiation towards adipocytes [33].

Acute hyperglycemia and consequent hyperosmolarity suppress expression of osteocalcin and uptake of calcium by osteoblasts in vitro [34], while in patients with DM they may enhance bone resorption [35] and osmotic diuresis induced by glycosuria and consequent hypercalciuria may determine a negative calcium balance [36].

Interestingly, osteocyte alterations have also been suggested in patients with DM. Sclerostin, a molecule known for its inhibitory activity on bone formation through the Wnt–β-catenin pathway, was found significantly higher; it correlated with thinner cortical thickness and was associated with vertebral fractures in postmenopausal women with T2DM as compared to healthy controls [37].

It is well known that insulin is fundamental for insulin-like growth factor-I (IGF-I) secretion, and, therefore, insulin deficiency, as it is in patients with T1DM, may be associated with functionally low IGF-I levels. IGF-I plays a key role in longitudinal growth, achievement of adequate peak bone mass and bone remodeling [38]; therefore, in patients with DM low IGF-I may contribute to additional impairment on bone homeostasis. Indeed, insulin directly, and through IGF-I, activates the IGF-I receptor, promoting osteoblast differentiation and activity. It is noteworthy that AGEs and chronic hyperglycemia may blunt IGF-I-mediated anabolic actions as well as determine osteoblast insensitiveness to IGF-I [39,40]. Moreover, incretins and, particularly, glucagon-like peptide 1 (GLP-1), whose receptor is expressed on bone marrow stromal cell surface and immature osteoblasts [41], exert a crucial role not only in stimulating insulin secretion but also mesenchymal stem cell proliferation and differentiation towards osteoblasts [42].

In addition, patients with DM have a high risk for falls, caused by multiple, specific factors with a complex interplay, that include high glycemic variability with hyper and hypoglycemic events (in particular those under insulin treatment), prolonged disease duration, blood pressure variations with orthostatic hypotension, hypovitaminosis D, impaired balance, low ability in physical performance, functional hypogonadism in men, visual impairment due to diabetic retinopathy, microalbuminuria and impaired renal function, peripheral neuropathy, micro and macrovascular peripheral disease with foot amputation, autoimmune malabsorption (such as celiac disease) in T1DM, sarcopenia and obesity in T2DM and gut microbiota alterations [27,43,44,45]. It is worth mentioning that since osteoporosis is generally considered more prevalent in females, and particularly in postmenopausal women, most of the reports have a sex selection bias, albeit the annual trend of publication has kept on growing since the first study on bone quality in DM published in 2013 [46].

3. Limitations of Conventional Assessment of Bone Health in Patients with Diabetes Mellitus

Bone can be regarded as a complex system, characterized by specific structural and geometrical features that enable it to fulfill its primary function of mechanical support by resisting compressive, torsional, and flexural forces. From an engineering perspective, numerous skeletal factors must be considered to explain bone strength, and their analysis is essential for improving our ability to predict structural failure [47].

In any structure subjected to external loading, the magnitude and distribution of internal stress are influenced by the loading conditions, the system geometry, and the material properties. To prevent permanent damage or rupture, stress and strain levels must remain below the material yield point. Bone adheres to the same mechanical principles; its resistance is determined by factors such as density, geometry, internal trabecular architecture, and cortical thickness, all of which can be evaluated using radiological imaging. These parameters can be measured using volumetric imaging (e.g., computed tomography) or planar imaging techniques, with traditional radiography and Dual Energy X-ray Absorptiometry (DXA) being the most commonly used. X-ray images can be analyzed through various methods [47,48,49], from traditional beam models typically applied to long bones [47,50,51] to more sophisticated approaches like the finite element model (FEM) [52].

While low BMD explains approximately 70% of fragility fractures observed in clinical settings, there is considerable overlap in BMD values between individuals with and without osteoporotic fractures [53,54]. This suggests that additional factors contribute to bone strength, such as texture, geometry, and the material ability to deform under load;properties that influence the mechanical behavior of all materials, including bone [55]. These aspects bear relevance in those diseases where BMD is not significantly reduced, like secondary osteoporosis [56,57] and, between these, DM [27,46].

