Trabecular Rod Thickness by Direct Measurement From 3D SEM Anaglyphs

Transcription

1 THE ANATOMICAL RECORD PART A 271A: (2003) Trabecular Rod Thickness by Direct Measurement From 3D SEM Anaglyphs VINOTHA VIJAYAPALAN, 1 PETER SUTTON-SMITH, 1,2 IAN H. PARKINSON, 1,2 * R. BRUCE MARTIN, 3 AND NICOLA L. FAZZALARI 1,2 1 Division of Tissue Pathology, Institute of Medical and Veterinary Science, Adelaide, Australia 2 Department of Pathology, University of Adelaide, Adelaide, South Australia 3 Orthopaedic Research Laboratories, University of California at Davis Medical Center, Sacramento, California ABSTRACT This study presents a methodology for measuring the thickness of trabecular rods directly from anaglyphs. Macerated sagittal slices of T12 vertebral bodies from 15 subjects were examined by scanning electron microscopy (SEM). Two digital images (the second image tilted 5 ) were recorded, and a 3D anaglyph was created. The thickness of the trabecular rods (Tb.Th (rods) ), and the anatomical orientation of the trabecular rods were measured using an image analyser. Conventional 2D histomorphometry was performed on adjacent bone slices. A total of 1,559 rod measurements were made from the 15 vertebral bone slices, with a mean Tb.Th (rods) of m. The rod thickness in males ( m) was significantly greater than that in females ( m, P 0.001). Tb.Th (rods) changed significantly with age in the males: the thicker rods in the younger men reduced with age to a thickness similar to that in women. 3D measurements were significantly larger than the 2D estimates, and there was no correlation between the two methods of measurement. An inverse correlation was found between the number of rods and the bone volume fraction (BV/TV), indicating that decreased BV/TV is associated with an increased number of rods. The vertical rods ( m) were significantly thicker than the horizontal rods ( m, P 0.001). The determination of rod numbers, and their orientation and individual thicknesses enables a greater understanding of cancellous bone architecture in both individuals and populations, and will allow more reliable finite element modelling. Direct measurements from 3D anaglyphs of intact specimens provide new data that show previously unrecognised age- and sex-related changes. Anat Rec Part A 271A: , Wiley-Liss, Inc. Key words: trabecular thickness; trabecular orientation; trabecular rods; 3D anaglyphs; vertebral body bone The accurate measurement of trabecular thickness is essential to elucidate the biomechanics of cancellous bone structure. Histological sections have been used to estimate trabecular thickness (Tb.Th) using the plate or rod model of trabecular bone architecture (Parfitt et al., 1987). This model-based histologic method has shown that changes in trabecular architecture occur with age and/or disease. In particular, trabecular bone loss has been shown to involve both trabecular thinning and increased trabecular separation (Ding and Hvid, 2000). These findings support the hypothesis that trabecular bone loss with age is due to a reduction in the number of trabecular elements and/or a thinning of these elements (Mosekilde, 1988; Jayasinghe et al., 1994; Amling et al., 1994). The derived parameters from 2D section-based mathematical modelling do not provide information about the 3D variability of individual trabecular elements. The increased focus on determining the mechanical properties of *Correspondence to: Ian Parkinson, Division of Tissue Pathology, Institute of Medical and Veterinary Science, Frome Road, Adelaide, South Australia 5000, Australia. Fax: ian.parkinson@imvs.sa.gov.au Received 16 October 2001; Accepted 2 November 2002 DOI /ar.a Published online 7 March 2003 in Wiley InterScience ( WILEY-LISS, INC.

