Update: New techniques for assessing bone health

Center for Muscle and Bone Research
Charité Medical School
Free University of Berlin and Humboldt University of Berlin

New techniques for assessing bone health
by D. Felsenberg,Germany

Peripheral and vertebral fractures are the greatest hazard facing patients with osteoporosis, reducing their physical mobility, quality of life, and life expectancy. There is increasing evidence, supported by fundamental physics, that more than dual energy x-ray absorptiometry is required to estimate individual fracture risk and monitor treatment response. The determinants of bone strength are geometry, structural properties (bone distribution), material properties, and direction of force. It is therefore essential to develop and implement more sophisticated techniques such as in vivo microcomputed tomography. Scanco Medical AG’s peripheral quantitative computing tomography system, XtremeCT, is a relatively new device for the in vivo approximation of geometry, structural parameters, and distal radial and tibial density. Structural parameters include trabecular thickness, trabecular separation, structural model index, connectivity, anisotropy, and cortical thickness. Other calculations include bone volume/tissue volume ratios and subregional and cortical bone mineral densities, expressed in mg/cm3. Finite element analysis based on 3-D reconstructions of 110 slices is used for stress mapping. The technique’s main limitations are movement artifacts and the fact that calculation of the structural parameters is density-based.

Medicographia. 2010;32:429-434 (see French abstract on page 434)

Since bone is a multifunctional tissue, assessment of its health involves multiple parameters. This paper reviews the latest techniques for evaluating two of these parameters, bone strength and fracture risk, based on the material and structural properties of bone and the direction of force. Mass does not come into it, although physicians persist in ignoring the basic mechanics involved, using terms such as bone quality to simplify a complex system.

According to Dalzell et al,1 the material properties of bone cannot currently be studied noninvasively. Of course there is dual energy x-ray absorptiometry (DXA), which determines bone mineral density (BMD), but the latter is very different from actual physical density. DXA scanners mainly measure bone mass, which theWorld Health Organization (WHO) has deemed important for classifying the clinical impact of osteoporosis treatments. However, we know from the experience of major pivotal studies that DXA is of very limited use in monitoring treatment effect, largely because it fails to separate measures for trabecular and cortical bone. Other limitations include the absence of geometric data for calculating cross-sectional moment of inertia (CSMI) and of structural and material property data.

Figure 1
Figure 1. The stress/strain curve as a material property.

Strain is the relationship between change of length and original length. With glass higher stresses are needed to achieve the same strain as with bone or rubber.

Bone strength is dependent on a material property, expressed by the elastic (Young’s) modulus (E), and the CSMI. The elastic modulus is given by the slope of the stress/strain curve (stress on the y axis and strain on the x axis) during the linear elastic phase (Figure 1). It is constant for a given material and is expressed in N/mm2 (1 newton/ mm2=1 megapascal [MPa], and 1 kilonewton/mm2=1 gigapascal [GPa]) Bone has an E value of 18 to 21 GPa. For comparison, glass has a very steep slope, with an E value of 50 to 90 GPa, whereas the E value of silicone rubber is 0.01 to 0.1 GPa. In other words, the higher the E value the more rigid the material. However, E values also depend on temperature, humidity, and speed of deformation.

Fracture risk is a function not only of bone’s material properties, but also of its geometry and the direction of force. The CSMI reflects the dependence on geometry. These calculations are mostly important in long bones (humerus, radius, ulna, femur, tibia, fibula, femoral neck, etc) where the direction of force is not uniaxial, along the long axis, but in all the other directions for which the bone is not adapted. The CSMI depends mostly on the distance of the bone mass to the neutral surface: I =fy2ΔA (Figure 2). Bone strength, expressed as bending stiffness (EI), is given by the product of E and CSMI.