Studies performed enrolling patients with T1DM show an average lower areal BMD as compared to age-matched healthy subjects. This is, at least in part, explained as a result of an impairment in peak bone mass acquisition [58]. Consistently, in adults with child-onset T1DM and in teenagers, lower trabecular bone density and number of trabeculae together with trabecular augmented space and heterogeneous distribution have been reported when bone microstructure was evaluated with micro-Magnetic Resonance Imaging (microMRI) [59] and High-Resolution peripheral Quantitative Computed Tomography (HR-pQCT) [60,61]. Albeit no significant cortical alterations were described in patients with non-complicated T1DM as compared to healthy controls, cortical porosity both at the tibia [62] and distal radius [63] was increased in patients with diabetic neuropathy and correlated positively with the severity of neuropathy and negatively with nerve conduction amplitude and velocity.

On the other side, patients with T2DM have normal or even exaggerated BMD [27], probably due to increased mechanical loading in overweight/obese patients and higher estrogen levels due to the stimulated action of aromatase enzymes [15], which are highly expressed in the fat tissue and convert male sex steroid hormones into female ones, finally determining low bone turnover [64]. Different analyses highlighted that fracture risk assessment tool (FRAX^®), a tool that helps clinicians in predicting fracture risk in patients with primary osteoporosis, relevantly (~30%) underestimates the risk for major and hip fragility fractures in patients with T2DM [65]. Subsequent observation tried to overcome this pitfall, proposing different corrections (e.g., reducing 0.5 SD T-score levels or using rheumatoid arthritis as a proxy for T2DM), with better, but not always satisfactory, results [66].

Recent times are now characterized by aging populations, which have the right to aspire to reach healthy aging. Both DM and osteoporosis have become important disorders to be addressed from a multidisciplinary perspective; therefore, prevention and assessment of bone fragility in DM are urgent issues, and measuring BMD represents an insufficient and inadequate tool in this specific setting. The aim of this paper is to narratively review the available literature on DXA bone quality in patients with DM, highlighting its clinical role in predicting fragility fractures, beyond bone mass-derived estimation.

4. Trabecular Bone Score (TBS) and Diabetes Mellitus

TBS, an indirect DXA bone texture index used since 2008, is a DXA-derived bone quality index that can explain fracture events in patients with a normal or slightly reduced BMD, like those affected by DM [67], where bone quality derangement is more relevant than bone quantity loss. DM is an in vivo model of this particular condition where TBS, a textural indicator of good bone microarchitecture, is a better fracture predictor than BMD [68]. TBS is an essential tool to investigate bone quality status, but it is inferred only from the lumbar spine scan and does not provide data about bone quality status at the hip.

TBS calculation is based on the fact that DXA image areas with soft gray variations are typical of a dense trabecular structure. Conversely, big dark areas are characteristic of low connectivity, low trabecular number and wide space between trabeculae [69]. The calculation of TBS is based on the same mathematical matrix DXA source used for BMD measurement, but it represents a different feature of bone status and can discriminate between patients with similar BMD but different trabecular microarchitecture [70]. Figure 1 shows an example of a comparison between a DXA-BMD study and the corresponding TBS analysis.

TBS can discriminate fractured patients and can predict fracture partially independently from BMD [70]. More recently, a study enrolling almost 74,000 subjects has confirmed that BMD and TBS are complementary in their ability to predict fracturative risk, independently from the other clinical risk factors that are included in the FRAX^® algorithm [71]. The use of TBS to adjust FRAX^® performance has been recently introduced also in daily clinical practice [72], and the magnitude of this adjustment (~5% for major fractures’ 10-year risk and ~1% for hip fractures) is substantial enough to reclassify treatment in many subjects who are close to the treatment threshold [73], and, intriguingly, patients with T2DM are more prone to experience a reclassification of fracture risk [74].