2 MEASUREMENT OF TRABECULAR THICKNESS 287 cancellous bone from finite element models relies on estimates for the distribution of trabecular thickness. Frost (1999) discussed the need for more precise and accurate measurement of trabecular elements so that the mechanical effect of changes in trabecular dimensions can be determined. The importance of knowing the distribution of trabecular measurements is highlighted by the finding that increased variability in trabecular thickness is associated with increased risk of failure in the trabecular network (Yeh and Keaveny, 1999). Specialised bone imaging techniques, such as computed tomography (CT) and magnetic resonance imaging (MRI), can provide some in vivo information on the variability of trabecular thickness (Hipp et al., 1996; Link et al., 1998). However, only in vitro techniques and micro-ct are capable of providing the resolution necessary to accurately resolve trabecular structures (Ding and Hvid, 2000). Whilst the data provided by these techniques, such as the mean and standard deviation (SD), are valuable, only overall trabecular dimensions are measured. To date, the classification of trabecular elements into plates or rods for the analysis of thickness distribution using 3D microscopybased morphological methods has not been reported. Measuring trabecular thickness directly from projected images of intact specimens can provide detailed information on individual trabecular elements. These direct measurements do not require derivation or calculation from a mathematical model, because the projected length on a computer screen (appropriately calibrated) is a real measurement. Trabecular dimensions have been reported as indirect measurements from whole macerated bone slices (Amling et al., 1994), as direct measurements from 2D imaging of embedded bone slices (Mosekilde, 1988), and as derived parameters from computerised 3D reconstructions (Borah et al., 2000). The dimensions obtained from direct measurements are up to 50% larger than those obtained from 2D-histomorphometric analysis using the plate model of trabecular architecture (Day et al., 2000). This implies that the model of trabecular architecture is limited, and that accurate trabecular dimensions are yet to be determined. Scanning electron microscopy (SEM) of cancellous bone provides high-resolution images with a large depth of field. Electron beam imaging provides an ideal means of visualising the fine detail of macerated bone slices. The image of a specimen positioned in the microscope at a large working distance has a depth of field much greater than that obtained with the light microscope. This combination of characteristics allows detailed SEM examination of trabecular structure to a depth of 2 or 3 mm without having to adjust the focus. Construction of 3D anaglyphs provides a true stereo image. Anaglyphs are composite steroscopic images viewed in superimposed complementary colours. Humans have stereo vision, which relies on the transformation by the brain of differential information from each eye into 3D images. 3D objects that are visualised from a computer screen or book are only 2D projections of these objects. The viewer does not see the third dimension. Constructing 3D anaglyphs of SEM images of bone allows the eye brain interface to function normally, and the trabecular plates and rods, and their orientations, can be visualised more clearly compared to other methods (Russ, 1990). The aim of this study was to assess a technique to reliably measure trabecular thickness and anatomical orientation of rod-like structures in vertebral cancellous bone, using 3D anaglyphs of SEM images. This technique can determine the distribution of trabecular thicknesses in individuals data that is not available from conventional 2D histomorphometric analysis. METHODS Macerated sagittal slices of T12 vertebral bodies from 15 subjects (seven females, years old, mean years; and eight males, years old, mean years) were examined in a JEOL JSM-5300 scanning electron microscope. These vertebral bodies were retrieved at necropsy from patients with no history of bonerelated disease or medication, which would influence bone structure. A 5-mm-thick sagittal slice was taken from the whole vertebrae with an Isomet low-speed saw. The slice was then macerated by boiling it in sodium hydroxide, bleaching it in hydrogen peroxide, and soaking it in petroleum spirit to remove residual fat. The maceration process uses highly alkaline solutions, which do not remove bone mineral (and hence no evidence of surface etching is revealed on SEM). The specimens were chosen to encompass the adult life span of both sexes. Nine contiguous fields (each 35 mm 2, totalling 315 mm 2, and covering most of the sagittal slice cancellous bone area) were imaged at 15 magnification. Two 1,024 * 768 pixel digital images (the second image tilted through 5 ) were recorded for each region, and a 3D anaglyph was created on the computer screen for viewing with red/green stereo glasses. The bone sample was placed in the SEM at a large (45 mm) working distance to increase the depth of field, and the images were focused 1 mm below the cut surface. The working magnification was determined using a 3-mm transmission electron microscope copper grid for calibration. The distortion of the microscope image through the focal plane was 5%. The thickness of the trabecular rods (Tb.Th (rods) ), and the angle of the trabecular rods with respect to the plane of the vertebral body end-plate were measured using a Quantimet 500MC image analyser (Leica, Cambridge, UK). At an image analyser resolution of 7.8 m/pixel, each anaglyph was displayed on the screen with a randomly applied grid overlay. The image was viewed with red/ green stereo glasses, and the thickness of each trabecular rod, which intercepted a grid line, was measured by drawing a line across each rod at its intersection with the grid line, perpendicular to its long axis. The perspective of the 3D image was used to identify the rods, but the measurements were of the 2D projections of each rod. The 3D anaglyphs enabled clear identification of trabecular rods as smooth, roughly cylindrical structures, with a length at least three times their width (Fig. 1). The overall rod population statistics were calculated by pooling the measurement data from all the specimens. Rod number was represented by the number of measurements performed. Correlation analysis was performed for the mean measurement, Tb.Th (rods), for each individual (Table 1) with Tb.Th and BV/TV from conventional 2D histomorphometry. Sex differences were determined by Student s t-test, and age-related changes were determined by least-squares linear regression analysis. All statistical tests were performed with Microsoft Excel (Microsoft Corp., Seattle, WA), and statistical significance was accepted at P 0.05.