Peripheral computed tomography (pCT) systems perform these calculations routinely. Beck et al2 devised an interactive hip structure analysis (HSA) program that derives femoral neck geometry from raw bone mineral image data in order to estimate hip strength using single plane engineering stress analysis. The purpose of the program was to improve the predictive value of hip bone mineral data for osteoporosis fracture risk assessment. The authors reported a series of experiments with an aluminum phantom and cadaver femora designed to test the accuracy of derived geometric measurements and strength estimates. HSA-computed femoral neck cross-sectional areas (CSA) and CSMI on an aluminum phantom agreed closely with actual values (r>0.99). HSA-computed cross-sectional properties of three human cadavers were compared with measurements derived from sequential CT cross-sectional images. Discrepancy between the two methods averaged less than 10% along the length of the femoral neck. The breaking strengths of 20 femora showed better agreement with HSA-predicted strength (r=0.89) than with femoral neck BMD (r=0.79).2

Figure 2
Figure 2. Mechanics of bending.

Bending a bar or cylinder compresses the material on the inner side of the curve and tensions the material on the outer side. The one surface in the model
subject to neither compression nor tension is termed the neutral surface and is
more or less parallel to the axis of the bar or cylinder. A cylinder such as a long
bone increases its bending stiffness (with no change in mass) by distributing its
mass over an increasing diameter as far as possible from the neutral surface.
The integral power of two of all distances to all different cross-sectional areas
(pixel) is called cross-sectional moment of inertia.

It takes more sophisticated devices to calculate the trabecular network in vitro and answer another question of general interest: how does the most typical fracture in osteoporosis, vertebral compression fracture (often referred to as “sintering” in German-language osteoporosis literature, a term taken from metallurgy), relate to microarchitectural deterioration in other skeletal regions? Epidemiological studies have shown that osteoporotic (vertebral and nonvertebral) fracture incidence relates not only to vertebral fracture prevalence, but also to prevalent vertebral fracture severity, suggesting that vertebral fracture severity is a marker for increased bone fragility at all skeletal sites. Bone architecture, defined as the distribution of bone mass in a trabecular network, can be directly assessed by histomorphometry or microcomputed tomography (mCT) analyses of invasively obtained bone biopsy samples or by in vivo assessment of the distal forearm or distal tibia.

Since no in vivo measurements are available we have to focus on in vitro data to determine the relationship between vertebral compression fracture and microarchitectural deterioration elsewhere. Genant et al3 conducted a semiquantitative analysis of baseline vertebral fracture severity on spinal radiographs from 190 postmenopausal women with osteoporosis. Bone structure indices were obtained by 2-D histomorphometry and 3-D mCT analyses in transiliac bone biopsy samples taken at baseline in a subset of patients from the Multiple Outcomes of Raloxifene Evaluation trial (MORE)4 and the teriparatide Fracture Prevention Trial (FPT).5 After adjustment for age, height, and spinal DXA-BMD, there were significant trends for 3-D bone volume, trabecular number, trabecular separation, and connectivity density: mCT bone volume was significantly lower (P<0.05) in women with mild (-23%), moderate (–30%), and severe fractures (–51%) than in women with no fractures. Trabecular number was lower (P<0.05) in women withmild (–14%),moderate (–18%), and severe (–28%) vertebral fractures compared to women without vertebral fractures, while trabecular separation was higher (P<0.05) in those with mild (33%), moderate (42%), and severe (55%) vertebral fractures. These data show a clear relationship between vertebral fracture severity and microstructural deterioration in transiliac bone biopsies. The task for the future is to determine how closely these data match the structural deterioration of the distal forearm and tibia as assessed by in vitro mCT.

The device

The only in vivo mCT system currently available for human measurements is the XtremeCT (Scanco Medical, 8303 Bassersdorf, Switzerland [www.scanco.ch/systems-solutions /preclinical-systems/xtremect.html]). It scans the distal radius or tibia in 2.8 minutes, acquiring a 9 mm-high stack of 110 slices at a resolution of 82 ìm. Its ability to accommodate specimen sizes up to 150 mm (height)×126 mm (diameter) provides scope for clinical applications (Figures 3 and 4).

Figure 3
Figure 3. XtremeCT peripheral quantitative computed tomography scanner at the Charité Hospital, Benjamin Franklin Campus, Berlin.

The scanner can be used for both basic research (mouse, rat) and interventional clinical studies.

Figure 4
Figure 4. Microcomputed tomography (mCT) scan of a Russian cosmonaut at the Charité before leaving for the International Space Station (ISS).


Each measurement takes about 2.8 minutes, plus another couple of minutes for calculations that include analysis of the regional or subregional BMD data: total bone, cortical bone, trabecular bone, and some trabecular bone subregions (Figure 5, page 432). Measurement is standardized to a defined distance from the joint endplates of the radiocarpal junction.

During measurement the forearm or tibia is fixed in a cast. The structural parameters measured include trabecular number, trabecular thickness, trabecular separation (Figure 6), structural model index (Figure 7), connectivity, anisotropy, and cortical thickness (Figure 5). All structural parameter calculations are based on density measurements.

The mCT methodology has been used in bending tests in rats to determine the relevance of geometry and cortical thickness. It provides data that accurately describe cortical bone geometry and parallel cortical bone strength results obtained by the 3-point bending method. These data meet the criteria of providing quick, reproducible, and accurate answers regarding cortical bone geometry as a predictor of cortical bone strength.6

Reference values and reproducibility

The first population-based normative data for in vivo measurements of bone microstructure, published in 2006,7 were obtained using a prototype of the current mCT system. The results may therefore be less robust than subsequent data. The first reference data obtained with the current device were reported by Dalzell et al.1 In 2005, Boutroy et al8 published mCT precision values of 0.7% to 1.5% for total, trabecular, and cortical densities and 2.5% to 4.4% for trabecular architecture. Postmenopausal women had lower density, trabecular number, and cortical thickness than premenopausal women (P<0.001) at both radius and tibia. Osteoporotic women had lower density, cortical thickness, and increased trabecular separation than osteopenic women (P<0.01) at both sites. Furthermore, although spine and hip BMD were similar, fractured osteopenic women had lower trabecular density and more heterogeneous trabecular distribution (P<0.02) at the radius than nonfractured osteopenic women.

Figure 5
Figure 5. Microcomputed tomography (mCT) analysis of bone mineral density in the distal tibia.

The scan shows specific subregions and includes trabecular and cortical density. Cortical thickness is as readily assessable as cortical circumference.

Clinical applications

Recent clinical applications of in vivo mCT include space research (effect of bed rest) and therapeutic trials with various bisphosphonates (risedronate, ibandronate), denosumab, odanacatib, and strontium ranelate, most of which have just been completed and hence are only available as posters or abstracts.

Bed rest studies have measured the effect of weightlessness on bone density and structure in young healthy female and male volunteers. In the Women International Space simulation for Exploration (WISE) study, subjects remained in bed at 6° head down tilt for 60 days.9 ThemCT data showed a clear tendency to a decline in all structural parameters and an increase in trabecular separation, but at 60 days the values did not differ significantly from baseline.We found no significant differences between the control group (no exercise), the exercise group (resistive exercise plus endurance training), and the nutrition group (specific amino acid-enriched diet). In the Berlin Bed Rest-2 (BBR2) study, mCT revealed significant tibial cortex loss in the control group.10 The bone loss observed in both studies showed high interindividual variation, with no conclusive pattern. A randomized double-blind prospective study compared strontium ranelate (SrR) and alendronate (70 mg once weekly) in postmenopausal women with osteoporosis (spine and/or total hip T-scores ≤–2.5 SD) over 24 months.11 Preplanned interim analysis of the mCT data at 12 months documented a 5.3% increase in cortical thickness in the SrR group compared to the alendronate group, which did not differ from baseline (P=0.001). The blood volume/tissue volume (BV/TV) ratio increased by 2.1% over baseline compared with the alendronate group (P=0.002), which again did not differ significantly from baseline.

