- Email: [email protected]
Trabecular architecture in women and men of similar bone mass with and without vertebral fracture: II. three-dimensional histology
Bone Vol. 27, No. 2 August 2000:277–282
Trabecular Architecture in Women and Men of Similar Bone Mass With and Without Vertebral Fracture: II. Three-Dimensional Histology J. E. AARON,1 P. A. SHORE,1 R. C. SHORE,2 M. BENETON,3 and J. A. KANIS3 1
School of Biomedical Sciences, Worsley Medical and Dental Building, University of Leeds, Leeds, UK Division of Oral Biology, Dental Institute, University of Leeds, Leeds, UK 3 Centre for Metabolic Bone Diseases at Sheffield (WHO Collaborating Centre), Medical School, University of Sheffield, Sheffield, UK 2
Introduction We recently developed a simple and inexpensive method that complements established bone histomorphometry procedures by enabling the two-dimensional imaging of cancellous bone to be viewed within its three-dimensional context with the marrow tissue in place and without detriment to the material for other histological purposes. The method, based on the preparation and superficial staining of slices 300 m thick, enables “real” (i.e., unstained) trabecular termini to be separated from “artifactual” (i.e., stained) termini, providing a direct measure of cancellous connectivity in osteopenic bone. The technique was applied to osteopenic age-matched, white, postmenopausal women (31 with and 22 without vertebral compression fractures) with a similar bone status, as measured at the spine by absorptiometry and at the iliac crest by histology (see part I of this study). Despite the similarity in the mass of trabecular bone at either site, the results showed a significant difference (p < 0.05) in the number of “real” trabecular termini between the groups, such that the fracture group had almost four times as many termini (mean ⴞ SE: 1.98 ⴞ 0.51/30 mm2) at the iliac crest as the nonfracture group (mean ⴞ SE: 0.53 ⴞ 0.31/30 mm2). Previous histomorphometry of the same material failed to detect a structural distinction between the two groups using established variables. It was concluded that a mass-independent trabecular discontinuity contributes to skeletal failure and that determination of the number of “real” disconnections (i.e., unstained termini) by the direct method proposed may provide a more sensitive discriminant of fracture than the present indirect procedures. A group of fracture and nonfracture men (see part I) suggested a similar distinction (fracture: 0.69 ⴞ 0.30/30 mm2; nonfracture: 0.18 ⴞ 0.18/30 mm2), although the difference was not significant. (Bone 27: 277–282; 2000) © 2000 by Elsevier Science Inc. All rights reserved.
It is well established that subjects with osteoporotic fractures tend to have lower mean bone mass than those without, whether their bone status is determined by noninvasive means or histologically using iliac crest bone biopsies. However, separation on the basis of bone mass alone is incomplete; that is, a substantial overlap in the data indicates that some subjects with an apparently adequate bone status sustain osteoporotic fractures, whereas others with an apparently inadequate bone status do not. Many other factors that contribute to atraumatic fracture have been recorded in the literature. Prominent among these factors is the trabecular architecture and, in particular, trabecular connectivity, because a well-interconnected structure is more resistant to bending forces than one that is not. A number of mathematical procedures have been applied to the measurement of trabecular connectivity indirectly in order to overcome the well-recognized intrinsic problem that, in thin histological sections, apparently disconnected bars may derive entirely from the plane of section. This means that many of the “apparent” trabecular termini in the two-dimensional (2D) histological image are artifactual. Analysis of the three-dimensional (3D) image evident in unsectioned material may be expected to overcome the difficulty; however, such images tend to consist of superimposed and complex trabecular networks that are not easy to evaluate. Moreover, removal of marrow tissue is invariably a requirement for a well-delineated image. This will be achieved at the cost of any skeletal elements that are detached from the main trabecular framework and which will disappear together with the eluted marrow cells. Although this is unlikely to affect the substantial interconnected framework characteristic of young healthy bone, it may affect the appraisal of elderly, disconnected, fragile, and osteopenic material (see, e.g., Kinney et al.12). A satisfactory compromise for analysis may lie somewhere between the thin histological section and the gross sawn segment favored by Atkinson5 and Amling et al.4 Their work serves to illustrate the potential a simple method might have, particularly if it was also inexpensive, nondisruptive to soft tissues, and applicable to small samples such as transiliac bone biopsies. Recently, such a method has been developed and which is not destructive for other histological purposes. On the contrary, it complements established histomorphometrical procedure by combining the 2D and 3D images and providing essential insight into the third dimension. It is based on the differential surface staining of thick, optically acceptable slices of cancellous bone.14
Key Words: Bone histology; Trabecular termini; Iliac crest biopsy; Two-dimensional (2D) and three-dimensional (3D) image analysis; Vertebral fracture; Osteoporosis.