3 288 VIJAYAPALAN ET AL. Fig. 1. SEM image ( 15) overlaid with randomly applied sampling grid, with lines indicating measurement of trabecular rod thickness. The rods were classified as having horizontal or vertical orientation based on the projected angle of the orientation on the plane of the image. For analysis, all specimens were orientated such that the vertebral body end-plate was aligned with the horizontal scan axis of the image analyser video monitor. Horizontal rods were assigned an angle of orientation between 0 and 45, and vertical rods were assigned an angle of orientation between 45 and 90. Consistent with a recent digital radiographic analysis of vertebral bone structure (Caldwell et al., 1995), 13% of the trabeculae had an oblique orientation (30 60 ), which confirms that the trabecular rods have a predominantly orthogonal architecture. RESULTS Figure 1 shows a conventional SEM image of trabecular bone with a grid overlay. Lines delineating trabecular rod thickness are drawn across some of the rods. 1 A total of 1,559 rod measurements were made from the 15 vertebral bone slices, with a mean Tb.Th (rods) of m. Rod thickness in males ( m (from 660 measurements)) is significantly (P 0.001) greater than that in females ( m (from 899 measurements)). Table 1 shows the summary statistics for the 15 individuals, and includes the Tb.Th (rods) distribution data (mean SD), range, and the number of trabeculae measured for each vertebral slice. The minimum Tb.Th (rods) was 80 m for all subjects, and seven of the 15 had a minimum of 55 m. BV/TV and Tb.Th results from the conventional 2D histomorphometry are also shown in Table 1. 1 To see an anaglyph in 3D with red/green stereo glasses, go to our web site at Figure 2 shows the relation between Tb.Th (rods) and age for both sexes. Tb.Th (rods) did not change with age in the seven female subjects. However, Tb.Th (rods) decreased significantly with age in the male population, and Tb.Th (rods) in older men was similar to that in females. A comparison of the rod-thickness data with the morphometry estimates based on Parfitt et al. s (1987) plate model reveals that the 3D measurements are significantly larger than the 2D estimates ( vs , respectively; P ), and there is no statistical correlation between the two methods of measurement (P 0.28). There is a trend for the rod number to increase with age. However, an inverse correlation exists between the number of rods and the 2D bone volume estimates (Fig. 3), indicating that decreased BV/TV is associated with an increased number of rods. Trabecular rods were classified into two groups: horizontal ( 45 ) and vertical ( 45 ), with respect to the horizontal scan axis of the image analyser video monitor. Table 2 shows the summary statistics when the data are subdivided into horizontal and vertical rods. Vertical rods are significantly thicker than the horizontal rods in the pooled male and female groups (P 0.001). Whilst neither the vertical nor the horizontal rods in females showed age-related changes, the horizontal and vertical rods both decreased significantly in thickness with age in the male group (Fig. 4). These decreases in horizontal and vertical rod thickness occurred at the same rate (slope of each regression is statistically the same: P 0.6). DISCUSSION This study presents a methodology for the measurement of trabecular rod thicknesses directly from projected images. The precise visualisation of trabecular rods afforded by the 3D anaglyph allows rods to be unambiguously identified. The 3D perspective also facilitates accurate recording of the anatomical orientation of the trabecular rods, thereby enabling their classification as either horizontal or vertical (Fig. 1). This is relevant because, according to our current understanding, vertebral body cancellous bone is composed predominantly of vertical trabeculae with interconnecting horizontal trabecular rods. Trabecular