Figure 6
Figure 6. Synchrotron computed tomography image of the trabecular network in a vertebra.

Trabecular distance (50 μ to 200 μ) and trabecular thickness (150 μ to 600 μ) in a healthy subject (red arrows). (Photo courtesy of Felsenberg and Giehl.)

Figure 7
Figure 7. The structural model index reflects the relationship between plates (*) and rods (+) in trabecular bone.

The increasing number of rods is typical of bone loss and indicates increasing plate resorption. (Photo courtesy of Felsenberg, Giehl and Ritter.)

At 24 months, cortical thickness increased by 6.3% (±9.5%) in the SrR group and by 0.9% (±6.2%) in the alendronate group (P=0.004). The comparative increases in BV/TV ratio [2.5% (±5.1%) vs 0.8% (±3.8%)] also differed significantly in favor of SrR (P=0.040), as did those in trabecular and cortical BMD: 2.5% (±5.1%) vs 0.9% (±4.0%) (P=0.048) and 1.4% (±2.8%) vs 0.7% (±2.1%) (P=0.045). The marker of bone formation, bone alkaline phosphatase, showed an 18% increase over baseline (P<0.001), while the marker of bone resorption, serum C-telopeptide crosslinked collagen type I, decreased –16% vs baseline (P=0.002), thereby confirming the dual mode of action of SrR. The results point to significant structural benefit in the distal tibia in women with postmenopausal osteoporosis treated for 2 years with SrR compared to alendronate.

Finite element analysis

Scanco provides specific finite element analysis (FEA) software for their image format. (FEA is a mathematical technique that originated from the need to solve complex elasticity and structural problems in civil and aeronautical engineering.) It is used to simulate tests and measure mechanical and elasticity properties such as stiffness, estimated failure load, trabecular/ cortical load distribution, and changes in mechanical properties. For example, the “von Mises stress distribution calculation” generates important data about stress risers in the trabecular network, sites of treatment effect, and increases/ decreases in structural strength (Figure 8).1,12

Figure 8
Figure 8. Microcomputed tomography (mCT) images of a vertebra: (A) Stress areas; (B) Stress distribution.

(A) mCT scan of a vertebral body. Stress regions are color-coded (high: red; low: yellow and green). High stress areas are seen in thin trabeculae and low stress areas in plate-like structures. (B) Stress distribution in the entire vertebral body with endplate deformation. High stress areas are seen in midvertebral structures concentrated on endplate cortical bone. (Both images by courtesy of Scanco Medical, Switzerland.)

The Pros and Cons

_ Pros
In vivo mCT determines the structural and material properties of peripheral bone at the cost of acceptable whole-body radiation (<15 μSv) within a few minutes, image processing included. The resulting data extend beyond DXA-BMD measurement and provide an estimate of bone strength that is grounded in fundamental mechanics. The mCT images can be processed by FEA to simulate mechanical testing and derive estimates of stress distribution and failure load. The technique measures cortical bone separately from trabecular bone.

Physical density (mass/volume) reflects the material properties of bone. However, we are not yet able to match structural information to mechanical strength tests.

Figure 9
Figure 9. Microcomputed tomography (mCT) scan of the distal radius in an interventional study (A-C).

(A & B) Follow-up scans showing grade 2 movement artifacts in the second scan with cortical discontinuity in places (red arrows) and multiple contour-blurring horizontal lines (yellow arrows). (C) Severe grade 3 movement artifacts caused the scan to be discarded.