In the Manitoba Registry, one of the largest cohorts ever published (almost 30,000 postmenopausal women) was investigated with regard to bone health. Among them, almost 10% were affected by DM, and this subgroup (compared to normoglycemic subjects) had higher lumbar and femoral BMD adjusted for different covariates (including age, BMI, glucocorticoid use, prior major fracture, rheumatoid arthritis, chronic obstructive pulmonary disease, smoking habit, alcohol abuse and osteoporosis treatment), but significantly reduced TBS, which was the only parameter associated with incident fracture risk both in diabetic and non-diabetic women [75]. Relevantly, BMD strongly correlates with body mass index (BMI), whereas reduced TBS was associated with higher fracture risk. BMI and the serum marker of bone formation P1NP were negative predictors of TBS, while the serum marker of bone resorption CTX was a positive one [76].

An Italian cross-sectional case-control study confirmed significantly lower TBS levels with a three-fold higher risk for fragility fractures in postmenopausal women with DM [77]. TBS alone or TBS-adjusted FRAX^® were predictors of fractures after different corrections, such as age, BMI, BMD and glucocorticoid use [77,78]. A stimulating result comes from another cross-sectional study showing that both trabecular (represented by low TBS) and cortical bone (which is predominant in the forearm DXA scan) parameters were significantly reduced in postmenopausal women with DM as compared to non-diabetic ones, with no significant differences at the hip or spine BMD [79].

This evidence was later extended to men of European ancestry with DM, who exhibited lower TBS [80] and, at the same time, a tendency toward higher BMD compared to non-diabetic men [81]. However, this trend was not statistically confirmed in a recent American study, likely due to the limited male sample size (78 men vs. 433 women) [82].

Asian females with DM do have lower TBS, albeit superimposable or even higher BMD, as compared to controls. Nevertheless, data on Asian males are more contradictory, also taking into account the age differences among enrolled cohorts [79,83], as Vietnamese [84] and Taiwanese [85] populations have normal TBS, whereas Korean [86] and Chinese [87] men with DM exhibited lower TBS. In 2019, these inconclusive results were therefore checked in a meta-analysis, without reaching definitive conclusions [88]. Nonetheless, it is worth mentioning that all reported TBS differences, even when statistically significant, were particularly modest as compared to non-diabetic subjects, and the number of enrolled men was extremely lower than women, hence limiting the sample power. This is also the explanation for two very small trials going against this evidence, where women with DM had similar TBS levels to women with normoglycemia [89,90].

As for many other DM complications, it is well known that fracture risk exponentially increases with poor disease control (as evaluated with HbA1c) and hypoglycemic episodes [91,92]; therefore, one could argue whether inadequate glucose-lowering treatment may have detrimental effects on TBS, too. Indeed, a non-uniform spectrum of results identified a possible connection between higher TBS values and appropriate glycemic control [93,94]; however, few studies were specifically built to address this issue [95]. Moreover, it is still a matter of debate what the exact glycemic target is to achieve the most beneficial results on bone health [96].

A cross-sectional observation in postmenopausal women evidenced that higher HbA1c levels and longer T2DM duration were independently associated with increased femoral bone mass and reduced lumbar spine TBS [97]. Similarly, although there was no statistically significant BMD difference in the whole cohort, the mean TBS value was significantly lower in the T2DM group in comparison to those with pre-diabetes or normoglycemic individuals, and both HbA1c and DM duration demonstrated a significant negative correlation with TBS [98].

While Dhaliwar and colleagues accredited a correlation between lower TBS, high BMD and inadequate DM control (defined by HbA1c > 7.5%) [93], Ballato and colleagues stratified men with T2DM based on glycemic control (good vs. poor control according to HbA1c level ≤7% or >7%, respectively) and compared them to non-diabetic subjects, but due to the modest sample size (169 diabetic men), no significant differences were detailed in terms of BMD and TBS [99]. Oppositely, TBS showed a negative correlation with HbA1c, fasting glycemia, fasting plasmatic insulin levels and insulin resistance in Korean women younger than 65 years old and in men [86]. Better glucose control was associated with higher TBS levels, and a weak but significant negative correlation between TBS and several indices of insulin resistance, including HOMA-IR, was subsequently noted in a cross-sectional report of 105 postmenopausal women [100].

Finally, in a recent, small, Brazilian study, TBS was lower in the T2DM group as compared to age-, sex-, and BMI-matched controls. HbA1c was negatively associated with TBS and body fat percentage with TBS and total hip BMD. Femoral BMD was positively associated with intramuscular lipids, whereas a trend of negative association was observed between intramuscular lipids and TBS [101]. Serum HbA1c and waist circumference were independently associated with TBS in a cross-section of 64 postmenopausal women with T2DM as compared to 175 postmenopausal women with a normal glycemic profile [102].

Since a relationship between glycemic control and TBS has been highlighted by several studies, the reader could argue if there is a correlation between the classical diabetic vascular complication and bone quality. Actually, a significant association between lower spine TBS levels and renal function (as estimated glomerular filtration rate (eGFR)) has been described [103]. Moreover, in a specific subset of frail patients under hemodialysis, DM was highly prevalent (more than 40% of enrolled patients), and TBS was compromised in older subjects and in those who previously suffered from cardiovascular events [104]. In a new Spanish cross-sectional study, no difference in both spine and hip BMD levels was observed, whereas TBS has been indicated as significantly lower in 92 patients with T2DM and microvascular disease vs. 269 patients with non-complicated DM, after multiple corrections, including disease duration and HbA1c. Mean TBS was partially degraded in both groups, and the reported difference between complicated vs. non-complicated DM was really modest (1.235 vs. 1.287, respectively). Nonetheless, patients with diabetic microvascular disease and TBS < 1.23 had significantly higher HbA1c [105]. Finally, lower TBS is not just a possible marker of microvascular disease, but it was also correlated with a higher mortality [100].

In this setting, the role of lipids has been investigated also by Dule and colleagues in a cohort of diabetic women. Authors identified a positive correlation between TBS, high-density lipoprotein (HDL) cholesterol, and circulating levels of the steroid hormone calcifediol; remarkably, plasmatic HDL levels were the most important predictor of TBS impairment, irrespective of patients’ age, menopausal status, BMI, waist circumference, physical activity, and statin exposure [106].

Previous studies on overweight and obese individuals have shown a significant linear decline in TBS as the number of metabolic syndrome components increases in both men and women [107]. However, TBS becomes less reliable in individuals with a BMI greater than 37 Kg/m² [108], as excessive abdominal soft tissue can lead to an underestimation of raw TBS values by obscuring the region of interest [109]. Furthermore, different fat distribution (android vs. gynoid) may exert a clinically relevant role on bone health, as verified in a study of postmenopausal women with reduced TBS levels and lower gynoid fat mass. Authors observed that 44.8% of enrolled females with a long-standing T2DM had pathological TBS (i.e., ≤1.31) and those with reduced TBS had more frequently at least one prevalent fragility fracture (32.6% vs. 11.3%) and lower BMI [110].

In this context, central adiposity has been suggested as a potential confounding factor affecting TBS. This was highlighted by Palomo and colleagues, who initially found an association between lower TBS and higher HbA1c in unadjusted models, as well as in models adjusted for age, lumbar BMD, and BMI. However, when TBS was corrected for soft tissue thickness, the association was no longer statistically significant [97]. Inverse correlations between TBS and metabolic syndrome, waist circumference [106] or central adiposity as expressed with lower gynoid and increased android fat mass [110] have been invariably confirmed. In a Brazilian study, TBS was significantly lower in patients with DM as compared to healthy subjects; a low score correlates with sarcopenia as measured with both hand-grip and gait-stand tests, and whole body fat mass was strongly associated with it [111].

This supposed confounding effect on TBS has been recently hypothesized to contribute to the discrepancies between TBS and HR-pQCT [112]. Nevertheless, it must be underscored that TBS (and other DXA-derived parameters, as discussed below) is actually the unique parameter that undoubtedly is associated with fragility fractures in diabetic-induced osteopathy.