rod thickness distributions are presented for each individual, male and female groups, and the whole sample. Overall, Tb.Th (rods) is greater in males than in females. However, only younger males ( 50 years) have thicker rods, which decrease significantly with age (Fig. 2). The measurements in older males ( 50 years) are similar to those in the female group, which did not change with menopause. Vertical trabecular rods are thicker than horizontal rods, and may reflect the plate origin of these rods. The dimensions of the structural elements in cancellous bone are influenced by the mechanical environment, and therefore horizontal and vertical elements may have different structural roles. The vertical trabecular elements are aligned with the principal load-bearing axis and are subject to compressive forces. Horizontal trabecular elements provide support to the more robust vertical trabeculae, and the interconnections ensure that the length of the unsupported vertical trabeculae is minimised, which helps prevent failure caused by buckling. In males, the thickness of both the horizontal and vertical trabecular rods decreases with age. It has been shown that bone loss in the iliac crest and vertebral bodies

4 MEASUREMENT OF TRABECULAR THICKNESS TABLE 1. Descriptive statistics for each individual of measurements from 3D anaglyphs Anaglyphs Histology Age/Sex Mean SD ( m) N Min Max ( m) CV BV/TV (%) Tb.Th ( m) 17F F F F F F F M M M M M M M M Mean BV/TV and mean Tb.Th for each individual measured from histological sections. 289 Fig. 2. Scatter plot of age vs. Tb.Th (rods) for both sexes, with regression line showing a statistically significant decrease in Tb.Th (rods) with age for the males (Tb.Th (rods) 0.53* Age 155.9, r 0.77, P 0.02). F males; E females. Fig. 3. Scatter plot of number of rods vs. BV/TV, with regression line showing decreasing BV/TV with increase in number of rods with age (BV/TV 0.32* Rods 12.9, r 0.54, P 0.04). of males is predominantly due to thinning of trabeculae, whereas females tend to lose whole trabecular elements as a consequence of trabecular perforation (Aaron et al., 1985; Mellish et al., 1989). This is consistent with the current observation that the thicker trabeculae in males became thinner with age. The trabeculae thickness in females did not change with age, and was similar to that in older males. The lower limit of thickness of trabecular rods appears to be bounded. The thinnest Tb.Th (rods) was 55 m, which is consistent with reported observations from histological sections (Wakamatsu and Sissons, 1969; Parfitt, 1987). No very thin trabeculae or entirely disconnected fragments were seen. This suggests that a single resorption episode is capable of removing at least 55 m of bone so that thinner elements are perforated and removed. This is consistent with a previously estimated bone resorption depth of 60 m in normal bone (Ericksen et al., 1985). There was no correlation between Tb.Th (rods) and Tb.Th as estimated by conventional histomorphometric analysis, and the magnitude of Tb.Th (rods) was greater than that of Tb.Th, as calculated from 2D sections. Day et al. (2000) found a very good correlation between the 2D and 3D estimates of trabecular rod thickness (r , P 0.001); however, Zielinski and Karnicki (2000) did not. The magnitudes of Tb.Th (rods) determined in the present study are similar to those reported in the above-cited studies; however, differences in image acquisition, and segmentation between groups may account for the discrepancy in the results. Also, estimates of Tb.Th are derived from sections that are much thinner than the trabecular rods. This results in tangential trabecular profiles that are not representative of true trabecular thickness. In addition, measurements of perimeter from sections of cancellous bone are prone to overestimation, which results in underestimation of Tb.Th (Parkinson and Fazzalari, 1994).