_ Cons
The local radiation dose is quite high. In the event of procedural error (wrong positioning, movement artifacts, etc), measurement can be repeated only twice, to a total of three measurements of the same region. Despite the short scan time (2.8 minutes), the very high resolution, visualizing structures down to 82 μm in diameter, produces multiple movement artifacts (MA) (Figure 7). We have identified MA in 38% of a total of several thousand mCT scans, 79% of which were in the forearm and 21% in the tibia. Even after repeating the forearm measurements twice, only 40% were MA-free. An MA grading system may help to increase the quality of forearm analyses. We have detected no pattern to the MA seen in repeated measurements. The problem is less evident with the tibia but still present. It can be decreased by shortening the scan time even further.

Another limitation is the peripheral measurement region. To measure bending stiffness together with cortical thickness and density, measurements should ideally be taken at the midshaft of radius and tibia. But the design of the device allows only very distal measurements. Reference values are not yet robust because of the relatively few normal subjects who have been scanned. A final limitation is that most of the programs for analyzing structural parameters are density based (Figure 9, page 433). _

1. Dalzell N, Kaptoge S, Morris N, et al. Bone micro-architecture and determinants of strength in the radius and tibia: age-related changes in a population based study of normal adults measured with high-resolution pQCT. Osteoporos Int. 2009;20(10):1683-1694.
2. Beck TJ, Ruff CB, Warden KE, Scott WW Jr, Rao GU. Predicting femoral neck strength from bone mineral data. A structural approach. Invest Radiol. 1990; 25(1):6-18.
3. Genant HK, Delmas PD, Chen P, et al. Severity of vertebral fracture reflects deterioration of bone microarchitecture. Osteoporos Int. 2007;18(1):69-76.
4. Ettinger B, Black DM, Mitlak BH, et al. Reduction of vertebral fracture risk in postmenopausal women with osteoporosis treated with raloxifene: results from a 3-year randomized clinical trial. Multiple Outcomes of Raloxifene Evaluation (MORE) Investigators. JAMA. 1999;282(7):637-645.
5. Neer RM, Arnaud CD, Zanchetta JR, et al. Effect of parathyroid hormone (1-34) on fractures and bone mineral density in postmenopausal women with osteoporosis. N Engl J Med. 2001;344(19):1434-1441.
6. Bagi CM, Andresen C, Pero R, Lariviere R, Turner CH, Laib A. The use of micro- CT to evaluate cortical bone geometry and strength in nude rats: correlation with mechanical testing, pQCT and DXA. Bone. 2006;38(1):136-144.
7. Khosla S, Riggs BL, Atkinson EJ, et al. Effects of sex and age on bone microstructure at the ultradistal radius: a population-based noninvasive in vivo assessment. J Bone Miner Res. 2006;21(1):124-131.
8. Boutroy S, Bouxsein ML, Munoz F, Delmas PD. In vivo assessment of trabecular bone microarchitecture by high-resolution peripheral quantitative computed tomography. J Clin Endocrinol Metab. 2005;90(12):6508-6515.
9. Defontaine M, Nasser-Eddin M, Rittweger J, Lazerges M. A 60 days Bed Rest study: preliminary QUS BUA changes at the calcaneus site. Ultrasonics Symposium, 2005 IEEE, Sept 18-21 2005;4:2014-2017.
10. Felsenberg D. Results of Berlin BedRest Study II. Personal communication. 2010.
11. Reginster JY, Felsenberg D, Boonen S, et al. Effects of long-term strontium ranelate treatment on the risk of nonvertebral and vertebral fractures in postmenopausal osteoporosis: Results of a five-year, randomized, placebo-controlled trial. Arthritis Rheum. 2008;58(6):1687-1695.
12. Boutroy S, van Rietbergen B, Sornay-Rendu E, Munoz F, Bouxsein ML, Delmas PD. Finite element analysis based on in vivo HR-pQCT images of the distal radius is associated with wrist fractures in postmenopausal women. J Bone Miner Res. 2008;23:392-399.

Keywords: osteoporosis; fracture; vertebral fracture; fracture risk; bone strength; dual-energy x-ray absorptiometry; microcomputed tomography; peripheral quantitative computing tomography