Address for correspondence and reprints: Dr. Jean E. Aaron, School of Biomedical Sciences, Worsley Medical and Dental Building, University of Leeds, Leeds LS2 9JT, UK. E-mail: [email protected]@leeds.ac.uk © 2000 by Elsevier Science Inc. All rights reserved.
277
8756-3282/00/$20.00 PII S8756-3282(00)00328-8
278
J. E. Aaron et al. Three-dimensional histology and vertebral fracture
Figure 1. Diagram illustrating apparent (*) and real (⫹) trabecular termini and an “apparent” (ApIs) and a “real” (ReIs) island (isolated trabeculae) in a histological section (10 m thick; 2D image) and a slice (300 m; 3D image) of cancellous bone.
The present study applies this novel technique to the comparison of trabecular connectivity in transiliac biopsies from two previously analyzed groups of postmenopausal women11 with matched bone mass and with and without minimal trauma vertebral compression fractures. Earlier histomorphological examinations of 8-m-thick sections have been unable to demonstrate any significant structural difference between them. Methods The material was the same as that used in part I of this study.11 Transiliac bone biopsies, 8 mm in diameter, were taken from 31 postmenopausal white women (mean ⫾ SD: 66 ⫾ 9 years) with one or more nontraumatic vertebral compression fracture(s) and 22 similar women (64 ⫾ 9 years) without fractures. As reported in part I, no significant difference was recorded in dual-energy X-ray absorptiometry (DXA) of the spine between the two groups, and none of the subjects was on treatment likely to affect the bone. Biopsies from a group of 27 men with and without vertebral fracture (see part I) were similarly examined. After preservation in 70% alcohol, embedding in methylmethacrylate,3 and the removal of 8-m-thick sections for histomorphometry,2,11 the blocks were transferred to a Microslice 2 (Metals Research, Ltd., Cambridge, UK) that cuts by means of a watercooled, diamond-impregnated rotating disk. A single slice, 300 m thick, was collected from each specimen, such that the biopsy cylinder was cut in the usual longitudinal plane (illustrat-
Bone Vol. 27, No. 2 August 2000:277–282
ed in part I). The thick slices were superficially stained on both sides with the same dye, or by immersing one side in 1% Alizarin red for 5 min, followed by immersion of the other side in 1% light green for 5 min (Shore et al.14). The slices were viewed under the optical microscope in plain or partly polarized light either unmounted or after mounting in XAM organic medium (BDH, Poole, UK), which eliminated the parallel striations occasionally produced by the slicing process. Using a hand counter, the number of real (i.e., unstained) trabecular termini (Re.Tm) and the numbers of real isolated islands of bone (i.e., unstained throughout; Re.Tm-Re.Tm) were counted within the depth (or third dimension) of the slice (Figure 1). Also counted were the apparent isolated trabeculae, which in this case included the real unstained ones that were sparse in number, together with other single struts with one or both termini stained (an example is evident in the upper left corner of Figure 3a). The area of interest (i.e., the spongiosa between the two cortices) was defined within a window of recorded size prepared from a masked coverslip (Figure 2). Results were compared with the corresponding TAS analysis (trabecular analysis system2) of apparent termini (sometimes called free ends8) and islands measured together with other structural variables in an adjacent thin section (see part I for details). Intraobserver variation in the number of real termini was determined by one observer (P.A.S.) in a blinded manner, analyzing thick slices from 20 randomly selected biopsies on two occasions separated by a time interval of 3 months. Interobserver variation was also determined in a blinded manner on the same day by two independent observers (P.A.S. and J.E.A.) using the aforementioned 20 thick slices. Results As described in part I a similarity in the bone status of the fracture and nonfracture groups of women was indicated at the spine in terms of the bone mineral density (BMD) measured by absorptiometry and at the iliac crest in terms of the relative bone volume measured histologically (fracture 9.9 ⫾ 4.0% and nonfracture 10.2 ⫾ 3.7% using TAS). The same situation was repeated in the specimens from the men (fracture 10.7 ⫾ 4.5% and nonfracture 10.1 ⫾ 3.7% using TAS). When undecalcified 8-m-thick histological sections from each group were viewed under the optical microscope they presented the discontinuous cancellous image characteristic of idiopathic osteopenia in which trabecular termini were a common feature. Some of the trabecular termini, particularly at the periphery of the specimen, had arisen as a result of mechanical damage during the removal of the biopsy. These typically had sharp angular, and sometimes frag-
Figure 2. Low-power view of a typical 300-m-thick slice from an osteopenic transiliac bone biposy. The cancellous area of interest is outlined. Upper and lower surfaces stained with Alizarin red. (a) Plain light and (b) partly polarized light showing clearer 3D structure. Original magnification: ⫻10.