5 290 VIJAYAPALAN ET AL. TABLE 2. Group statistics for horizontal and vertical trabecular rods Class Tb.Th(rods) mean SD ( m) Fig. 4. Scatter plot of age vs. Tb.Th (rods) of horizontal and vertical rods in males, with regression lines showing a statistically significant decrease in Tb.Th (rods) with age for both horizontal and vertical rods (Tb.Th (rods) 0.44* Age 147, r 0.71, P 0.05; and Tb.Th (rods) 0.48* Age 161, r 0.79, P 0.01, respectively). F horizontal; E vertical. The cancellous bone becomes more rod-like as bone volume decreases (Fig. 3). This data is consistent with the findings of Day et al. (2000), who reported that samples with low BV/TV are more rod-like and plate-like as BV/TV increases. These observations are consistent with the idea that rods are created as trabecular plates perforate, modulated by a net imbalance in bone remodelling that results in decreased bone volume. The present study describes a method that uses conventional SEM facilities and yields values for trabecular thickness that are more precise and accurate compared to methods that use 2D images. The determination of rod number, orientation, and individual thickness enables a greater understanding of cancellous bone architecture in both individuals and populations. The ability to produce distributions of measurements for trabecular structures opens the way for analysis, which heretofore has not been possible. Finite element models of cancellous structure have shown theoretically that increased variability in trabecular dimensions results in increased risk of fracture of the cancellous structure (Yeh and Keaveny, 1999). The ability to differentiate between horizontal and vertical rods further increases the sensitivity of the analysis, which will be of great benefit to biomechanical modelling studies. The data show that this method of measuring trabecular thickness directly from 3D anaglyphs of intact N Min Max ( m) Total Horizontal Vertical a Male Horizontal Vertical a Female Horizontal Vertical a a Significantly different to horizontal, P vertebral bone has sufficient sensitivity to detect gender dimorphism in age-related changes in the cancellous architecture. Techniques that enable more accurate measurement of trabecular structures may aid the identification of individuals at risk of bone fracture. We hypothesise that the distribution of trabecular thickness in the vertebral body will add a hitherto inaccessible dimension to the mass-based paradigm of osteoporotic fracture risk. LITERATURE CITED Aaron JE, Makins NB, Sagreiya K The microanatomy of trabecular bone loss in normal againg men and women. Clin Orthop Rel Res 215: Amling M, Grote HJ, Posl M, Hahn M, Delling G Polyostotic heteregeneity of the spine in osteoporosis. Quantitative analysis and three dimensional morphology. Bone Miner 27: Borah B, Dufresne TE, Cockman MD, Gross GJ, Sod EW, Myers WR, Combs KS, Higgins RE, Pierce SA, Stevens ML Evaluation of changes in trabecular bone architecture and mechanical properties of minipig vertebrae by three-dimensional magnetic resonance microimaging and finite element modeling. J Bone Miner Res 15: Caldwell CB, Willett K, Cuncins AV, Hearn TC Characterisation of vertebral strength using digital radiographic analysis of bone structure. Med Phys 22: Day JS, Ding M, Odgaard A, Sumner DR, Hvid I, Weinans H Parallel plate model for trabecular bone exhibits volume fractiondependent bias. Bone 27: Ding M, Hvid I Quantification of age-related changes in the structure model type and trabecular thickness of human tibial cancellous bone. Bone 26: Ericksen EF, Mosekilde L, Melsen F Trabecular bone resorption depth decreases with age: differences between normal males and females. Bone 6: Frost HM On the trabecular thickness -number problem. J Bone Miner Res 14: Hipp JA, Jansujwicz A, Simmons CA, Snyder BD Trabecular bone morphology from micro-magnetic resonance imaging. J Bone Miner Res 11: Jayasinghe JAP, Jones SJ, Boyde A Three-dimensional photographic study of cancellous bone in human fourth lumbar vertebral bodies. Anat Embryol 189: Link TM, Majumdar S, August P, Lin JC, Newitt D, Lu Y, Lane NE, Genant HK In vivo high resolution MRI of the calcaneus: differences in trabecular structure in osteoporosis patients. J Bone Miner Res 13: Mellish RWE, Garrahan NJ, Compston JE Age-related changes in trabecular width and spacing in human iliac crest biopsies. Bone Miner 6: Mosekilde L Age-related changes in vertebral trabecular bone architecture assessed by a new method. Bone 9: Parfitt AM Trabecular bone architecture in the pathogenesis and prevention of fracture. Am J Med 82(Suppl B): Parfitt AM, Drezner ML, Glorieux FH, Kanis JA, Malluche H, Meunier PJ, Ott SM, Recker RR Bone histomorphometry: standardization of nomenclature, symbols, and units. J Bone Miner Res 2: Parkinson IH, Fazzalari NL Cancellous bone structure analysis using image analysis. Aust Phys Eng Sci Med 417: Russ JC Stereoscopy. Computer-assisted microscopy. In: Russ JC, editor. The measurement and analysis of images. New York: Plenum Press. p Wakamatsu E, Sissons HA The cancellous bone of the iliac crest. Calcif Tissue Res 4: Yeh OC, Keaveny TM Biomechanical effects of intraspecimen variations in trabecular architecture: a three-dimensional finite element study. Bone 25: Zielinski KW, Karnicki F Comparison of radiometric and histometric features of bone samples from ribs. Anal Quant Cytol Histol 22: