Aging Clinical and Experimental Research (2019) 31:1375–1389 https://doi.org/10.1007/s40520-019-01294-4 REVIEW Radiofrequency echographic multi‑spectrometry for the in‑vivo assessment of bone strength: state of the art—outcomes of an expert consensus meeting organized by the European Society for Clinical and Economic Aspects of Osteoporosis, Osteoarthritis and Musculoskeletal Diseases (ESCEO) Adolfo Diez‑Perez1 · Maria Luisa Brandi2,3 · Nasser Al‑Daghri4 · Jaime C. Branco5 · Olivier Bruyère6 · Loredana Cavalli2,3 · Cyrus Cooper7 · Bernard Cortet8 · Bess Dawson‑Hughes9 · Hans Peter Dimai10 · Stefano Gonnelli11 · Peyman Hadji12 · Philippe Halbout13 · Jean‑Marc Kaufman14 · Andreas Kurth15,16 · Medea Locquet17 · Stefania Maggi18 · Radmila Matijevic19,20 · Jean‑Yves Reginster4,6 · René Rizzoli21 · Thomas Thierry22,23 Received: 17 June 2019 / Accepted: 24 July 2019 / Published online: 17 August 2019 © The Author(s) 2019 Abstract Purpose The purpose of this paper was to review the available approaches for bone strength assessment, osteoporosis diag- nosis and fracture risk prediction, and to provide insights into radiofrequency echographic multi spectrometry (REMS), a non-ionizing axial skeleton technique. Methods A working group convened by the European Society for Clinical and Economic Aspects of Osteoporosis and Osteoarthritis met to review the current image-based methods for bone strength assessment and fracture risk estimation, and to discuss the clinical perspectives of REMS. Results Areal bone mineral density (BMD) measured by dual-energy X-ray absorptiometry (DXA) is the consolidated indicator for osteoporosis diagnosis and fracture risk assessment. A more reliable fracture risk estimation would actually require an improved assessment of bone strength, integrating also bone quality information. Several different approaches have been proposed, including additional DXA-based parameters, quantitative computed tomography, and quantitative ultrasound. Although each of them showed a somewhat improved clinical performance, none satisfied all the requirements for a widespread routine employment, which was typically hindered by unclear clinical usefulness, radiation doses, limited accessibility, or inapplicability to spine and hip, therefore leaving several clinical needs still unmet. REMS is a clinically available technology for osteoporosis diagnosis and fracture risk assessment through the estimation of BMD on the axial skeleton reference sites. Its automatic processing of unfiltered ultrasound signals provides accurate BMD values in view of fracture risk assessment. Conclusions New approaches for improved bone strength and fracture risk estimations are needed for a better management of osteoporotic patients. In this context, REMS represents a valuable approach for osteoporosis diagnosis and fracture risk prediction. Keywords Bone strength assessment · Osteoporosis diagnosis · Fracture risk · REMS · Ultrasound · Lumbar spine · Femoral neck Nasser Al-Daghri, Jaime C. Branco, Olivier Bruyère, Loredana Cavalli, Cyrus Cooper, Bernard Cortet, Bess Dawson-Hughes, Introduction Hans Peter Dimai, Stefano Gonnelli, Peyman Hadji, Philippe Halbout, Jean-Marc Kaufman, Andreas Kurth, Medea Locquet, Osteoporosis is a systemic skeletal disease characterised by Stefania Maggi, Radmila Matijevic, Jean-Yves Reginster, René the reduction in bone mass and the degeneration of bone Rizzoli, Thomas Thierry have contributed equally to this work. structure that leads to an increased risk of fracture [1], with Extended author information available on the last page of the article important social, medical and community care costs. 13 Vol.:(0123456789) 1376 Aging Clinical and Experimental Research (2019) 31:1375–1389 Bone strength, which is a measure of the resistance to These considerations argue for the introduction in clinical bone fracture [2], is determined by a composite summation routine of further tools in order to better define and estimate of numerous skeletal characteristics that can be divided bone strength and to predict the risk of fracture. into four basic components: composition, microarchitec- To address this, a working group convened by the Euro- ture, size, and shape (Fig. 1). Several factors contribute pean Society of Clinical and Economic Aspects of Oste- to the definition of each skeletal characteristic, such as oporosis and Osteoarthritis (ESCEO) met to review the cellular density, mineralization, collagen crosslinking at data on the currently available imaging techniques for the a nanoscale level for the definition of bone composition; in vivo bone strength assessment and to discuss principles, trabecular and cortical properties including porosity, thick- applications and perspectives of a new technology called ness, connectivity at microscale level for the definition of Radiofrequency Echographic Multi-Spectrometry (REMS) bone microarchitecture; age, genetics, gender, and habits on the basis of the latest available data. Specific attention for the definition of bone size and shape at a macroscale has been paid to identify the clinical needs related to bone level. Understanding how these variables interact and con- strength assessment, fracture risk prediction and osteoporo- tribute to bone strength is critical in the development of tic patient management that are not completely satisfied by fracture prediction tools [3, 4]. the techniques currently employed in clinical routine and to Available imaging techniques are able to capture a their possible resolution by the introduction of innovative measure of the aforementioned factors and could also approaches. provide information about bone strength. Therefore, in principle, several tools are necessary to assess the differ- ent bone properties, such as spectroscopy and microscopy Clinical assessment of bone strength at nano-scale level, high-resolution quantitative computed tomography (HR-QCT) or µCT at micro-scale level, dual- Several approaches have been developed to provide an esti- energy X-ray absorptiometry (DXA) or CT at macro-scale mation of bone strength, including imaging devices, risk level [3, 4]. Nowadays, in clinical practice, bone strength fracture calculators, bone biopsy techniques, and laboratory is indirectly estimated by bone mineral density (BMD) tests (Table 1). In this review, we will focus on the imaging measured by DXA, which currently represents a well- devices: the main features of the most common approaches recognized technology for the diagnosis of osteoporosis are described and discussed in the following paragraphs. worldwide. Nonetheless, several studies demonstrated that BMD only accounts for about 50–70% of the bone strength Dual‑energy X‑ray absorptiometry variation, being able to describe features related to the quantity of bone tissue but overlooking information about The most widely used technique to obtain information about bone quality [4–6]. bone strength is DXA, currently considered as the gold Fig. 1  Factors contributing to bone strength. The ultimate definition of bone strength is complex, but four main categories of bone characteristics that contribute to bone strength can be outlined at different scale level: size, shape, architecture and composition [3] 13 Aging Clinical and Experimental Research (2019) 31:1375–1389 1377 Table 1  Main currently available tools for the in vivo assessment of predict fracture risk independently of current BMD, found bone strength that routine clinical DXA measurement in untreated women Category Tool cannot accurately characterise the rate of BMD loss due to the dominating effect of measurement errors [15]. Imaging devices Dual X-ray absorptiometry (DXA) BMD alone for the assessment of fracture risk has a high Quantitative CT (QCT) specificity but low sensitivity, with many fragility frac- High resolution peripheral QCT (HR- tures occurring in patients who did not have a diagnosis for pQCT) osteoporosis as defined by the WHO classification based on Magnetic resonance imaging (MRI) T-score value, but whose BMD lay in the osteopenic range Quantitative ultrasound (QUS) [16–18]. Moreover, in the diabetes paradigm, BMD is unable Radiofrequency echographic multi spec- trometry (REMS) to explain the increased risk of fracture: in particular, for Risk fracture calculators FRAX® Type 1 diabetes, the highly increased risk of hip fracture was Qfracture® only partially explained by the observed BMD reduction, Garvan® whereas in Type 2 diabetes the increased risk of fracture is DVO risk ­calculator® not captured by the paradoxically higher BMD [19]. Bone biopsy techniques Static histomorphometry Thus, a screening program for osteoporosis based on Dynamic histomorphometry DXA-measured BMD alone cannot be recommended for Laboratory tests Bone turnover markers (BTMs) the whole population [20, 21], and further studies evaluat- Single nucleotide polymorphisms (SNPs) ing the most cost-effective screening strategy are warranted Micro-ribonucleic acid (miRNA) [22]. These findings also suggest the hypothesis that multi- Others Microindentation ple factors, other than BMD, contribute to fracture risk and highlight the importance of BMD-independent determinants of bone strength and fracture risk assessment. standard densitometric technique for the diagnosis of osteo- DXA scanners, if equipped with dedicated software porosis. Through the acquisition of two-dimensional (2D) modules, can also measure additional parameters related to X-ray scans with an effective dose in the range of 1–13 µSv textural features and hip geometry, as described in the sub- [7–9], it estimates an areal BMD expressed in g/cm2 (i.e., sequent paragraphs. the ratio between bone mineral content estimated through X-ray absorption and 2D projection of the scanned bone Trabecular bone score area), which is a surrogate measure of bone strength [3]. Ex vivo studies performed on isolated human bones showed The Trabecular Bone Score (TBS), calculated by the soft- that areal BMD accounts for about two thirds of the variation ware package TBS ­iNsight® (Medimaps Group SA, Geneva, of bone strength as evaluated by shear test of the femoral Switzerland), is a unit-less indirect index of trabecular neck or compression of the vertebrae [5, 10]. microarchitecture based on pixel grey level variations in the In clinical practice, DXA is employed to categorise the DXA images [23, 24]. The large clinical database of the patient by comparing the measured BMD value with the Province of Manitoba (Canada), including 29,407 women average BMD of a reference healthy population of young aged over 50 years at baseline, has been used for several individuals, which allows to define the patient’s T-score investigations aimed at assessing the ability of TBS in frac- as the number of standard deviations (SDs); the measured ture risk prediction. Although TBS (along with BMD and BMD differs from the reference value. Thus, patients are additional clinical risk factors) helps to identify individuals classified as healthy (normal BMD, T-score ≥ − 1), osteo- at high risk of fracture and to guide initiation of osteoporosis penic (− 2.5 < T-score < − 1), osteoporotic (T-score ≤ − 2.5) treatment [25], change in lumbar spine TBS is not a useful or severely osteoporotic (T-score ≤ − 2.5 in the presence of indicator of fracture risk regardless of osteoporosis treatment one or more fragility fractures), according to the definitions [26]. Another retrospective study on the Manitoba database given by the World Health Organization (WHO) [11, 12]. showed the ability of lumbar spine TBS to predict osteoporo- Moreover, DXA is employed to provide prognostic informa- tic fractures in patients with type 2 diabetes, whereas BMD tion on probability of fracture and to monitor bone status was paradoxically increased in these patients [27]. in longitudinal studies [13]. A meta-analysis of prospective TBS values can be also input into the Fracture Risk cohort studies correlating the baseline BMD with the subse- Assessment tool F ­ RAX® [28]. This tool, available online quent follow-up for fractures showed that the measurement at www.shef.ac.uk/FRAX [29], aims to predict osteoporo- at the spine and at the hip seemed to have a better predictive tic fractures on the basis of several risk factors optionally ability for spine and hip fractures, respectively [14]. Les- including femoral neck BMD or spinal TBS [17]. A recent lie et al. investigating whether the rate of BMD loss would meta-analysis showed that TBS, adjusted for time since 13 1378 Aging Clinical and Experimental Research (2019) 31:1375–1389 baseline measurement and age, has a significant correlation with major osteoporotic fractures, which is slightly improved when ­FRAX® probability is taken into account, supporting the use of TBS both as a stand-alone fracture risk prediction tool and as a valuable add-on to ­FRAX® [30]. Nonetheless, some limitations of TBS should be also mentioned: increased noise in DXA images results in an artefactually reduced TBS [31]; TBS, but not BMD, is dependent on the used scan mode (standard versus thick), at least for GE Lunar Prodigy scanners [32]; increases in soft-tissue thickness overlying the spine lead to diminished TBS [33]; TBS reliability in men depends on BMI [34, 35]. Furthermore, a debate is still open in the scientific lit- erature as regards the proportion in which TBS explains the variability in vertebral strength. The very few available bio- mechanical studies [36, 37], which anyway represent only an indirect approach to what happens in vivo, found poor or no direct correlation between TBS and vertebral failure load, although reporting in some cases significant correla- tions with other elastic properties (e.g., r = 0.64 with bone stiffness [36]). Fig. 2  Representation of the Hip Axis Length (HAL) definition, i.e. Overall, lumbar spine TBS is an evolving software that the distance from the base of the greater trochanter to the inner pelvic has been shown to be useful as a complementary frac- brim (segment a–b). Angle c is the neck shaft angle, i.e. the angle ture risk prediction tool. However, the magnitude of TBS between the derived axes of the femoral neck and shaft increase in osteoporosis treatment is smaller than that of BMD. The relationship between change in TBS and frac- ture risk reduction remains to be elucidated with recent and region [43], HAL and neck shaft angle [44] might have a ongoing studies that are helping to refine its clinical utility role in fracture risk assessment for specific categories of [28, 38]. patients. Hip axis length and hip structural analysis Quantitative computed tomography The hip axis length (HAL), i.e. the distance from the base Quantitative computed tomography (QCT) allows to obtain a of the greater trochanter to the inner pelvic rim (Fig. 2), is a tomographic 3D reconstruction of the scanned bone and pro- further non-BMD parameter that has been considered for the vides a volumetric bone mineral density (vBMD), expressed assessment of hip bone strength through DXA [39]. A study in g/cm3, which is less susceptible to osteo-arthritis artefacts by Leslie et al. [40] showed that the risk of hip fracture, than the areal BMD measured by DXA. Depending on the adjusted for age and femoral neck BMD, increases by 3.6% specific adopted protocol, the exposure dose for a lumbar in men (p = 0.022) and by 4.6% in women (p < 0.001) for QCT scan is in the range of 50–500 µSv, which is clearly every millimetre of increase in HAL. However, as pointed lower than a standard abdominal CT scan (approximatively out by the International Society of Clinical Densitometry 8 mSv) [7, 8]. In contrast to DXA, this analysis allows sepa- (ISCD) at the 2015 Position Development Conference, rate estimation of trabecular and cortical BMD. Since tra- the association between HAL and hip fracture is actually becular bone has a substantially higher metabolic turnover ambiguous, with 14 published studies showing a positive than cortical bone, QCT is more sensitive to changes in association between HAL and hip fracture, 11 studies show- BMD than DXA [45]. The American College of Radiology ing no association and 1 study reporting a negative associa- (ACR) has recently introduced practice guidelines for the tion [41]. On this basis, the Hip Structural Analysis (HSA) use of QCT, with the definition of equivalent WHO clas- software was developed to simultaneously evaluate both hip sification of osteoporosis classes based on spine trabecular geometrical information and corresponding mineral mass volumetric BMD measurements and an association between data derived from DXA images [42]. Whether HSA could volumetric BMD and risk of fracture has also been estab- be actually useful for clinical assessment of bone fragility is lished. Moreover, the correlation between BMD measured currently under investigation, with few studies showing that by QCT and derived by DXA has also been assessed. Thus, some factors, i.e. cortical thickness of the intertrochanteric QCT might be used in case of non-availability of DXA. It 13 Aging Clinical and Experimental Research (2019) 31:1375–1389 1379 has also some important advantages over DXA that are use- microarchitecture, thus performing the so-called virtual ful in specific clinical indications, i.e. abnormal body size biopsy. A micro-FEA for peripheral bones might be thus (small/large body frame) or very obese patients, advanced derived from HR-pQCT scans. A cross-section of the distal degenerative spine disease (e.g. idiopathic skeletal hyperos- radius or tibia is scanned at a resolution sufficiently high to tosis, degenerative disc disease, facet arthropathy), and need resolve the trabecular microstructure (in the order of 80 μm). for accurate monitoring of trabecular metabolic activity [46]. Based on these images, micro-FE models are generated and In addition, QCT allows the development of finite-ele- a mechanical test is simulated to derive the stiffness of the ment analysis (FEA), a computational method in which scanned region and to estimate the strength of the whole each voxel represents a finite element with assigned material bone. In validation studies, micro-FEA showed its ability properties (e.g., elastic modulus). Load simulations might to predict bone failure load better than any density-based be performed with the estimation of the load at which struc- parameter, but this improved strength prediction did not tural failure occurs, so identifying a biomechanical fracture result in a better separation of subjects with and without threshold [47]. Indeed, the prospective 5-year AGES-Reykja- distal radius fractures in retrospective studies, probably due vik case–control study of 1110 patients reported significant to the reduced reproducibility, and consequently reduced correlation between bone strength estimated by FEA and specificity and sensitivity, in population retrospective stud- fracture occurrence, with age-adjusted odds ratios for low ies with respect to validation studies performed on cadav- vertebral strength and incident spine fractures of 2.8 (95% ers [57]. Currently, several HR-pQCT devices are available CI 1.8–4.3) and 2.2 (2.2, 95% CI 1.5–3.2), and age-adjusted worldwide [58, 59] and their use is mainly reserved for odds ratios for low femoral strength and incident hip frac- research purposes [60, 61], although there is potential to tures of 4.2 (95% CI 2.6–6.9) and 3.5 (95% CI 2.3–5.3) for use this technique in the clinical diagnosis and management women and men, respectively [48]. The FEA performed on of osteoporosis irrespectiwhether its ability in fracture pre- QCT images also provides information on femoral strength diction will be fully demonstrated [59, 61]. changes associated with therapies, as in the “PaTH” study evaluating the effects of PTH(1–84) and/or alendronate in Magnetic resonance imaging osteoporotic women [49]. It was shown that strength was mainly influenced by trabecular density changes, regardless In magnetic resonance imaging (MRI), and high-resolution of the pharmacological treatment, although changes in the MRI as well, the hydrogen atoms in water molecules are cortical bone density and overall bone geometry also con- exploited to obtain a contrast between marrow signal (hyper- tributed to femoral strength changes, in that case depending intense) and trabecular structure (hypo-intense). With this on the drug used. Due to the limited clinical availability of non-ionizing imaging modality, structural trabecular param- QCT and to the associated high costs and radiation doses eters (i.e., trabecular number, thickness, connectivity, anisot- [50], in the past years, several studies investigated the pos- ropy and shape) might be obtained and FEA can be applied sibility to obtain finite-element models from DXA imaging, similarly to the case of QCT [62]. However, some important but their clinical feasibility remains very limited and their limitations such as high costs, long scan time, motion arte- actual usefulness should be further explored [51–53]. facts and partial volume effects (in case of insufficient image Peripheral QCT (pQCT) has a more widespread avail- resolution that results in reduced contrast between adjacent ability and examines appendicular skeleton sites with a tissues and blurring effects) make this imaging technique lower effective radiation dose than central QCT (lower than scarcely employed and investigated in osteoporosis diagno- 0.01 mSv [8]). Through the simultaneous calculation of sis [63]. Furthermore, to the best of our knowledge, no sig- BMD, bone and muscle geometrical parameters, and bio- nificant longitudinal studies have been conducted to assess mechanical parameters, it provides an evaluation of the fracture risk prediction through MRI. “muscle-bone unit”, with the potential of representing a valid functional approach for bone health assessment [54]. Quantitative ultrasound However, with respect to DXA, pQCT has anyway a more limited accessibility, being also sensitive to motion artefacts. Quantitative ultrasound (QUS) devices measure bone prop- erties using differential reflections and attenuation of pulsed High‑resolution peripheral QCT ultrasound waves. According to the ISCD Official Positions, the only validated skeletal site for the clinical use of QUS in High-resolution peripheral QCT (HR-pQCT), with a sin- osteoporosis management is the heel, which, in conjunction gle scan exposure dose of about 3–4 µSv [55, 56], allows with clinical risk factors, can be used in patients aged over one to obtain volumetric images of distal bones, i.e. radius 65 in order to identify low-risk populations [64, 65]. The and tibia, with a simultaneous measure of cortical and most common parameters of interest for this non-ionizing trabecular vBMD and a deeper representation of bone technique are those derived from the combination of the 13 1380 Aging Clinical and Experimental Research (2019) 31:1375–1389 speed of sound (SOS) and the broadband ultrasound attenu- In a study by Kopperdahl et al. [48], derived from the ation (BUA), like the stiffness index (SI) and the quantitative AGES-Reykjavik case-control cohort, the large majority of ultrasound index (QUI). SOS, which relies on the accurate fracture events seems to occur when femoral neck T-score measurement of the time employed by a sound wave to pass of areal BMD was below about − 1.5. However, the abil- through the heel, is directly proportional to the BMD. Anal- ity of this parameter to predict hip fractures significantly ogously, BUA, which measures the reduction in intensity at improved only if combined with femoral strength estima- different frequencies of a broadband ultrasound pulse sent tion for a net reclassification improvement of 33%. Bone through the bone, shows a greater attenuation of the higher strength estimated from QCT-derived FEA was signifi- frequencies in strong bones than in weak bones [66]. The cantly correlated with fracture risk, when adjusted for age SI, for instance, was empirically derived as the sum of the and sex, independently from baseline BMD, indicating that normalized BUA and SOS values, with a comparable con- a reduced bone strength might be a predictive factor for tribution from the two factors [67]. high risk of osteoporotic fracture. Population studies have shown the capabilities of heel The predictive ability over a long period of areal BMD QUS to predict osteoporotic fractures, especially for hip, was evaluated in the Study of Osteoporotic Fractures with similar sensitivity but lower specificity than DXA (SOF) [74], which assessed BMD and risk factors in in discriminating fractured from non-fractured subjects almost 8000 women between 1988 and 1990, with a clini- [68–71]. A meta-analysis on 14 prospective studies showed cal follow-up for fractures lasting for 25 and 20 years for that the estimated relative fracture risk ranged between hip and any non-vertebral fracture, respectively. Although 1.23 and 1.94 for each SD decrease in QUS measurements BMD alone might be predictive of fracture risk, the over- depending on the measured QUS parameter and the type all prediction would be improved by taking into account of fracture, suggesting that heel QUS might be used as an other risk factors and other bone quality descriptors more alternative to DXA in the assessment of non-vertebral frac- closely related to bone strength. ture risk [72]. However, it has been also reported that QUS The sensitivity of areal BMD to treatment changes was predictive value wanes with time, being for instance more assessed in studies evaluating drug treatments, such as reliable at 1 year after baseline measurement than at 5 years the FREEDOM study (Denosumab versus placebo) [75] [73]. and the HORIZON study (Zoledronic versus placebo) The main advantages of QUS with respect to DXA are [76], showing that an increase in total hip BMD after lower cost, smaller space required, portability and absence 3 years of treatment was correlated with a decrease in of ionizing radiation. However, QUS approaches are appli- risk of non-vertebral fracture and, interestingly, the risk cable only to peripheral skeletal sites (calcaneus, tibia, pha- of fracture showed similar relationships (slopes) with per langes, radius), and their diagnostic performances are typi- cent change in total hip BMD for both drug and placebo cally inferior to the corresponding DXA ones. Moreover, the groups (Fig. 3, [75]). Moreover, a meta-regression of 38 availability of several QUS devices that differ substantially placebo-controlled trials with follow-up ranging from 1 to from each other in terms of measured parameters has limited 8 years showed that increments of 2–6% in total hip BMD their acceptance. Furthermore, QUS results are also depend- were significantly associated with a 28–66% reduction in ent on operator, anatomical location, and relative position- vertebral fractures, as well as with a 16–40% reduction ing of bone and ultrasound transducer: these drawbacks cur- in hip fractures, but not with reductions in non-vertebral rently limit the employment of QUS as a clinical diagnostic fractures [77]. In general, in the osteoporosis treatment, tool, except for screening purposes [47, 64]. the time-interval of repeated DXA-BMD measurements must be long enough to discern real BMD increments from precision error of repeated measurements (in the order of 1–2%) [78], with reasonable interval not less than 2 years. Role of bone strength in the assessment In summary, areal BMD estimated by DXA imaging of fracture risk: state of the art showed reasonable diagnostic performances being dis- criminative, predictive and sensitive to treatment changes. Areal BMD However, it is widely accepted that a deeper analysis of bone biomechanics is needed to increase the accuracy of Modelling bone strength might have implications for the fracture risk prediction by taking into account the actual understanding of osteoporosis mechanism, the assessment bone strength [3, 17, 79, 80]. of fracture risk and the prediction, and monitoring of drug treatments. As expressed above, the main clinically available indicator to predict fracture risk is areal BMD estimated by DXA imaging. 13 Aging Clinical and Experimental Research (2019) 31:1375–1389 1381 [83], including 8 large cohorts of patients for a total of 7254 individuals, investigated the association between indices derived from HR-pQCT and fracture risk. During an average follow-up of 4.6 years, 86% of incident frac- tures were registered in patients with femoral neck BMD in the osteopenic or normal range, but in whom deficits in bone microstructure were found with HR-pQCT. The results indicate that trabecular and cortical bone density and microstructure, as well as estimated failure load meas- ured at the peripheral skeleton, predicted incident fractures independently of femoral neck BMD and ­FRAX®. Moreover, a large multimodal investigation was per- formed by Biver et al. [60] in order to assess the contribu- tion of bone phenotype parameters to the prediction of low-trauma fractures in PMW: peripheral trabecular and cortical volumetric BMD and microstructure, as well as the estimated bone strength by HR-pQCT, showed high prediction rate independently of femoral neck BMD, ­FRAX®, and TBS. As BMD assessed by DXA at the ultra- Fig. 3  Association between incidence of non-vertebral fracture distal radius [84] showed very good performance for frac- and total hip BMD percent change from baseline at 36 months in Denosumab and placebo cohorts. The risk of non-vertebral fracture ture prediction, the authors suggested that measuring ultra- decreased with increasing per cent change in total hip BMD with distal radius BMD in addition to femoral neck BMD could similar relationships (slopes) for both treatment groups. The density be useful to refine fracture prediction for PMW in this curves at the bottom represent the distributions of total hip BMD age range, especially when bone microstructure cannot be change at 36 months for each treatment group [75] specifically assessed due to the limited clinical availabil- ity of HR-QCT [57–59]. Similarly, a cross-sectional case- High‑resolution peripheral QCT (HR‑pQCT) control study found that bone microstructure and strength assessment were able to discriminate patients on long-term The main clinical issue associated with the use of BMD glucocorticoids with vertebral fracture independently of alone is the twilight zone of osteopenic patients, where the BMD: noteworthy, with a reduction in total volumetric majority of fractures occurs, suggesting that the standard BMD and cortical thickness at the distal tibia, it was able classification may overlook some important biomechani- to identify patients at high risk of vertebral fractures in the cal characteristics that could improve the assessment of subgroup considered to have low fracture risk as assessed fracture risk [81]. For instance, the discriminative value by DXA or ­FRAX® [85]. of parameters derived from HR-pQCT at the distal radius The main difficulty in the full assessment of bone strength (i.e., total and trabecular density and heterogeneity of tra- is due to the numerous bone components involved in its defi- becular network) was assessed in a setting of osteopenic nition. The introduction of advanced tools in clinical prac- postmenopausal women (PMW), with and without frac- tice is needed in order to collect additional information for tures, in which BMD of the lumbar spine and femoral neck the modelling of bone strength and fracture risk. did not significantly differ [82]. Statistically significant dif- ferences were found between fractured and non-fractured subjects, with the former having − 10% in total density, FEA derived from QCT − 12.3% in trabecular density, and +25.6% in heterogene- ity of trabecular network [82]. This paper, on the one hand, Several attempts have been performed by FEA derived from underlined the importance of the description of trabecular QCT [86], sometimes including bone anisotropy informa- microarchitecture of peripheral sites for the enhancement tion [87], in order to model bone strength and to assess a of the prediction of fracture risk in osteopenic patients, personalized biomechanical fracture threshold as the ratio but, on the other hand, given the preliminary nature of of external forces applied to the bone and the FE-computed the study and the small number of enrolled patients, it bone strength [81]. Micro-FEA derived from HR-pQCT also emphasized the need of additional studies to better was also used in the prospective STRAMBO study, which understand the actual usefulness of the mentioned bone enrolled 825 men that were followed for 8 years, showing microarchitecture parameters. A recent study by the Bone that the distal radius trabecular number was the parameter Microarchitecture International Consortium (BoMIC) most strongly associated with fracture risk [88]. 13 1382 Aging Clinical and Experimental Research (2019) 31:1375–1389 Microindentation [95], the identification of the most cost-effective imaging technique through thorough economic analyses (including The recently developed in vivo impact microindentation different fracture states, country-specific aspects, etc.) is still may have a role in the explanation of fracture propensity debated in literature [21, 96] and the definition of a specific in case of poor bone quality associated with normal BMD, cost-effectiveness model is beyond the scope of this paper. such as type 2 diabetes, atypical femoral fractures, stress Anyway, there is a series of clinical needs that are not fractures, glucocorticoid treatment and HIV infection [89, effectively satisfied because of the previously mentioned 90]. For instance, in a study by Rozental et al. [91], lower limitations of the currently employed techniques and bone strength index measured by impact microindentation at would require the introduction of reliable methods for bone the anterior surface of the mid-tibia diaphysis in PMW was strength evaluation and fracture risk estimation. The most associated with distal radius and hip fractures and, if con- urgent of these unmet needs include the following: osteopo- firmed by prospective trials, might be integrated in models rosis diagnosis in patient categories for which X-ray exami- for the identification of women at risk of fragility fractures. nations are not feasible (e.g., paediatric subjects, pregnant and breastfeeding women, etc.); under-diagnosis and late DXA‑based 3D modelling diagnosis; effective bone quality assessment integrated with a reliable fracture risk prediction available in clinical rou- The most promising future approach could be the integra- tine; short-term follow-up of patients under treatment, since tion of the available technologies, such as DXA-based 3D the present techniques require at least 1 year between two modelling, as shown in a retrospective open-label study aim- measurements. ing to assess the effects of treatments on average changes in volumetric BMD and trabecular architecture over 24 months using DXA-based 3D modelling [92]. Nevertheless, the Radiofrequency echographic multi request for a long time to carry out the FEA hampers its use spectrometry (REMS) in the clinical practice, which will perhaps have to wait for the improvement of the underlying technology and mathe- Basic principles matical-physical processes. REMS technology is a non-ionizing axial approach for oste- Currently unmet clinical needs oporosis diagnosis. The operating principle is based on the analysis of native raw unfiltered ultrasound signals, the so- Despite the mentioned specific added values of some called radiofrequency (RF) ultrasound signals, acquired dur- imaging methods alternative to DXA, the main routinely ing an echographic scan of lumbar vertebrae and/or femoral employed parameter to estimate bone strength and to predict neck. The analysis of native unfiltered ultrasound signals fracture risk is still represented by DXA-measured BMD. allows to retain the maximum information about the char- In fact, currently there is not a widely accessible and cost- acteristics of the investigated tissues, which are normally effective tool capable of significantly improving fracture risk filtered out during the conventional process of B-mode estimation in individuals with osteopenic or normal BMD, image reconstruction. The bone health status is assessed which account for about 50% of fragility fractures [93]. through the comparison of the analysed signal spectra with Typically, the described approaches increase the dura- previously derived reference spectral models for the consid- tion of the diagnostic procedure and ionizing radiation dose ered pathological and normal conditions [97, 98]. The large without a clear improvement of the clinical outcome. For amount of collected data related to internal bone structure instance, a very recent study [94] has evaluated the per- provides both quantity- and quality-related information, formance of QCT-based FEA as a predictor of hip fracture being thus theoretically suitable for the estimation of bone risk, documenting that, compared to the traditional DXA, strength and the prediction of fracture risk. it would imply a 15% increment in the total costs and an In more detail, in a REMS investigation, the probe is over 2000 times higher radiation dose, while reducing the placed on the abdomen or on the hip in order to visualize fractured patients by less than 5%. This is also coupled with of the target bone interface and the operator has to set the the important accessibility issues that are common to all appropriate values of scan depth and transducer focus. Sub- the QCT-based techniques, which require all the strictest sequently, the software detects the sought bone interfaces in precautions typical of X-ray-bearing techniques, including the sequence of acquired frames and identifies the regions also HR-pQCT, the use of which is further hindered by their of interest (ROIs) for the diagnostic evaluation (Fig. 4). One very limited availability in clinical contexts [57–59]. key-feature lies in the exploitation of B-mode images for However, although the cost-effectiveness of osteoporosis the identification of target bone interfaces and related ROIs, prevention through early diagnosis has been demonstrated combined with the diagnostic analyses performed on the 13 Aging Clinical and Experimental Research (2019) 31:1375–1389 1383 Fig. 4  Software-guided REMS acquisition on femoral neck. Before part of the echographic field of view, immediately below the focus starting the acquisition, the operator sets transducer focus and scan position. The software automatically detects the bone interface and depth in order to visualize the target bone interphase in the central identifies the region of interest (ROI) Fig. 5  REMS analysis is charac- terised by the parallel process- ing of the native raw unfiltered signals of several scan lines, deriving one spectrum from each scan line (sample spectra are shown on the right) RF data. REMS analysis is then characterised by the paral- lel processing of the unfiltered signals of several scan lines (Fig. 5). The analysis of single scan line spectra