Bone Vol. 27, No. 2 August 2000:277–282
J. E. Aaron et al. Three-dimensional histology and vertebral fracture
279
Figure 3. “Apparent” and “real” trabecular termini demonstrated by (a) superficial staining on upper and lower surfaces with Alizarin red, or (b) Alizarin red and fast green when apparent termini (arrows) are colored and real termini (arrowheads) are unstained (i.e., white). Original magnification: ⫻20. (c) Discontinuity may arise from a microfracture (arrowed; original magnification: ⫻60) and may stimulate (d) callus repair (arrowed; original magnification: ⫻20) (e) A real terminus (single arrowhead) together with a real island (double arrowhead) are unstained. Arrows show a superficially stained island of bone that is clearly part of an interconnected system in the 3D image. This contrasts with the superficially stained apparent island of (a) (top left corner) that remains detached in the 3D image (although it may interconnect elsewhere). Original magnification: ⫻60. (f) Disconnections are rare in healthy cancellous bone. Original magnification: ⫻20.
mented, edges, and generally occurred in compacted groups.14 They contrasted with others that had smoother rounded contours and were surrounded by intact marrow tissue. Only the latter were generally included in the present quantitation, because, although a proportion of the former may have existed in vivo1 as a result of localized microfissuring, their identification using small biopsies is generally unreliable. When 300-m-thick slices of undecalcified bone were examined under the light microscope the trabecular termini within the intact spongiosa could be divided into two groups. “Apparent” termini stained red or green were clearly distinguishable from the “real” termini, which were unstained (Figure 3a,b). Occasionally, microfractures were observed (Figure 3c), their occurrence confirmed by the presence of sites of callus repair (Figure 3d), as
described previously by our group1 and others.4,15 Also evident were unstained (i.e., real) isolated islands of bone (Figure 3e), although these were uncommon. In some biopsies, trabecular disconnection was rare with no evidence of discontinuity in the regularly anastomosing system (Figure 3f). A comparison of cancellous termini, as demonstrated in women by 2D and 3D imaging, is shown in Figure 4. In the 2D image measured by TAS (Figure 4a,b) apparent isolated trabeculae (mean 14 per 30 mm2 field) and apparent trabecular termini (mean 50 per 30 mm2 field) appeared with similar frequency in both fracture and nonfracture subjects. In the 3D image, on the other hand, the number of apparent isolated trabeculae (i.e., single unbranched struts) stained at one or both termini was reduced (collective mean 2.5 per 30 mm2 field), with the majority (i.e., 85%) now
280
J. E. Aaron et al. Three-dimensional histology and vertebral fracture
Bone Vol. 27, No. 2 August 2000:277–282
Figure 4. Histograms of apparent and real trabecular disconnections in fractured (F) and nonfractured (NF) women with the same percent bone volume. (a,b) Two-dimensional image: Cancellous termini. In (a) and (b), thin sections (which do not differentiate real from apparent) showed no significant difference in either the number of isolated trabeculae (bone islands) or termini. (c,d) Three-dimensional image: Cancellous termini. In (c), using thick slices, most of the islands of bone observed in the 2D image above were interconnected in the 3D image and, although more apparent islands (single unconnected struts both with as well as without terminal staining) remained detached in F than NF, the difference did not reach significance. (The proportion of these that were demonstrably real isolated trabeculae [i.e., unstained throughout] was too small for comparison, amounting to only five instances in the combined groups.) Significance was reached, however, in (d) where the number of real (unstained) trabecular termini was higher in fracture than in nonfracture cases. *p ⬍ 0.05; data expressed as mean ⫾ SE. (The field size of 30 mm2 is arbitrary and was selected because it fits within the cancellous area of a typical bone biopsy section.) Filled bars (NF), nonfractured; hatched bars (F), fractured.
being demonstrated as part of the major network, particularly in the nonfracture group, although the difference remained nonsignificant (Figure 4c). At the same time, the number of real termini was invariably low, such that fewer than 2 in 50 apparent termini were real. In other words, 96% of apparent trabecular termini in thin sections of the aforementioned material were not real. Similarly, only a fraction of the apparent trabecular islands (five throughout the entire collection of samples and too few for statistical analysis) was confirmed as real (i.e., unstained at both termini). However, there was a significant difference in the number of real trabecular termini between the two groups of women, with the fracture subjects having four times the number found in the nonfracture group (Figure 4d). Comparison with the samples from the men suggests that they tended to have fewer real termini than did the women, and although the mean number was greater in those with fractures (0.69 ⫾ 0.30/30 mm2) than
those without (0.18 ⫾ 0.18/30 mm2) the difference was not quite significant. That the reproducibility of the novel method applied was acceptable, and the results reliable, is supported by the intraobserver and interobserver coefficients of variation (CV%), which were calculated to be 1.70 and 1.95, respectively. Discussion Trabecular disconnection may occur histologically for a number of reasons. It may result from localized osteoclastic activity leading to trabecular perforation, from the progressive attenuation of a bony bar due to inadequate apposition, or from the proliferation of microfissures into fatigue fractures and localized sites of dissolution and dispersion that are sometimes followed by callus repair.1,15 Disconnection may also arise as a prepara-
Bone Vol. 27, No. 2 August 2000:277–282
tive artifact related to peripheral trephine damage, an inadequate histology technique, or an inherent fragility of the material. Criteria used in defining and eliminating areas of artifactual damage have been described elsewhere.14 Whatever the etiology, the aforementioned evidence suggests that a separate role for trabecular disconnection independent of bone mass is likely in vertebral osteoporotic fracture.10 This conclusion is contrary to that of Flautre and Hardouin7 whose fracture subjects (bone volume 15%) had fewer trabeculae than nonfracture subjects (bone volume 19%), yet, paradoxically, had better connectivity (in terms of the trabecular bone pattern factor, although not in terms of the trabecular plate density—a measurement now called trabecular number). In the present study, subjects were matched skeletally in terms of BMD at the spine as well as histologically at the ilium, suggesting similar biological behavior at each site. Microstructural differences that were independent of bone mass and consistent with those just noted have been reported previously by Kleerekoper et al.13 using histological sections and variables derived indirectly from area and perimeter measurements. Both of the aforementioned groups of investigators acknowledged the limitations imposed by 2D histological sections and the need for a rigorous measurement of connectivity in the skeletal context (whether defined topologically or morphologically). In contrast to the observations of Kleerekoper et al.,13 the trabecular plate density (now known as the trabecular number), together with several other established variables previously applied to the material in part I (e.g., the trabecular node:terminus ratio,8 the trabecular pattern factor,9 the marrow star volume16), showed no more than a common trend that failed to reach significance.11 That the “real” trabecular terminus number succeeded in demonstrating a significant difference between the same groups suggests the direct method applied here is more sensitive. It follows that, as a measure of trabecular disconnection, the number of real trabecular termini in a defined field is apparently a better discriminator of fracture than the bone volume alone. Moreover, the information can be obtained at a modest cost using standard hard tissue laboratory equipment, in contrast to the technologically complex and expensive alternatives described in the literature.6,12 Young healthy bone is less usefully examined in this way, because in a thick slice it presents an image in which the trabeculae are too dense and superimposed. Furthermore, while in the atrophied material noted earlier some 96% of the apparent trabecular termini were not real, in normal young adult spongiosa the evidence suggests that this figure is probably 100%, with any disconnection being a transient rarity. As is the case for all architectural variables, accuracy in determining the number of real trabecular termini is dependent upon the integrity of the bone biopsy, and the trauma of extraction may create fractured trabeculae and displaced fragments. This led to the exclusion of a small number of specimens from otherwise suitable subjects. The retention of the marrow tissue throughout was considered an essential aspect of the method in ensuring that naturally detached components of the cancellous network and isolated trabeculae were not lost. Isolated trabeculae were uncommon in the relatively small sample provided by the typical iliac bone biopsy. At the same time, any detached elements exceeding the 300-m-thick dimension of the slice would not be perceived as such as the termini would lie outside the window of observation, hence the need for the thickest slice possible.14 In the osteopenic cancellous skeleton as a whole, however, discrete islands of bone may persist for longer than is generally supposed and together with disconnected systems will contribute to the bone mass measurements but not to the bone strength. For example, slices of bone 30 mm2 (area) ⫻ 0.3 mm (thick) from 53 female subjects represents a total cancellous