Once a bone-active drug is prescribed, improvements in TBS values are usually modest and less evident than BMD changes, and the role of different medications on bone health is still a matter of debate in the diabetic scenario. In addition, it is unclear whether the role of DXA improvement in BMD and/or derived bone quality parameters could be interpreted as a marker of medication response. So far, no specific therapeutic indications have arisen from expert groups or National/International Academic Societies, and the management of DM-induced osteoporosis is superimposable on those with low BMD and normoglycemia. Indeed, very few studies aim at investigating the role of bisphosphonates in T2DM: in a prospective, interventional trial, 150 mg of intravenous ibandronate was administered monthly for 12 months in osteoporotic patients with or without DM. In both groups a similar increase in BMD and TBS was detected with similar modification in both formation and resorption bone turnover markers (specifically, P1NP and CTX), without any significant change in terms of HbA1c or fasting glycemia [90]. In an Indian trial with diabetic patients and controls receiving a single infusion of 4 mg intravenous zoledronic acid, the gain in lumbar BMD was significantly lower in the T2DM group as compared to the non-diabetic one, and a comparable change in bone turnover markers and TBS in both groups was discovered after one year of follow-up [113].

Recently, in postmenopausal women with osteoporosis and T2DM, 12 months of romosozumab significantly improved lumbar BMD and TBS independently of abdominal fat, and to a greater extent than alendronate. The greater gain with romosozumab was maintained after transition to alendronate and persisted significantly at months 24 and 36 as compared to alendronate alone. Percentage changes in TBS corrected for abdominal fat thickness were weakly correlated to lumbar BMD percentage changes from baseline to month 36 [114]. Regarding the possible role of anti-diabetic drugs, in an Italian longitudinal study, one-year treatment with subcutaneous dulaglutide or semaglutide resulted in a significant reduction in weight and BMI and a significant increase in bone turnover markers and adiponectin, while myostatin values showed a modest but significant reduction. Lumbar BMD significantly declined, whereas TBS values showed a marginal increase, demonstrating a preservation in bone quality despite bone mass reduction [115].

Fewer data are actually available regarding TBS in T1DM settings, where cohorts are eventually less ample. Notwithstanding the well-demonstrated low BMD, mean TBS has been reported to be similar [116] or lower (although within the normal ranges) [117] as compared to healthy subjects, and, when deranged, it negatively correlated with components of metabolic syndrome and insulin resistance, but not with disease control or waist circumference. Notably, most of the patients affected by T1DM are not obese; therefore, the fat tissue thickness is not so relevant in this peculiar subgroup. In early-onset T1DM, TBS was superimposable on the general population and correlated with both radial and tibial trabecular mass [63]. Considering these inconsistent results, two different meta-analyses have been performed, confirming lower TBS as compared to healthy subjects [118], but in one of them, this reduction was not large enough to justify such a higher fracture risk in patients with T1DM [119].

5. Hip Geometry and Diabetes Mellitus

HSA is a bone geometry parameter that performs as an index of bone derangement in primary and secondary osteoporosis [120,121,122]. It is inferred from a DXA hip scan and provides a geometric description of the functional mechanics of three femoral areas of interest (narrow neck, intertrochanteric and femur shaft): the cross-sectional area (CSA), the cross-sectional moment of inertia (CSMI), the section modulus (Z) and the buckling ratio (BR) [123].

CSA is an index of bone resistance to axially directed loads, whereas CSMI reflects the flexural strength of structural rigidity and can be calculated by multiplying the area in the cross-section by the square of its distance from the centroid. BR is calculated as the ratio between the maximum distance between the center of mass and the outer cortex and the average cortical thickness. BR provides a stability index of the cortex under compressive loads, bending included [122]. Figure 2 shows an explanation of how HSA allows the evaluation of femoral bone geometry and strength parameters from DXA scans.

Research has shown that HSA can predict the occurrence of hip fractures [121,124]. However, its application in clinical practice remains limited due to challenges in interpreting structural parameters and a lack of sufficient evidence from real-world clinical settings [125]. As a result, current scientific guidelines do not recommend the routine use of HSA for assessing hip fracture risk [125]. DXA images can also be used to automatically extract two additional geometric parameters: the neck-shaft angle (NSA) and hip axis length (HAL). Several studies have identified a positive association between increased HAL and a higher risk of hip fracture, suggesting that this parameter may be an important predictor independent of BMD [125]. The NSA, defined as the angle between the femoral neck and shaft, has been investigated as a potential independent predictor of hip fracture risk. While some studies support this association, findings become less consistent when analyses are adjusted for BMD. As such, current evidence remains insufficient to support the routine clinical use of NSA, particularly since its predictive value may not surpass that of BMD alone [125].