allows the auto- matic exclusion of signals corresponding to artefacts, such as calcifications or osteophytes, thanks to the identification of unexpected spectral features. The selected measured data are finally synthetized in a patient-specific spectrum of the considered bone target, which undergoes an advanced comparison with gender-, age-, site- and BMI-matched reference spectral models Fig. 6  Patient-specific spectra undergo advanced comparisons with extracted from a dedicated database (Fig. 6). age-, sex-, BMI- and site-matched spectral models of pathologic and Actually, the spectral modifications introduced by the healthy conditions physical properties of the bone structure that has backscat- tered the ultrasound signals are identified by the compari- REMS approach has the potential to calculate also param- son procedure, resulting in a BMD estimation and in the eters different from BMD, derived from bone quality indi- consequent diagnostic classification as healthy, osteopenic cators and further related to bone strength. Among these, or osteoporotic. The adopted approach is based on the cal- Fragility Score (FS) has been developed as an independ- culation of the Osteoporosis Score, which corresponds to ent indicator of bone quality that provides an estimation of the percentage of analysed spectra that were classified as fracture risk independently of BMD. It is a dimensionless “osteoporotic” through the dedicated spectral analyses [97, number in the range 0–100 obtained from the comparison 98]. Linear equations are then employed to transform Osteo- between the patient-specific spectrum and reference spectral porosis Score into BMD values. models obtained from patients with and without osteoporotic 13 1384 Aging Clinical and Experimental Research (2019) 31:1375–1389 fractures. Recent preliminary studies have shown that FS to be r = 0.94 for lumbar spine and r = 0.93 for femoral neck has a good accuracy in the discrimination between fractured (p < 0.001 for both). and non-fractured subjects, and also that FS is significantly Moreover, the agreement between BMD values obtained correlated with the fracture risk calculated by ­FRAX® when by DXA and REMS was calculated through the Bland–Alt- the latter included also the results of the femoral neck den- man method [104]: the average difference (expressed as sitometry [99–101]. bias ± 2 SDs) was − 0.004 ± 0.088 g/cm2 for lumbar spine Ongoing developments of REMS are addressed to the and − 0.006 ± 0.076 g/cm2 for femoral neck. The standard investigation of other musculoskeletal tissues, such as mus- error of the estimate (SEE) resulted equal to 0.044 g/cm2 cles and cartilage, in order to monitor the onset and evo- (5.3%) for lumbar spine and 0.038 g/cm2 (5.8%) for femoral lution of diseases like sarcopenia and arthrosis by early neck. detecting the corresponding tissue deteriorations through The results obtained from the referred multicentre clinical dedicated REMS parameters. study showed that REMS technology has a high sensitivity and specificity in osteoporotic patient identification, with a significant diagnostic agreement with the gold standard Clinical validation studies DXA in the classification of patients as healthy, osteopenic or osteoporotic (total diagnostic concordance was 88.8% for The application of REMS technology for osteoporosis diag- lumbar spine and 88.2% for femoral neck). Both intra- and nosis has been clinically validated through an observational inter-operator variability associated with REMS investiga- multicentre clinical trial involving 7 Italian centres that tions were better than the corresponding values typically enrolled over 1900 PMW [102]. The aim of the study was reported in literature for the employed comparative gold to assess the precision and diagnostic accuracy of REMS in standard DXA [105, 106]. comparison with the clinical gold standard reference, repre- Thanks to its radiation-free approach, REMS might be sented by DXA. All DXA acquisitions were double-checked applied for population mass investigations or prevention pro- to avoid possible errors as identified by Messina et al. [103]. grams, early diagnosis in clinical practice and therapeutic Similarly, a quality control was performed on all the REMS short-term follow-up. reports in order to verify the correct selection of scan depth In the meantime, in October 2018 the first device imple- and transducer focus. The analysis of the temporal distri- menting the REMS approach has received the clearance of bution of the errors indicated the possibility of improving the U.S.A. Food and Drug Administration (FDA) for the the clinical practicability of REMS through a more rigorous measurement of the diagnostic parameters BMD, T-score training of the operators, aimed at reducing the time span of and Z-score, and for monitoring bone changes in the clini- the learning curves and the related initial error rates. After cal routine. the cross-check, 1195 spinal and 1373 femoral comparable cases were analysed. Further studies and clinical perspectives The REMS intra-operator precision, expressed as root- mean-square coefficient of variation (RMS-CV) was 0.38% An International Multicentre Clinical Trial, called “Echo- (95% confidence interval: 0.28–0.48%) for lumbar spine and Bone” and focused on an extended study of the REMS 0.32% (0.24–0.40%) for femoral neck. The corresponding approach for osteoporosis diagnosis in a wide European least significant change for the 95% confidence level, calcu- clinical context through a head-to-head trial with DXA, is lated via the ISCD precision calculator (available at http:// currently ongoing in five reference centres for osteoporosis www.iscd.org/resour​ ces/calcul​ ators​ /), was 1.05% for lumbar management. At an interim analysis on 595 patients enrolled spine and 0.88% for femoral neck. The inter-operator repeat- in Barcelona (Spain) [107], preliminary data confirm the ability was 0.54% for spine and 0.48% for femoral neck. recently published results [102]. Afterwards, diagnostic accuracy of REMS was assessed Then, another recently completed longitudinal study by determining the concordance of REMS and DXA in [108] has quantified the predictive value of REMS T-score in the discrimination between osteoporotic, osteopenic and identifying patients at risk for osteoporotic fractures by per- healthy patients through the calculation of the correspond- forming a follow-up of up to 5 years for the possible occur- ing Cohen’s kappa (k), obtaining k = 0.82 for lumbar spine rence of fragility fractures in a population of 1370 women and k = 0.79 for femoral neck. who had undergone baseline spinal REMS and DXA scans. Sensitivity of REMS in discriminating osteoporotic sub- After the follow-up period, all the subjects were divided jects from non-osteoporotic ones turned out to be 91.7% into two Groups: Group A (with incident fractures) and for spine and 91.5% for femur, while specificity was 92.0% Group B (without incident fractures). As expected, signifi- and 91.8%, respectively. Furthermore, the degree of cor- cant differences between the two groups were found in both relation between DXA and REMS T-score values resulted REMS T-score (− 2.68 ± 1.28 in Group A vs − 2.03 ± 1.23 13 Aging Clinical and Experimental Research (2019) 31:1375–1389 1385 in Group B, p < 0.001) and DXA T-score (− 2.52 ± 1.20 and femoral BMD. Available scientific evidences describe in Group A vs − 2.08 ± 1.17 in Group B, p < 0.001). The REMS-estimated BMD as an accurate diagnostic parameter, employment of the typical threshold of T-score ≤ − 2.5 as a which resulted also a predictor of incident clinical fracture cut-off value for both the techniques produced the follow- risk in a representative sample of female subjects. Moreover, ing results in the identification of Group A patients by the REMS has shown a further potential in the assessment of two techniques: sensitivity = 65% and specificity = 57% for skeletal fragility based on bone structure quality through REMS (OR = 2.6); sensitivity = 57% and specificity = 57% the Fragility Score parameter, which is independent from for DXA (OR = 1.7). This performance of REMS T-score the densitometric evaluation. in the identification of patients at high risk for osteoporotic fractures will have to be further verified on a wider popula- Acknowledgements The ESCEO Working Group was funded by the ESCEO. The ESCEO receives unrestricted educational grants to sup- tion through the outcome of a similar study that is currently port its educational and scientific activities from non-governmental ongoing in six clinical centres. organizations, not-for-profit organizations, non-commercial or corpo- In general, the available data suggest that REMS technol- rate partners. The choice of topics, participants, content and agenda ogy might have a beneficial impact on current diagnostic of the Working Groups as well as the writing, editing, submission and reviewing of the manuscript are the sole responsibility of the ESCEO, protocols and subsequent patient management in the clinical without any influence from third parties. routine. Further clinical scenarios are envisaged for addi- tional REMS applications, including in particular fracture Funding The ESCEO Working Group was funded by the ESCEO as risk assessment in paediatric patients and pregnant women, detailed in the “Acknowledgements section”. No other funding was and also in patients at risk of secondary osteoporosis (e.g., received. diabetic, nephropathic, oncological patients). REMS adoption in the clinical routine is expected to Compliance with ethical standards increase in the near future because of the mentioned evi- Conflict of interest A Diez-Perez has been speaker or advisor for dences and advantages, and also because the technology AMGEN, UCB, Lilly, Roche, Echolight and Gilead and owns shares is currently undergoing the evaluation procedures to be of Active Life Sci. ML Brandi has received honoraria from Amgen, included in the relevant international guidelines. Notewor- Bruno Farmaceutici, Kyowa Kirin; Academic grants and/or speaker thy, this technology does not require radiological protec- from: Abiogen, Alexion, Amgen, Bruno Farmaceutici, Eli Lilly, Ky- owa Kirin, MSD, NPS, Servier, Shire, SPA; Consulting fees from: tion, which might have been a problem for the reception Alexion, Bruno Farmaceutici, Kyowa Kirin, Servier, Shire. B Cortet in primary care in some countries. Finally, the portability has received fees for occasional interventions as an expert or speaker facilitates the employment on hospitalized fractured patients for Amgen, Expanscience, Ferring, Lilly, Medtronic, MSD, Mylan, not-transferrable to the densitometry units and on patient Novartis, Roche Diagnostics, and UCB. S Gonnelli has received lec- ture fees or paid advisory boards by Lilly, UCB, Abiogen, Amgen. J-Y follow-up at home. Reginster has received Consulting fees or paid advisory boards from IBSA-GENEVRIER, MYLAN, RADIUS HEALTH, PIERRE FA- BRE; Lecture fees when speaking at the invitation of sponsor IBSA- Conclusions GENEVRIER, MYLAN, CNIEL, DAIRY RESEARCH COUNCIL (DRC); Grant Support from Industry (All through Institution): IBSA- GENEVRIER, MYLAN, CNIEL, RADIUS HEALTH. R Rizzoli has The impact of osteoporosis and the resulting bone fracture been Speaker Bureau or Member of Scientific Advisory Boards for on both patient’s life and healthcare systems is constantly Danone, Echolight, Effryx, Mylan, Nestlé, ObsEva, Pfizer, Radius increasing: means for early diagnosis and monitoring of this Health, Sandoz and TEVA/Theramex. T Thomas has received per- sonal fees for lectures and expertises from Abbvie, Amgen, Arrow, disease are urgently needed in order to prevent and reduce BMS, Chugai, Expanscience, Gilead, HAC-Pharma, LCA, Lilly, Me- the occurrence of fractures. BMD measurement based on dac, MSD, Pfizer, Theramex, Thuasne, TEVA and UCB and financial DXA remains the current gold standard for osteoporosis support or fees for research activities from Amgen, Bone Therapeutics, classification, as defined by WHO. However, the above- Chugai, HAC-Pharma, MSD, Novartis, Pfizer, and UCB. All the other authors declare that they have no conflicts of interest concerning this mentioned clinical situation and the reported limitations of article. this technique boosted the scientific research to identify reli- able and accurate alternative approaches for bone strength Statement of human and animal rights No humans or animals have estimation and identification of individuals at high risk of been involved in this study. fracture. Those efforts are leading to a deeper understand- Informed consent For this type of study formal consent is not required. ing of intrinsic bone characteristics, related not only to bone quantity but also to bone quality, which might contribute to Open Access This article is distributed under the terms of the Crea- improve the accuracy of bone strength measurement and tive Commons Attribution 4.0 International License (http://creat​iveco​ fracture risk assessment. mmons​.org/licen​ses/by/4.0/), which permits unrestricted use, distribu- In this context, REMS represents the first clinically avail- tion, and reproduction in any medium, provided you give appropriate able method for direct non-ionizing measurement of lumbar 13 1386 Aging Clinical and Experimental Research (2019) 31:1375–1389 credit to the original author(s) and the source, provide a link to the 18. Wainwright SA, Marshall LM, Ensrud KE et al (2005) Hip frac- Creative Commons license, and indicate if changes were made. ture in women without osteoporosis. J Clin Endocrinol Metab 90:2787–2793 19. Schacter GI, Leslie WD (2017) DXA-based measurements in dia- betes: can they predict fracture risk? Calcif Tissue Int 100:150– 164. https​://doi.org/10.1007/s0022​3-016-0191-x References 20. Kanis JA, Borgstrom F, De Laet C et al (2005) Assessment of fracture risk. Osteoporos Int 16:581–589 1. Anonymous (1993) Consensus development conference: diag- 21. Müller D, Pulm J, Gandjour A (2012) Cost-effectiveness of dif- nosis, prophylaxis and treatment of osteoporosis. Am J Med ferent strategies for selecting and treating individuals at increased 94:646–650 risk of osteoporosis or osteopenia: a systematic review. Value 2. Armas LAG, Lappe JM, Heaney RP (2010) Calcium, bone Health 15:284–298. https​://doi.org/10.1016/j.jval.2011.11.030 strength and fractures. In: Orwoll ES, Bilezikian JP, Vander- 22. Nayak S, Roberts MS, Greenspan SL (2011) Cost-effectiveness schueren D (eds) Osteoporosis in men, 2nd edn. Academic Press, of different screening strategies for osteoporosis in postmeno- London, pp 235–241 pausal women. Ann Intern Med 155:751–761 3. Choksi P, Jepsen KJ, Clines GA (2018) The challenges of diag- 23. Pothuaud L, Carceller P, Hans D (2008) Correlations between nosing osteoporosis and the limitations of currently available grey-level variations in 2D projection images (TBS) and 3D tools. Clin Diabetes Endocrinol 4:12. https​://doi.org/10.1186/ microarchitecture: applications in the study of human trabec- s4084​2-018-0062-7 ular bone microarchitecture. Bone 42:775–787. https​: //doi. 4. Fonseca H, Moreira-Gonçalves D, Coriolano HJ et al (2014) org/10.1016/j.bone.2007.11.018 Bone quality: the determinants of bone strength and fragil- 24. Silva BC, Leslie WD, Resch H et al (2014) Trabecular bone ity. Sports Med 44:37–53. https ​ : //doi.org/10.1007/s4027​ score: a noninvasive analytical method based upon the DXA 9-013-0100-7 image. J Bone Miner Res 29:518–530. https​://doi.org/10.1002/ 5. Amman P, Rizzoli R (2003) Bone strength and its determi- jbmr.2176 nants. Osteoporos Int 14:13. https​: //doi.org/10.1007/s0019​ 25. Hans D, Goertzen AL, Krieg MA et al (2011) Bone micro- 8-002-1345-4 architecture assessed by TBS predicts osteoporotic fractures 6. Hart NH, Nimphius S, Rantalainen T et al (2017) Mechani- independent of bone density: the Manitoba study. J Bone Miner cal basis of bone strength: influence of bone material, bone Res 26:2762–2769. https​://doi.org/10.1002/jbmr.499 structure and muscle action. J Musculoskelet Neuronal Interact 26. Leslie WD, Majumdar SR, Morin SN et al (2017) Change in 17:114–139 trabecular bone score (TBS) with antiresorptive therapy does 7. Fogelman I, Blake GM (2000) Different approaches to bone den- not predict fracture in women: the Manitoba BMD cohort. sitometry. J Nucl Med 41:2015–2025 J Bone Miner Res 32:618–623. https ​ : //doi.org/10.1002/ 8. Damilakis J, Adams JE, Guglielmi G et al (2010) Radiation jbmr.3054 exposure in X-ray-based imaging techniques used in osteopo- 27. Leslie WD, Aubry-Rozier B, Lamy O et al (2013) TBS (trabecu- rosis. Eur Radiol 20:2707–2714. https​://doi.org/10.1007/s0033​ lar bone score) and diabetes-related fracture risk. J Clin Endo- 0-010-1845-0 crinol Metab 98:602–609. https​://doi.org/10.1210/jc.2012-3118 9. [No authors listed] (2003) Prevention and management of osteo- 28. Martineau P, Silva BC, Leslie WD (2017) Utility of trabecular porosis. World Health Organ Tech Rep Ser 921:1–164 bone score in the evaluation of osteoporosis. Curr Opin Endo- 10. Bousson V, Le Bras A, Roqueplan F et al (2006) Volumetric crinol Diabetes Obes 24:402–410. https​: //doi.org/10.1097/ quantitative computed tomography of the proximal femur: rela- MED.00000​00000​00036​5 tionships linking geometric and densitometric variables to bone 29. Kanis JA, Johnell O, Oden A et al (2008) FRAX and the assess- strength. Role for compact bone. Osteoporos Int 17:855–864 ment of fracture probability in men and women from the UK. 11. World Health Organization (1994) Assessment of fracture risk Osteoporos Int 19:385–397. https​: //doi.org/10.1007/s0019​ and its application to screening for postmenopausal osteoporosis: 8-007-0543-5 report of a WHO study group [meeting held in Rome from 22 to 30. McCloskey EV, Odén A, Harvey NC et al (2016) A meta-analysis 25 June 1992] of trabecular bone score in fracture risk prediction and its rela- 12. Kanis JA, Melton LJ, Christiansen C et al (1994) The diagnosis tionship to FRAX. J Bone Miner Res 31:940–948 of osteoporosis. J Bone Miner Res 9:1137–1141 31. Winzenrieth R, Michelet F, Hans D (2013) Three-dimensional 13. Blake G, Adams JE, Bishop N (2013) DXA in adults and chil- (3D) microarchitecture correlations with 2D projection image dren. In: Rosen CJ, Delmas P (eds) Primer on the metabolic bone gray-level variations assessed by trabecular bone score using diseases and disorders of mineral metabolism, 8th edn. American high-resolution computed tomographic acquisitions: effects of Society of Bone and Mineral Research (ASBMR), Washington, resolution and noise. J Clin Densitom 16:287–296 pp 152–158 32. Chen W, Slattery A, Center J et al (2016) The effect of changing 14. Marshall D, Johnell O, Wedel H (1996) Meta-analysis of how scan mode on trabecular bone score using lunar prodigy. J Clin well measures of bone mineral density predict occurrence of Densitom 19:502–506 osteoporotic fractures. BMJ 312:1254–1259 33. Amnuaywattakorn S, Sritara C, Utamakul C et al (2016) Simu- 15. Leslie WD, Majumdar SR, Morin SN et al (2015) Why does rate lated increased soft tissue thickness artefactually decreases of bone density loss not predict fracture risk? J Clin Endocrinol trabecular bone score: a phantom study. BMC Musculoskelet Metab 100:679–683. https​://doi.org/10.1210/jc.2014-3777 Disord 17:17. https​://doi.org/10.1186/s1289​1-016-0886-1 16. Siris ES, Chen YT, Abbott TA et al (2004) Bone mineral density 34. Schousboe JT, Vo TN, Langsetmo L et al (2017) Association of thresholds for pharmacological intervention to prevent fractures. trabecular bone score (TBS) with incident clinical and radio- Arch Intern Med 164:1108–1112 graphic vertebral fractures adjusted for lumbar spine BMD 17. Unnanuntana A, Gladnick BP, Donnelly E et al (2010) The in older men: a prospective cohort study. J Bone Miner Res assessment of fracture risk. J Bone Joint Surg Am 92:743–753 32:1554–1558. https​://doi.org/10.1002/jbmr.3130 35. Langsetmo L, Vo TN, Ensrud KE et al (2016) The association between trabecular bone score and lumbar spine volumetric 13 Aging Clinical and Experimental Research (2019) 31:1375–1389 1387 BMD is attenuated among older men with high body mass cross-sectionalstudy. Osteoporos Int 29:191–200. https​://doi. index. J Bone Miner Res 31:1820–1826. https:​ //doi.org/10.1002/ org/10.1007/s0019​8-017-4232-8 jbmr.2867 53. Leslie WD, Luo Y, Yang S et al (2019) Fracture risk indices 36. Roux JP, Wegrzyn J, Boutroy S, Bouxsein ML et al (2013) The from DXA-based finite element analysis predict incident frac- predictive value of trabecular bone score (TBS) on whole lumbar tures independently from FRAX: The Manitoba BMD Registry. vertebrae mechanics: an ex vivo study. Osteoporos Int 24:2455– J Clin Densitom. https​://doi.org/10.1016/j.jocd.2019.02.001 2460. https​://doi.org/10.1007/s0019​8-013-2316-7 54. Stagi S, Cavalli L, Cavalli T et al (2016) Peripheral quantita- 37. Maquer G, Lu Y, Dall’Ara E et al (2016) The initial slope of the tive computed tomography (pQCT) for the assessment of bone variogram, foundation of the trabecular bone score, is not or strength in most of bone affecting conditions in developmental is poorly associated with vertebral strength. J Bone Miner Res age: a review. Ital J Pediatr 42:88 31:341–346. https​://doi.org/10.1002/jbmr.2610 55. MacNeil JA, Boyd SK (2008) Improved reproducibility of high- 38. Harvey NC, Glüer CC, Binkley N et al (2015) Trabecular bone resolution peripheral quantitative computed tomography for score (TBS) as a new complementary approach for osteoporo- measurement of bone quality. Med Eng Phys 30:792–799. https​ sis evaluation in clinical practice. Bone 78:216–224. https​://doi. ://doi.org/10.1016/j.meden​gphy.2007.11.003 org/10.1016/j.bone.2015.05.016 56. Burghardt AJ, Link TM, Majumdar S (2011) High-resolution 39. Edmondson CP, Schwartz EN (2017) Non-BMD DXA meas- computed tomography for clinical imaging of bone microarchi- urements of the hip. Bone 104:73–83. https​://doi.org/10.1016/j. tecture. Clin Orthop Relat Res 469:2179–2193. https​://doi. bone.2017.03.050 org/10.1007/s1199​9-010-1766-x 40. Leslie WD, Lix LM, Morin SN et al (2016) Adjusting hip frac- 57. van Rietbergen B, Ito K (2015) A survey of micro-finite ele- ture probability in men and women using hip axis length: the ment analysis for clinical assessment of bone strength: the first Manitoba bone density database. J Clin Densitom 19:326–331. decade. J Biomech 48:832–841. https​://doi.org/10.1016/j.jbiom​ https​://doi.org/10.1016/j.jocd.2015.07.004 ech.2014.12.024 41. Broy SB, Cauley JA, Lewiecki ME et al (2015) Fracture risk 58. Digby MG, Bishop NJ, Paggiosi MA et al (2016) HR-pQCT: a prediction by non-BMD DXA measures: the 2015 ISCD Official non-invasive ‘biopsy’ to assess bone structure and strength. Arch Positions Part 1: hip geometry. J Clin Densitom 18:287–308. Dis Child Educ Pract Ed 101:268–270. https​://doi.org/10.1136/ https​://doi.org/10.1016/j.jocd.2015.06.005 archd​ischi​ld-2015-30945​5 42. Beck TJ, Broy SB (2015) Measurement of hip geometry—tech- 59. Lespessailles E, Ibrahim-Nasser N, Toumi H et al (2018) Contri- nical background. J Clin Densitom 18:331–337. https​://doi. bution of high resolution peripheral quantitative CT to the man- org/10.1016/j.jocd.2015.06.006 agement of bone and joint diseases. Joint Bone Spine 85:301– 43. Zhuang H, Li Y, Lin J et al (2017) Cortical thickness in the inter- 306. https​://doi.org/10.1016/j.jbspi​n.2017.04.012 trochanteric region may be relevant to hip fracture type. BMC 60. Biver E, Durosier-Izart C, Chevalley T et al (2018) Evaluation of Musculoskelet Disord 18:305. https​://doi.org/10.1186/s1289​ radius microstructure and areal bone mineral density improves 1-017-1669-z fracture prediction in postmenopausal women. J Bone Miner Res 44. Lee DH, Jung KY, Hong AR et al (2016) Femoral geometry, bone 33:328–337. https​://doi.org/10.1002/jbmr.3299 mineral density, and the risk of hip fracture in premenopausal 61. Cheung AM, Adachi JD, Hanley DA et al (2013) High-resolution women: a case control study. BMC Musculoskelet Disord 17:42. peripheral quantitative computed tomography for the assessment https​://doi.org/10.1186/s1289​1-016-0893-2 of bone strength and structure: a review by the Canadian Bone 45. Adams JE (2009) Quantitative computed tomography. Eur J Strength Working Group. Curr Osteoporos Rep 11:136–146. Radiol 71:415–424 https​://doi.org/10.1007/s1191​4-013-0140-9 46. Link TM, Lang TF (2014) Axial QCT: clinical applications and 62. Chang G, Honig S, Liu Y et al (2015) 7 Tesla MRI of bone new developments. J Clin Densitom 17:438–448. https​://doi. microarchitecture discriminates between women without and org/10.1016/j.jocd.2014.04.119 with fragility fractures who do not differ by bone mineral density. 47. Manhard MK, Nyman JS, Does MD (2017) Advances in imaging J Bone Miner Metab 33:285–293. https​://doi.org/10.1007/s0077​ approaches to fracture risk evaluation. Transl Res 181:1–14. https​ 4-014-0588-4 ://doi.org/10.1016/j.trsl.2016.09.006 63. Griffith JF, Genant HK (2012) New advances in imaging osteo- 48. Kopperdahl DL, Aspelund T, Hoffmann PF et al (2014) Assess- porosis and its complications. Endocrine 42:39–51. https​://doi. ment of incident spine and hip fractures in women and men using org/10.1007/s1202​0-012-9691-2 finite element analysis of CT scans. J Bone Miner Res 29:570– 64. Hans D, Baim S (2017) Quantitative ultrasound (QUS) in 580. https​://doi.org/10.1002/jbmr.2069 the management of osteoporosis and assessment of fracture 49. Keaveny TM, Hoffmann PF, Singh M et al (2008) Femoral risk. J Clin Densitom 20:322–333. https​://doi.org/10.1016/j. bone strength and its relation to cortical and trabecular changes jocd.2017.06.018 after treatment with PTH, alendronate, and their combination 65. Shepherd JA, Schousboe JT, Broy SB et al (2015) Executive as assessed by finite element analysis of quantitative CT scans. summary of the 2015 ISCD position development conference J Bone Miner Res 23:1974–1982. https​: //doi.org/10.1359/ on advanced measures from DXA and QCT: fracture predic- jbmr.08080​5 tion beyond BMD. J Clin Densitom 18(3):274–286. https​://doi. 50. Areeckal AS, Kocher M, Sumam David S (2018) Current and org/10.1016/j.jocd.2015.06.013 emerging diagnostic imaging-based techniques for assessment of 66. Langton CM, Ali AV, Riggs CM et al (1990) A contact method osteoporosis and fracture risk. IEEE Rev Biomed Eng 12:254– for the assessment of ultrasonic velocity and broadband attenu- 268. https​://doi.org/10.1109/RBME.2018.28526​20 ation in cortical and cancellous bone. Clin Phys Physiol Meas 51. Ruiz Wills C, Olivares AL, Tassani S et al (2019) 3D patient-spe- 11:243–249 cific finite element models of the proximal femur based on DXA 67. Prins SH, Jørgensen HL, Jørgensen LV et al (1998) The role of towards the classification of fracture and non-fracture cases. quantitative ultrasound in the assessment of bone: a review. Clin Bone 121:89–99. https​://doi.org/10.1016/j.bone.2019.01.001 Physiol 18:3–17 52. Yang S, Leslie WD, Luo Y et al (2018) Automated DXA-based 68. Damilakis J, Maris TG, Karantanas AH (2007) An update on finite element analysis for hip fracture risk stratification: a the assessment of osteoporosis using radiologic techniques. Eur Radiol 17:1591–1602 13 1388 Aging Clinical and Experimental Research (2019) 31:1375–1389 69. Chan MY, Nguyen ND, Center JR et al (2013) Quantitative ultra- 85. Shen J, Griffith JF, Zhu TY et al (2018) Bone mass, micro- sound and fracture risk prediction in non-osteoporotic men and structure, and strength can discriminate vertebral fracture in women as defined by WHO criteria. Osteoporos Int 24:1015– patients on long-term steroid treatment. J Clin Endocrinol Metab 1022. https​://doi.org/10.1007/s0019​8-012-2001-2 103:3340–3349. https​://doi.org/10.1210/jc.2018-00490​ 70. McLeod KM, Johnson S, Rasali D et al (2015) Discriminatory 86. Johannesdottir F, Allaire B, Bouxsein ML (2018) Fracture pre- performance of the calcaneal quantitative ultrasound and oste- diction by computed tomography and finite element analysis: oporosis self-assessment tool to select older women for dual- current and future perspectives. Curr Osteoporos Rep 16:411. energy X-ray absorptiometry. J Clin Densitom 18:157–164. https​ https​://doi.org/10.1007/s1191​4-018-0450-z ://doi.org/10.1016/j.jocd.2015.02.006 87. Panyasantisuk J, Dall’Ara E, Pretterklieber M et al (2018) Map- 71. Zhang L, Lv H, Zheng H et al (2015) Correlation between param- ping anisotropy improves QCT-based finite element estimation eters of calcaneal quantitative ultrasound and hip structural anal- of hip strength in pooled stance and side-fall load configura- ysis in osteoporotic fracture patients. PLoS ONE 10:e0145879. tions. Med Eng Phys 59:36–42. https:​ //doi.org/10.1016/j.meden​ https​://doi.org/10.1371/journ​al.pone.01458​79 gphy.2018.06.004 72. Marín F, González-Macías J, Díez-Pérez A et al (2006) Rela- 88. Szulc P, Boutroy S, Chapurlat R (2018) Prediction of fractures in tionship between bone quantitative ultrasound and fractures: a men using bone microarchitectural parameters assessed by high- meta-analysis. J Bone Miner Res 21:1126–1135 resolution peripheral quantitative computed tomography—the 73. McCloskey EV, Johansson H, Kanis JA et al (2015) Predictive prospective STRAMBO study. J Bone Miner Res 33:1470–1479. ability of heel quantitative ultrasound for incident fractures: an https​://doi.org/10.1002/jbmr.3451 individual-level meta-analysis. Osteoporos Int 26:1979–1987. 