J. E. Aaron et al. Three-dimensional histology and vertebral fracture
281
volume of approximately 0.50 cm3, contained within which were 75 “real” termini and only 5 “real” islands. If the average skeleton weighs 20 kg, 20% of which is cancellous bone with a density of about 2 g/cm3, the cancellous volume of an average skeleton is 2000 cm3 and, based on the evidence given, would be predicted to contain 300,000 real trabecular termini and 20,000 real islands. This estimate can be compared with a similar evaluation of a large ilial autopsy specimen from an 80-year-old man without fractures, which was derived using a video reconstruction of the surface of a serially sectioned block (M. F. Wilson and J. E. Aaron, unpublished data). The projected cancellous skeleton was found to contain 80,000 real termini and 8000 real islands. The difference may be due in part to a more effectively sustained connectivity in men than in women.3 The presence of small isolated trabecular elements supported by marrow tissue has been described in the proximal tibia by Kinney et al.12 in ovariectomized rats 5 weeks after operation, with these amounting to about 1.5% of the total trabecular bone volume. The relative variations in connectivity just described may explain why some subjects with an apparently adequate bone mass suffer fractures, whereas differences in neuromuscular function, aspects of lifestyle, or other features of bone “quality” as yet too ill-defined for quantitation may explain why some subjects remain structurally intact despite an impoverished bone mass. An upward trend in trabecular thickness13 may be a compensatory mechanism after disconnection, although this was not confirmed by Flautre and Hardouin.7 Finally, despite a compelling trend, no significant difference in histological connectivity was found in the smaller group of men11 examined, possibly because numbers of suitable biopsies were too few and possibly because male osteoporosis is complicated by secondary factors. In conclusion, a simple histological method has been applied to the evaluation of trabecular termini in osteopenic women and men with and without vertebral compression fractures. This method demonstrated a mass-independent difference in women in terms of more numerous real trabecular termini in those with fractures. Comparison with established indirect methods suggests that the direct method explained herein is a more sensitive predictor of fracture predisposition and may provide a useful adjunct in the appraisal of bone fragility.
Acknowledgments: The project was funded by a grant from Action Research and was part of the master’s of science dissertation of P. A. Shore. We are grateful to V. Hawkins of the Division of Oral Biology, Dental Institute, University of Leeds, for valuable technical advice.
References 1. Aaron, J. E. Autoclasis—a mechanism of bone resorption and an alternative explanation for osteoporosis. Calcif Tissue Res 22(Suppl.):247–254; 1977. 2. Aaron, J. E., Johnson, D. R., Kanis, J. A., Oakley, B. A., and Paxton, S. K. An automated method for the analysis of trabecular bone structure. Comput Biomed Res 25:1–16; 1992. 3. Aaron, J. E., Makins, N. B., and Sagreiya, K. The microanatomy of trabecular bone loss in normal aging men and women. Clin Orthop 215:260 –271; 1987. 4. Amling, M., Hahn, M., Wening, V. J., Grote, H. J., and Delling, G. The microarchitecture of the axis as the predisposing factor for fracture of the base of the odontoid process. J Bone Jt Surg 76-A:1840 –1846; 1994. 5. Atkinson, P. Variation in trabecular structure of vertebrae with age. Calcif Tissue Res 1:24 –37; 1967. 6. Feldkamp, L. A., Goldstein, S. A., Parfitt, M., Jesion, G., and Kleerekoper, M. The direct examination of three-dimensional bone architecture in vitro by computed tomography. J Bone Miner Res 4:3–11; 1989. 7. Flautre, B. and Hardouin, P. Microradiographic aspect on iliac bone tissue in
282
8.
9.
10. 11.
12.
J. E. Aaron et al. Three-dimensional histology and vertebral fracture postmenopausal women with and without vertebral crush fractures. Bone 15:477– 481; 1994. Garrahan, N. J., Mellish, W. E., and Compston, J. E. A new method for the two-dimensional analysis of bone structure in human iliac crest bone biopsies. J Microsc 142:341–349; 1986. Hahn, M., Vogel, M., Pompesius-Kempa, M., and Delling, G. Trabecular bone pattern factor: A new parameter for simple quantification of bone microarchitecture. Bone 13:327–330; 1990. Heaney, R. P. The natural history of vertebral osteoporosis. Is low bone mass an epiphenomenon? Bone 13(Suppl.):S23–S26; 1992. Hordon, L., Raisi, M., Aaron, J. E., Paxton, S. K., Beneton, M., and Kanis, J. A. Trabecular architecture in women and men of similar bone mass, with and without vertebral fracture: I. Two-dimensional histology. Bone 27:271–276; 2000. Kinney, J. H., Lane, N. E., and Haupt, D. L. In vivo three-dimensional microscopy of trabecular bone. J Bone Miner Res 10:264 –270; 1995.