A few studies have investigated the role of HSA in evaluating bone geometry and mechanical strength in patients with DM, with contrasting results. In the case-control study by Naseri and colleagues conducted on 348 postmenopausal women with T2DM, most HSA parameters were not significantly different from those of non-diabetic controls after adjusting for age and BMI [79]. However, diabetic women exhibited significantly higher values in certain femoral shaft indices, including CSA, CSMI, Z, and subperiosteal diameter, suggesting potential compensatory adaptations in bone geometry [79].

Another study by Krishnan and colleagues on adolescents and young adults with T1DM found no significant differences in HSA parameters when compared to matched controls. Nevertheless, a negative correlation between DM duration and CSMI was observed, with authors suggesting that structural deterioration may begin early in the disease course, even before measurable changes in BMD [126]. This evidence points out the potential role of HSA in detecting subclinical alterations in bone strength in younger diabetic populations.

A third study by Ballato and colleagues evaluated postmenopausal women with T2DM and, similarly, found no significant differences in femoral neck geometry compared to non-diabetic controls after adjusting for BMI [99]. Still, diabetic women showed slightly increased values in CSMI and Z, consistent with findings from Naseri and colleagues. These results further support the hypothesis of structural compensation at the proximal femur level and emphasize the need for a multifactorial approach to fracture risk assessment in DM.

Taken together, these studies highlight the possible added value of HSA in understanding skeletal health in both T1DM and T2DM. It remains crucial to have a proper interpretation of geometric data in light of the HSA technical limitations, such as the 2D nature of DXA and the lack of lean mass adjustment.

6. Bone Strain Index (BSI) and Diabetes Mellitus

The DXA-based tools discussed so far (TBS and HSA) are valuable and commonly used techniques that offer indirect assessments of vertebral microarchitecture and femoral geometry, respectively. However, when used alongside BMD, these tools still provide an incomplete picture of bone mechanical resistance. Specifically, they lack detailed information on bone deformation and fatigue behavior, which are critical from a structural engineering perspective. One of the most comprehensive methods for evaluating bone mechanics is FEM. FEM enables the simulation of stress and strain distributions within a structure made of specific materials under defined loading conditions. Originally developed for engineering applications, this technique has been effectively adapted for medical use, including fracture risk prediction [127] and prosthetic implants simulation [128]. Recently, an innovative DXA-derived index called BSI has been conceived and developed. It is calculated using Finite Element Analysis (FEA) applied to the greyscale distribution of bone density measured on both spine and femoral scans [55,129,130]. BSI calculation incorporates information on density distribution, bone geometry and resistance to loadings on local areas. Unlike BMD and TBS, which primarily quantify bone mass and its spatial distribution, BSI also accounts for the shape of the anatomical site under investigation and the specific loading derived from the patient’s body weight. As such, BSI represents a novel approach in bone health assessment, offering valuable information to better understand bone quality derangement in metabolic bone diseases, particularly in secondary osteoporosis [55,131]. The FEM involves breaking down a complex object into smaller, simpler elements to which the principles of classical mechanics can be applied. When forces and constraints are applied to specific regions of the bone, they generate internal stresses and strains. These depend on the type and magnitude of the load, the bone geometry, and the stiffness of each individual element. Although numerous FEMs have been developed to evaluate bone condition and predict fracture risk, none have been adopted in routine clinical practice. This is mainly due to the lack of fully automated software and the absence of integration into standard clinical reporting workflows. In order to make FEM clinically relevant, its application must extend to the femoral and lumbar regions, which are typically assessed using DXA. In recent years, many studies have concentrated on FEM analysis of the proximal femur to estimate bone strength and improve hip fracture risk assessment [49,52]. However, only a limited number of studies have focused on the lumbar spine, showing that FEM provides superior predictive performance for vertebral strength compared to BMD measured by DXA [132]. The FEM analysis is automatically performed by applying forces and constraints to a triangular mesh generated from bone segmentation within the DXA software. For the lumbar site, each vertebra is loaded on the upper surface and constrained to the lower, following the model proposed by Colombo and colleagues [130]. Material properties of each triangular element in the model are assigned based on the experimental relationships described by Morgan and colleagues at the lumbar region [133]. The force applied to the upper plate of the vertebra is calculated using the patient-specific loading model described in the study by Han and colleagues [134]. In the femoral region, the BSI algorithm simulates a lateral fall scenario, with constraints applied to both the femoral head and the lower part of the shaft and with a subject-specific impact force (proportional to the individual’s body weight) applied to the greater trochanter [134]. Since the BSI value reflects the bone ability to withstand applied loads, it serves as an indicator of mechanical strength. As explained in the introduction, mechanical resistance to fracture should consider different variables: stiffness, texture, geometry, deformation capability and fatigue. See Figure 3 for an example of BSI figure.