89. Unal M, Creecy A, Nyman JS (2018) The role of matrix com- https​://doi.org/10.1007/s0019​8-015-3072-7 position in the mechanical behaviour of bone. Curr Osteoporos 74. Black DM, Cauley JA, Wagman R et al (2018) The ability of a Rep 16:205–215. https​://doi.org/10.1007/s1191​4-018-0433-0 single BMD and fracture history assessment to predict fracture 90. Herrera S, Diez-Perez A (2017) Clinical experience with over 25 years in postmenopausal women: the study of osteo- microindentation in vivo in humans. Bone 95:175–182. https​:// porotic fractures. J Bone Miner Res 33:389–395. https​://doi. doi.org/10.1016/j.bone.2016.11.003 org/10.1002/jbmr.3194 91. Rozental TD, Walley KC, Demissie S et al (2018) Bone mate- 75. Austin M, Yang YC, Vittinghoff E et al (2012) Relationship rial strength index as measured by impact microindentation in between bone mineral density changes with denosumab treat- postmenopausal women with distal radius and hip fractures. J ment and risk reduction for vertebral and nonvertebral fractures. J Bone Miner Res 33:621–626. https​://doi.org/10.1002/jbmr.3338 Bone Miner Res 27:687–693. https​://doi.org/10.1002/jbmr.1472 92. Winzenrieth R, Humbert L, Di Gregorio S et al (2018) Effects 76. Jacques RM, Boonen S, Cosman F et al (2012) Relationship of of osteoporosis drug treatments on cortical and trabecular bone changes in total hip bone mineral density to vertebral and nonver- in the femur using DXA-based 3D modeling. Osteoporos Int tebral fracture risk in women with postmenopausal osteoporosis 29:2323–2333. https​://doi.org/10.1007/s0019​8-018-4624-4 treated with once-yearly zoledronic acid 5 mg: the HORIZON- 93. Bliuc D, Alarkawi D, Nguyen TV et al (2015) Risk of subse- Pivotal Fracture Trial (PFT). J Bone Miner Res 27:1627–1634. quent fractures and mortality in elderly women and men with https​://doi.org/10.1002/jbmr.1644 fragility fractures with and without osteoporotic bone density: 77. Bouxsein ML, Eastell R, Lui LY et al (2019) Change in bone den- the Dubbo Osteoporosis Epidemiology Study. J Bone Miner Res sity and reduction in fracture risk: a meta-regression of published 30:637–646. https​://doi.org/10.1002/jbmr.2393 trials. J Bone Miner Res 1:1. https​://doi.org/10.1002/jbmr.3641 94. Viceconti M, Qasim M, Bhattacharya P et al. (2018) Are CT- 78. Kanis JA, Cooper C, Rizzoli R et al (2019) European guidance based finite element model predictions of femoral bone strength- for the diagnosis and management of osteoporosis in postmeno- ening clinically useful? Curr Osteoporos Rep 16:216–223. https​ pausal women. Osteoporos Int. 30:3–44. https:​ //doi.org/10.1007/ ://doi.org/10.1007/s11914​ -018-0438-8. Review. Erratum in: Curr s0019​8-018-4704-5 Osteoporos Rep. 2018 Jun 22 79. Osterhoff G, Morgan EF, Shefelbine SJ et al (2016) Bone 95. Kanis JA, Hiligsmann M (2014) The application of health mechanical properties and changes with osteoporosis. Injury technology assessment in osteoporosis. Best Pract Res Clin 47:S11–S20. https​://doi.org/10.1016/S0020​-1383(16)47003​-8 Endocrinol Metab 28:895–910. https ​ : //doi.org/10.1016/j. 80. McCreadie BR, Goldstein SA (2000) Biomechanics of fracture: beem.2014.04.001 is bone mineral density sufficient to assess risk? J Bone Miner 96. Minniti D, Davini O, Gualano MR et al (2014) Techniques for Res 15:2305–2308 diagnosing osteoporosis: a systematic review of cost-effective- 81. Keaveny TM, Bouxsein ML (2008) Theoretical implications ness studies. Int J Technol Assess Health Care 30:273–281. https​ of the biomechanical fracture threshold. J Bone Miner Res ://doi.org/10.1017/S0266​46231​40002​57 23:1541–1547. https​://doi.org/10.1359/jbmr.08040​6 97. Conversano F, Franchini R, Greco A et al (2015) A novel ultra- 82. Boutroy S, Bouxsein ML, Munoz F et al (2005) In vivo assess- sound methodology for estimating spine mineral density. Ultra- ment of trabecular bone microarchitecture by high-resolution sound Med Biol 41:281–300. https​://doi.org/10.1016/j.ultra​ peripheral quantitative computed tomography. J Clin Endocrinol smedb​io.2014.08.017 Metab 90:6508–6515 98. Casciaro S, Peccarisi M, Pisani P et al (2016) An advanced quan- 83. Samelson EJ, Broe KE, Xu H et al (2019) Cortical and trabecular titative echosound methodology for femoral neck densitometry. bone microarchitecture as an independent predictor of incident Ultrasound Med Biol 42:1337–1356. https​://doi.org/10.1016/j. fracture risk in older women and men in the Bone Microarchi- ultra​smedb​io.2016.01.024 tecture International Consortium (BoMIC): a prospective study. 99. Caffarelli C, Tomai Pitinica MD, Francolini V et al (2018) REMS Lancet Diabetes Endocrinol 7:34–43. https​://doi.org/10.1016/ technique: future perspectives in an Academic Hospital. Clin s2213​-8587(18)30308​-5. Erratum in: Lancet Diabetes Endo- Cases Miner Bone Metab 15:163–165 crinol 7:e1 100. Greco A, Pisani P, Conversano F et al (2017) Ultrasound fragility 84. Eastell R, Wahner HW, O’Fallon WM et al (1989) Unequal score: an innovative approach for the assessment of bone fragil- decrease in bone density of lumbar spine and ultradistal radius ity. Measurement 101:236–242 in Colles’ and vertebral fracture syndromes. J Clin Invest 83:168–174 13 Aging Clinical and Experimental Research (2019) 31:1375–1389 1389 101. Pisani P, Greco A, Conversano F et al (2017) A quantitative ultra- 106. Ravaud P, Reny JL, Giraudeau B et al (1999) Individual smallest sound approach to estimate bone fragility: a first comparison with detectable difference in bone mineral density measurements. J dual X-ray absorptiometry. Measurement 101:243–249 Bone Miner Res 14:1449–1456 102. Di Paola M, Gatti D, Viapiana O et al (2018) Radiofrequency 107. Ovejero Crespo D, Nogues X, Diez-Perez A (2019) The non- echographic multispectrometry compared with dual X-ray ionizing radiofrequency echographic multi spectrometry (REMS) absorptiometry for osteoporosis diagnosis on lumbar spine applied on a Spanish cohort for the osteoporosis diagnosis on and femoral neck. Osteoporos Int 30:391–402. https​: //doi. lumbar spine and femoral neck. WCO-IOF-ESCEO Abstract org/10.1007/s0019​8-018-4686-3 P795 103. Messina C, Bandirali M, Sconfienza LM et al (2015) Preva- 108. Adami G, Arioli G, Bianchi G et al (2019) Prediction of inci- lence and type of errors in dual-energy X-ray absorptiom- dent fragility fractures through radiofrequency echographic multi etry. Eur Radiol 25:1504–1511. https​://doi.org/10.1007/s0033​ spectrometry (REMS). Ann Rheum Dis 78(Suppl 2):933. https:​ // 0-014-3509-y doi.org/10.1136/annrh​eumdi​s-2019-eular​.6256 104. Altman DG, Bland JM (1983) Measurements in medicine: the analysis of method comparison studies. Statistician 32:307–317 Publisher’s Note Springer Nature remains neutral with regard to 105. Hopkins SJ, Welsman JR, Knapp KM (2014) Short-term preci- jurisdictional claims in published maps and institutional affiliations. sion error in dual energy X-ray absorptiometry, bone mineral density and trabecular bone score measurements; and effects of obesity on precision error. J Biomed Gr Comput 4:8–14 Affiliations Adolfo Diez‑Perez1 · Maria Luisa Brandi2,3 · Nasser Al‑Daghri4 · Jaime C. Branco5 · Olivier Bruyère6 · Loredana Cavalli2,3 · Cyrus Cooper7 · Bernard Cortet8 · Bess Dawson‑Hughes9 · Hans Peter Dimai10 · Stefano Gonnelli11 · Peyman Hadji12 · Philippe Halbout13 · Jean‑Marc Kaufman14 · Andreas Kurth15,16 · Medea Locquet17 · Stefania Maggi18 · Radmila Matijevic19,20 · Jean‑Yves Reginster4,6 · René Rizzoli21 · Thomas Thierry22,23 11 * Adolfo Diez‑Perez Department of Medicine, Surgery and Neurosciences, ADiez@parcdesalutmar.cat University of Siena, Siena, Italy 12 1 Frankfurter Hormon und Osteoporose Zentrum, Frankfurt, Department of Internal Medicine, Hospital del Mar/IMIM Germany and CIBERFES, Autonomous University of Barcelona, 13 Passeig Maritim 25‑29, 08003 Barcelona, Spain International Osteoporosis Foundation, Nyon, Switzerland 2 14 FirmoLab Fondazione F.I.R.M.O., Florence, Italy Department of Endocrinology, Ghent University Hospital, 3 Ghent, Belgium Department of Biological, Experimental and Clinical 15 Science, University of Florence, Florence, Italy Department of Orthopaedic Surgery and Osteology, 4 Klinikum Frankfurt, Frankfurt, Germany Chair for Biomarkers of Chronic Diseases, Biochemistry 16 Department, College of Science, King Saud University, Mayor Teaching Hospital, Charite Medical School, Berlin, Riyadh, Kingdom of Saudi Arabia Germany 5 17 NOVA Medical School, Universidade Nova de Lisboa, Department of Public Health, Epidemiology and Health Lisbon, Portugal Economics, University of Liège, Liège, Belgium 6 18 WHO Collaborating Centre for Public Health Aspects National Research Council, Aging Program, Institute of Musculoskeletal Health and Aging, University of Liège, of Neuroscience, Padua, Italy Liège, Belgium 19 Faculty of Medicine, University of Novi Sad, Novi Sad, 7 MRC Lifecourse Epidemiology Unit, Southampton General Serbia Hospital, University of Southampton, Southampton, UK 20 Clinical Center of Vojvodina, Clinic for Orthopedic Surgery, 8 Department of Rheumatology and EA 4490, Novi Sad, Serbia University-Hospital of Lille, Lille, France 21 Service of Bone Diseases, Geneva University Hospitals 9 Bone Metabolism Laboratory, Jean Mayer USDA Human and Faculty of Medicine, Geneva, Switzerland Nutrition Research Center on Aging, Tufts University, 22 Department of Rheumatology, Hospital Nord, CHU St Boston, MA, USA Etienne, St Etienne, France 10 Department of Internal Medicine, Division of Endocrinology 23 INSERM 1059, University of Lyon, St Etienne, France and Diabetology, Medical University of Graz, Graz, Austria 13