Can Novel Clinical Densitometric Techniques Replace or Improve DXA in Predicting Bone Strength in Osteoporosis at the Hip and Other Skeletal Sites?


  • The authors have no conflict of interest.


New peripheral techniques are now available for the diagnosis of osteoporosis, but their value in the clinical management of the disease remains controversial. This study tests the hypothesis that peripheral quantitative computed tomography (pQCT) at the distal radius and/or quantitative ultrasound (QUS) at the calcaneus can serve as replacement or improvement of current methodology (QCT and DXA) for predicting bone strength at the hip and other sites. In 126 human cadavers (age, 80.2 ± 10.4 years), DXA of the femur, spine, and radius and pQCT of the radius were acquired with intact soft tissues. QCT (spine) and QUS (calcaneus) were performed ex situ in degassed specimens. Femoral failure loads were assessed in side impact and vertical loading. Failure loads of the thoracolumbar spine were determined at three levels in compression and those of the radius by simulating a fall. Site-specific DXA explained approximately 55% of the variability in femoral strength, whereas pQCT and QUS displayed a lower association (15-40%). QUS did not provide additional information on mechanical strength of the femur, spine, or radius. All techniques displayed similar capability in predicting a combined index of failure strength at these three sites, with only QUS exhibiting significantly lower associations than other methods. These experimental results suggest that clinical assessment of femoral fracture risk should preferably rely on femoral DXA, whereas DXA, QCT, and pQCT display similar capability of predicting a combined index of mechanical strength at the hip, spine, and radius.


Osteoporotic fractures represent a major health care problem relevant to all geographic areas and populations around the world. They severely affect the quality of life and mortality of the elderly,(1, 2) with costs being estimated of more than $14 billion in the Unites States.(3) The clinical and socio-economic relevance of osteoporosis is highlighted by declaration of this decade as the bone and joint decade (2000 to 2010) by the World Health Organization and other institutions. Because effective treatment is available,(4, 5) accurate noninvasive prediction of bone strength and fracture risk and an efficient selection of patients to treat represent important goals in medical diagnostics.

Although several (standard and novel) techniques are currently available for diagnosis of osteoporosis clinically,(6) which method is best suited for this purpose is controversial. DXA is considered the gold standard, and quantitative measures of bone mass (bone mineral content [BMC]) can be derived directly at all sites of interest.(6) The disadvantage of DXA is, however, that the surrounding soft tissues can introduce relevant measurement errors,(7, 8) that bone mineral density (BMD, g/cm2) measurements are affected by variations in bone size,(9, 10) and that cortical and trabecular bone cannot be separated.(6) Quantitative computed tomography (QCT) is less prone to soft tissue errors and permits analysis of trabecular and cortical bone separately.(6, 11) However, QCT involves relatively high X-ray doses and relies on expensive equipment that is rarely available in outpatient or private practice settings.(6)

The advantage of novel peripheral measurement techniques is that they are relatively cheap and easy to use and that they potentially permit to measure bone properties beyond those of bone mass or density.(11) However, their value in the diagnostics of osteoporosis and in selecting patients for treatment has remained controversial. Peripheral QCT (pQCT) requires smaller amounts of radiation that are applied to less vulnerable (peripheral) sites, usually the distal radius.(12) In particular, pQCT offers the opportunity to determine specific geometric parameters that have been shown to be associated with whole (structural) bone strength.(11, 12) Because of the heterogeneity of bone properties throughout the skeleton,(13, 14) however, it is questionable whether peripheral measurements reflect bone properties of the hip and spine. Quantitative ultrasound (QUS) has been shown to display a small, but relevant, relationship with trabecular microstructure and mechanical properties.(15–17) This has stimulated hope that ultrasound can provide significant information for the prediction of bone strength, in addition to site-specific bone mass or density. The attractiveness of the technique also lies in the complete lack of ionizing radiation and in the low costs involved.(15, 17) As with pQCT, however, QUS can only be measured at peripheral sites.

The objective of the current study was therefore to comprehensively evaluate new peripheral (pQCT and QUS) methods that are currently used and approved for clinical bone densitometry in relation to conventional measurements with DXA and QCT. This was accomplished by comparing their ability to accurately predict experimental failure loads (as surrogates of fracture risk) at relevant skeletal sites. Because hip fractures have the most deleterious consequences on the individual's well-being and the highest socio-economic impact, focus of this paper is on fractures of the proximal femur. The correlation of densitometry with mechanical strength of the spine and distal radius has been previously reported(18, 19); in the current work, we will therefore also assess to which extent these techniques are capable of predicting a combined index of failure strength at the hip, spine, and distal radius. We will specifically test the hypothesis that pQCT at the distal radius and/or QUS at the calcaneus can serve as replacement or improvement of current methodology (DXA and QCT) and that QUS provides additional information in predicting bone strength at the hip and other relevant skeletal sites.


Study sample

126 formalin-fixed cadavers with intact skin and soft tissues were examined from a pool of 140 specimens used in a course of macroscopic dissection.(20) Specimens with bilateral hip endoprostheses (n = 3) and with bone diseases other than osteoporosis or osteopenia (n = 11) were excluded based on histomorphometry of iliac crest samples.(20) Forty-six specimens were male (age, 76.4 ± 11.4 years; mean ± SD) and 80 were female (age, 82.2 ± 9.0 years). The donors had dedicated their body to the institute several years before death and are therefore assumed to constitute a representative selection of the elderly population resident in Southern Germany.

Bone densitometry by DXA

DXA measurements (BMC, g; BMD, g/cm2) of the proximal femur (subregions: neck, trochanter, total femur), lumbar vertebra 2-4 (L2-L4), and the distal radius (distal 33%) were obtained under in situ conditions with a clinical scanner (DPX-L; Lunar Corp., Madison, WI, USA). The skin and soft tissues were left fully intact to reflect natural soft tissue errors occurring under clinical imaging conditions. pQCT was performed at the distal radius (4% and 20% from the wrist joint), also with intact soft tissues. An XCT 2000 (Stratec, Pforzheim, Germany) was used; the technical details and precision of the measurements were reported previously.(21) Both femora, the thoracolumbar spine, the right forearm, and the left calcaneus were excised. These were cleaned of muscles and soft tissues, except for the right forearm. All bones were radiographed in two planes to determine prior fracture. In these cases (n = 16 for L2-L4, n = 6 for the right forearm), the specific densitometric and biomechanical data of these bones were discarded.

QCT measurements were obtained at the midvertebral level of L2-L4 under ex situ conditions at 80 kV with a clinical CT (Somatom Plus 4; Siemens, Erlangen, Germany). Standard settings were applied, and a phantom was used to convert Hounsfield units into density values. To avoid artifacts from inclusions of air, the segments were degassed and measured within aqueous surrounding. QUS of the calcaneus was also obtained under ex situ conditions to avoid potential artifacts from gas in the soft tissues. The calcanei were dissected clean of the surrounding tissues and degassed as described above. Measurements were performed in a temperature-controlled water bath (37°C) at an anatomically defined position with a clinical scanner (Achilles+;Lunar), sealed for underwater usage.(22) The speed of sound (SOS), broadband ultrasound attenuation (BUA), and stiffness index (SI) were determined.(17)

Mechanical testing

Pairs of femora were sorted according to age; the left and right sides were assigned in alternating order to the two loading configurations. One side was tested by simulating a sideways fall on the greater trochanter,(23–25) whereas the contralateral femur was tested by applying a vertical load to the femoral head, parallel to the shaft.(9, 26) In cases where only one femur was available (unilateral hip arthroplasty; n = 11), specimens were assigned to one of both tests, so that 120 were tested in side impact and 121 in vertical loading. The tests were performed at 6.5 mm/s, using a Zwick 1445 (Zwick, Ulm, Germany) material testing machine. In the side impact configuration, 60 specimens displayed cervical fractures and 29 displayed trochanteric fractures. Specimens with clinically untypical fractures (crush fractures of the trochanter or head and shaft fractures; n = 31) were discarded. In vertical loading, 110 specimens displayed fractures through the femoral neck and 11 through the greater trochanter. Only neck fractures were considered for further analysis.

The vertebral bodies of thoracic vertebrae 6 and 10 and of lumbar vertebra 3 were tested to failure as functional spinal units (T5-T7, T9-T11, L2-L4) with intact ligaments and discs in axial compression without posterior elements.(18, 27, 28) On the right forearm, a fall was simulated(19, 29, 30) with the hand in 70° dorsiflexion and 10° radial abduction. Only typical Colles fractures were considered for further analyses (n = 93). A more detailed description of the mechanical testing protocols, the descriptive data on failure loads, and their correlation between sites have been reported previously.(18–20)

To derive a combined index of bone strength at the hip, spine, and distal radius, we computed the percentage deviation of each individual failure load from the mean failure load at this particular skeletal site. The combined index was then determined as the mean of these deviations at the femur (only side impact configuration), at the spine (average of T6, T10, and L3), and at the distal radius. Note that the average failure load for the spine was computed from the three vertebrae in the same way as for the combined index. In all 126 specimens, at least one failure value was available, and in 61 specimens, failure loads were available at all three sites.

Statistical analysis

Correlation coefficients were computed by linear regression analysis (Statview 4.5; Abacus Concepts, Berkeley, CA, USA) and were considered significant at a 1% level. Fisher z-transformation was used to assess whether certain coefficients were significantly higher than others. Stepwise multiple regression models (forward mode) were used to determine whether QUS was able to add significant, independent information to DXA.


The correlation coefficients of DXA, QCT, pQCT, and QUS with femoral failure loads are shown in Table 1, and selected coefficients of determination are displayed in Fig. 1. DXA of the femoral neck provided the highest correlation coefficients with mechanical strength of the femur in both loading configurations. For vertical loading, all non-site-specific measurements displayed a significantly lower association than site-specific analysis (Table 1). In the side impact configuration, non-site-specific DXA, QCT, pQCT, and QUS also exhibited significantly lower correlation coefficients, except for measurements at the radius, where differences were not significant at the 5% level. QUS was unable to add significant independent information to neck BMC in predicting femoral strength in a stepwise regression model for both loading configurations. Similar relationships were observed for separate subtypes of femoral side impact fracture (cervical vs. trochanteric fractures; Table 1).

Table Table 1. Correlation of Femoral Failure Loads With Densitometric Data
original image
Figure FIG. 1.

Bar graphs showing coefficients of determination (r2) for DXA, QCT, pQCT (distal radius), and QUS (calcaneus), with mechanical failure loads at the hip, thoracolumbar spine, distal radius, and a combined index of failure at these three skeletal sites.

The correlation of the densitometric variables with spinal failure loads are displayed in Table 2 and Fig. 1. QCT and spinal DXA had similar capability of predicting thoraco-lumbar strength, whereas all other techniques and/or sites displayed significantly (p < 0.01) lower correlation coefficients. Likewise, forearm DXA and pQCT of the radius displayed significantly higher correlations with failure of the radius than non-site-specific DXA, QCT, and QUS (Table 2; Fig. 1).

Table Table 2. Correlation of Vertebral and Radial Failure Loads and of a Combined Failure Index of All Three Sites With Densitometric Data
original image

Most techniques showed similar capability in predicting a combined index of bone strength at the three most relevant sites (r2, approximately 50-60%; Table 2; Fig. 1), with only trochanteric DXA, pQCT at 4% (rather than 20%), and QUS at the calcaneus displaying significantly lower associations than the other methods and/or measurement sites. Results were similar when considering specimens with at least one or with all three failure loads. Specific geometric parameters from pQCT were not superior to the BMC, and a combined ultrasound index (stiffness index) was not superior to SOS and BUA in predicting bone strength. A stepwise regression model (including DXA at the three sites and calcaneal QUS as independent variables) revealed that the combination of spinal and radial DXA provided the best estimate of the combined strength index (adjusted r2 = 61%) and that QUS was unable to add significant independent information in this context.


In this paper we present a comprehensive analysis of the ability of standard and novel (peripheral) densitometric techniques to estimate bone strength at the hip and other relevant skeletal sites. In particular, these techniques are related, for the first time, to a combined index of mechanical strength at all clinically relevant sites, namely the hip, the thoracolumbar spine, and the distal radius. We specifically tested the hypotheses that pQCT at the distal radius and/or QUS at the calcaneus can serve as replacement or improvement of current methodology (DXA and QCT) and that QUS can provide significant additional information to site-specific DXA in predicting bone strength. The advantages of novel peripheral measurement techniques are that they are relatively inexpensive, easy to use, apply either none or only small amounts of radiation, and permit determination of properties of bone beyond those of bone mass and density. These properties include geometric parameters of the cortical bone (pQCT), and potentially, microstructural and mechanical components of the trabecular compartment (QUS).

Our results show that failure loads of the hip are most accurately predicted by femoral DXA, both in a side impact and in a vertical loading configuration. Measurements at the neck provided the best estimate among regions of interest at the proximal femur. Peripheral measurements with pQCT or QUS displayed lower correlation coefficients with mechanical strength of the hip, and QUS was unable to add significant additional information to site-specific analysis. These results clearly favor conventional DXA measurements at the proximal femur for estimating hip fracture risk.

Similar observations applied to the spine and distal radius, in which site-specific DXA and QCT/pQCT provided similar estimates of bone strength, whereas non-site-specific measurements were significantly inferior. When computing a combined index of failure at all three sites, most techniques showed similar performance, but DXA of the trochanter, pQCT at the 4% radial site, and calcaneal QUS displayed significantly lower correlation coefficients.

We have selected an experimental (biomechanical) study design, because bone strength cannot be objectively determined in vivo, because (prospective) clinical studies require excessive time intervals for a sufficient number of fractures to occur and because it is problematic to subject large sets of volunteers to radiographic measurements at multiple skeletal sites. Strengths of the present study include the large sample size, the fact that the femur was tested both in a side impact and in a vertical loading configuration, that a combined index of failure was computed for the most relevant skeletal sites, that DXA was performed in situ (accounting for in vivo soft tissue artifacts), and that most currently approved techniques were directly compared in the same sample. Potential limitations include the lack of detailed medical history and the use of fixed cadavers. In contrast to specimens obtained from pathological dissection, however, this sample did not include a preselection of highly pathological cases. By obtaining X-rays of all ROIs before mechanical testing and by performing a histomorphometric analysis of the iliac crest, we were able to apply exclusion criteria similar to those of clinical studies.(20) Regarding the effect of fixation, Edmonston et al.(31) concluded that mechanical strength of entire bones is only minimally affected and that the correlation between bone status and mechanical strength is unchanged. We have shown previously that prolonged (10-month) formalin fixation has no significant effect on DXA under in situ conditions(8) and that QUS maintains a linear relationship with values obtained before embalmment.(22) The densitometric measurements on the current sample were in the range of those reported in vivo,(32) and the failure loads and their correlation with DXA was similar to those reported in fresh specimens.(23–25, 27, 29, 30, 33) The fracture patterns observed (ratio between cervical and trochanteric fractures) resembled those seen clinically, and the proportion of nontypical fractures (25%) was similar to that obtained in other experimental studies.(25) This investigation should thus not be critically affected by fixation. Nevertheless, one should be aware that, although the in situ comparison of bone densitometry with mechanical failure loads in cadavers is likely similar to the in vivo relationship with bone strength, other nonskeletal factors may alter the relationship of bone strength and clinical fracture risk.

We have previously reported a low correlation of failure loads among different skeletal sites.(20) These findings suggest that either the acquisition or the loss of mechanical competence of the human skeleton is governed by substantial regional variation and does not seem to represent a strictly systemic process. This provides a fundamental problem when attempting to estimate mechanical competence at one site from densitometric measurements at another location. This is also reflected in this study, in which site-specific techniques performed consistently and significantly better than non-site-specific ones, irrespective of the physical method used.

In line with our results, Augat et al.(34) reported in a clinical case-control study that pQCT at the distal radius displayed lower ability to discriminate between patients with and without femoral fractures than femoral DXA. In an experimental study, pQCT at the peripheral lower extremity (femur and tibia) did not provide a better estimate of femoral failure loads than measurements at the distal radius.(35) Despite the importance of geometric cortical properties in determining bone strength,(26, 30) pQCT seems to be unable to reach the predictive ability of site-specific DXA in the hip and spine. pQCT was able to predict the combined index of mechanical strength as well as spinal QCT and DXA, but geometric properties were not superior to the BMC as measured with pQCT.

As in several previous experimental studies,(36, 37) calcaneal QUS was unable to provide comparable or even additional information to site-specific DXA. Although earlier prospective, clinical studies have indicated a high predictive ability of calcaneal QUS,(38, 39) the current data suggest that QUS is not adequate for estimating bone strength at other sites. This is also in line with more recent results of prospective studies,(40) and this seems to not only apply to the femur but also to a combined index of mechanical strength at the clinically most relevant skeletal sites.

In summary, approximately 55% of the variability of femoral bone strength in side impact and vertical loading are explained by site-specific (femoral) DXA, with no other technique being able to reach a similar degree of prediction. New peripheral measurement techniques (pQCT and QUS) thus seem unable to replace DXA for assessing bone strength as a surrogate of fracture risk in the hip or to predict some of the variability in bone strength, unexplained by site-specific bone mass. However, all methods, except for calcaneal QUS, displayed similar capability in predicting a combined index of mechanical strength at the hip, thoracolumbar spine, and distal radius. According to these experimental results, clinical assessment of femoral fracture risk should continue to rely on femoral DXA, whereas DXA, QCT, and pQCT seem to be equally suited for estimating a combined risk of fracture at the three clinically most relevant sites of osteoporotic fracture.


We thank Prof G Delling and coworkers (Osteopathologie, UKE Hamburg, Germany) for the histomorphometric diagnosis, Dr Jan Grimm (Diagnostische Radiologie, Kiel, Germany) for reading the spinal X-rays, and Borjana Barth, Gudrun Goldmann, Stephanie Kranz, Nadine Krefting, Markus Bachmeier, Dominik Bürklein, Oliver Groll (Anatomische Anstalt München), and Albert Albrecht (AO Research Institute Davos) for their help with the densitometry and biomechanical testing. This study was supported by Grant DFG LO 730 2-1 from the German Research Society.