A recent case-control study evaluated BSI performance in a cohort of 153 postmenopausal women with T2DM, with the aim of assessing its ability to discriminate patients with prevalent fragility fractures [135]. Among the enrolled subjects, 22 out of 153 (14.4%) presented with at least one major fragility fracture. The study revealed significantly higher BSI values at both lumbar spine and proximal femur sites in fractured patients compared to non-fractured ones. Other DXA-derived parameters showed similar discriminative power; nevertheless, lumbar BSI exhibited the greatest relative difference between the two groups and was the third best-performing parameter in ROC analysis (AUC = 0.733), following TBS (AUC = 0.763) and femoral neck BMD (AUC = 0.756) [135]. Correlation analyses revealed only weak to moderate inverse associations between BSI and BMD, highlighting the complementary nature of BSI with respect to traditional densitometric indices. In multivariate analysis, classification models based on BSI alone achieved higher specificity (0.70) compared to those based on BMD (0.60) or TBS (0.64), and the highest overall performance was obtained when BSI was combined with both BMD and TBS (accuracy = 0.69, sensitivity = 0.69, specificity = 0.70) [135]. These findings underscore the potential role of BSI as a complementary metric for fracture risk assessment in T2DM, particularly in clinical contexts where BMD may fail to reflect underlying skeletal deterioration.

7. Other Diagnostic Technologies to Assess Bone Status in Patients with Diabetes Mellitus

Although DXA remains the gold standard to assess bone mass in primary osteoporosis, different tools have been developed to better estimate bone quality and microarchitecture in order to provide a more comprehensive evaluation of bone health. Albeit the role of these technologies is beyond the purpose of this narrative review, several data are now available regarding pQCT. HR-pQCT, with its scans either at the tibia or radius, has demonstrated its ability in predicting fragility and vertebral fractures in clinical research settings [136]; however, its clinical application, also considering high costs and ionized radiation exposure, strongly limited its application in routine clinical practice. In patients with T2DM, notwithstanding the differences among enrolled cohorts in terms of sex, age, disease duration and control, anti-diabetic drugs and presence of other DM-related complications [27], all the reports agree in describing a preserved trabecular volume density (BV/TV), an increase in the number and thickness of trabeculae as well, and lower trabecular separation as compared to subjects without T2DM, with no difference among those patients harboring fragility fractures [137] and/or microvascular disease [138]. In addition, an improved quality of trabecular bone microarchitecture has been confirmed in a recent meta-analysis, where a more homogeneous distribution of trabeculae has been spotted [139]. A possible explanation of this high fracture rate in patients with T2DM could derive from the analyses of cortical bone, which is thicker and more porous, particularly in fractured subjects or in those with microvascular complications [140]. Since cortical porosity increases with a superimposable rate in patients with T2DM and independently from prevalent fractures, it has been hypothesized that increased pore diameter and density could be the result of microvascular damage also in bone [141]. In contrast, very few data are available regarding patients with T1DM. No difference was observed between patients without microvascular complications and healthy controls, whereas those with microvascular complications had larger total and trabecular bone areas, lower total, trabecular and cortical volumetric BMD, thinner cortex at the radius, and lower total and trabecular BMD at the tibia as compared to patients with normal glycemia and microvascular comorbidities [60]. Furthermore, patients with microvascular disease also exhibited lower total and trabecular BMD, trabecular thickness, estimated bone strength, and greater trabecular separation and network inhomogeneity at both sites as compared to diabetic patients with no microvascular complications [60]. More recently, radiological bone parameters from HR-pQCT were used to estimate local trabecular bone formation and resorption: at the distal radius, they were, respectively, 47% and 59% lower in T1DM participants compared with healthy controls. Interestingly, bone formation correlated positively with nerve conduction amplitude and negatively with HbA1c [62,142].

A new generation of non-invasive ultrasound technology, also known as radiofrequency echographic multispectrometry (REMS), has been validated. With its low cost, easy-to-use device, no radiation exposure, and quick scans, it may represent a valid alternative, providing information on both bone quality and bone mass, which strongly correlate with BMD assessed by DXA, estimating all together the patient’s fracture risk [143]. Ninety postmenopausal women with T2DM were recently screened and compared to non-diabetic healthy controls, reporting a significantly higher BMD when measured by DXA. In contrast, REMS parameters were significantly lower. A significantly higher prevalence of osteoporosis was observed with REMS than DXA (47% vs. 28%, respectively), and diabetic women with previous fragility fractures had a significantly lower lumbar BMD assessed by REMS, as compared to non-fractured ones, with no difference observed with DXA [144]. More recently, as previously discussed, the same group reported a non-significant variation in lumbar BMD assessed by REMS, a marginally increase in TBS and a significant reduction in DXA BMD at the spine, as well as a statistically significant reduction in both DXA and REMS femoral mass in both women and men with T2DM after one year of injectable, weekly dulaglutide or semaglutide therapy [115]. Albeit exploratory, these data seem more promising than those previously reported with other ultrasound techniques performed at the heel or hand phalanges, which failed in predicting fracture risk, as they provide superimposable results with those obtained with DXA [145]. That is why REMS technology seems to succeed in overcoming the effects of structural internal artifacts and accurately estimating bone fragility not only in primary [146], but also in the specific context of DM-induced osteoporosis.

8. Conclusions

Osteoporosis has been increasingly recognized as a frequent complication in DM, contributing to reduced quality of life and decreased life expectancy in affected patients. Patients with T1DM or T2DM have fivefold and twofold higher risks for fragility fractures, respectively. In this specific context, BMD is often found to be normal or even elevated, making it insufficient for predicting fracture risk. In routine clinical practice, clinicians should carefully assess fracture risk, as bone quality is often more severely compromised than bone mass due to the characteristic low bone turnover in these patients. In this regard, DXA-derived software tools may provide valuable additional information. Findings on TBS have consistently demonstrated its reliability in predicting fracture risk, especially in postmenopausal women. Moreover, low TBS values have been associated with poor glycemic control and the concomitant presence of other diabetic microvascular complications, such as neuropathy or nephropathy. Furthermore, DXA may provide parameters related to hip geometry and load resistance, which are both deranged in patients with DM and associated with fragility fractures. Among them, BSI is an innovative DXA-derived parameter that integrates bone geometry and load resistance. In T2DM patients, BSI has shown promising results in identifying fragility fractures and may complement established indices such as BMD and TBS, improving fracture risk assessment in clinical practice. Therefore, given the substantial evidence supporting an increased fracture risk, the assessment of bone quality should be considered essential in patients with either T1DM or T2DM. Nevertheless, it remains too often overlooked in clinical practice. The bone specialist should be involved to define the most appropriate diagnostic work-up and therapeutic strategies, considering both the potential benefits of bone-active drugs [147] and their real risk, as in the case of romosozumab [148]. In addition, the effects of antidiabetic medications on bone health—either beneficial [149] or detrimental [150]—should also be carefully considered.