Bone Vol. 27, No. 2 August 2000:277–282 13. Kleerekoper, M., Villanueva, A. R., Stanciu, J., Rao, S. D., and Parfitt, M. The role of three-dimensional trabecular microstructure in the pathogenesis of vertebral compression fractures. Calcif Tissue Int 37:594 –597; 1985. 14. Shore, P., Shore, R. C., and Aaron, J. E. A histological method to directly determine the number of trabecular termini in cancellous bone. Biotechn Histochem. In press. 15. Todd, R. C., Freeman, A. R., and Pirie, C. J. Isolated trabecular fatigue fractures in the femoral head. J Bone Jt Surg 54-B:723–728; 1972. 16. Vesterby, A. Star volume of marrow space and trabeculae in iliac crest. Bone 11:149 –155; 1990.
Date Received: August 13, 1999 Date Revised: March 22, 2000 Date Accepted: March 22, 2000
Trabecular Architecture in Women and Men of Similar Bone Mass With and Without Vertebral Fracture: II. Three-Dimensional Histology J. E. AARON,1 P. A. SHORE,1 R. C. SHORE,2 M. BENETON,3 and J. A. KANIS3 1
School of Biomedical Sciences, Worsley Medical and Dental Building, University of Leeds, Leeds, UK Division of Oral Biology, Dental Institute, University of Leeds, Leeds, UK 3 Centre for Metabolic Bone Diseases at Sheffield (WHO Collaborating Centre), Medical School, University of Sheffield, Sheffield, UK 2
Introduction We recently developed a simple and inexpensive method that complements established bone histomorphometry procedures by enabling the two-dimensional imaging of cancellous bone to be viewed within its three-dimensional context with the marrow tissue in place and without detriment to the material for other histological purposes. The method, based on the preparation and superficial staining of slices 300 m thick, enables “real” (i.e., unstained) trabecular termini to be separated from “artifactual” (i.e., stained) termini, providing a direct measure of cancellous connectivity in osteopenic bone. The technique was applied to osteopenic age-matched, white, postmenopausal women (31 with and 22 without vertebral compression fractures) with a similar bone status, as measured at the spine by absorptiometry and at the iliac crest by histology (see part I of this study). Despite the similarity in the mass of trabecular bone at either site, the results showed a significant difference (p < 0.05) in the number of “real” trabecular termini between the groups, such that the fracture group had almost four times as many termini (mean ⴞ SE: 1.98 ⴞ 0.51/30 mm2) at the iliac crest as the nonfracture group (mean ⴞ SE: 0.53 ⴞ 0.31/30 mm2). Previous histomorphometry of the same material failed to detect a structural distinction between the two groups using established variables. It was concluded that a mass-independent trabecular discontinuity contributes to skeletal failure and that determination of the number of “real” disconnections (i.e., unstained termini) by the direct method proposed may provide a more sensitive discriminant of fracture than the present indirect procedures. A group of fracture and nonfracture men (see part I) suggested a similar distinction (fracture: 0.69 ⴞ 0.30/30 mm2; nonfracture: 0.18 ⴞ 0.18/30 mm2), although the difference was not significant. (Bone 27: 277–282; 2000) © 2000 by Elsevier Science Inc. All rights reserved.
It is well established that subjects with osteoporotic fractures tend to have lower mean bone mass than those without, whether their bone status is determined by noninvasive means or histologically using iliac crest bone biopsies. However, separation on the basis of bone mass alone is incomplete; that is, a substantial overlap in the data indicates that some subjects with an apparently adequate bone status sustain osteoporotic fractures, whereas others with an apparently inadequate bone status do not. Many other factors that contribute to atraumatic fracture have been recorded in the literature. Prominent among these factors is the trabecular architecture and, in particular, trabecular connectivity, because a well-interconnected structure is more resistant to bending forces than one that is not. A number of mathematical procedures have been applied to the measurement of trabecular connectivity indirectly in order to overcome the well-recognized intrinsic problem that, in thin histological sections, apparently disconnected bars may derive entirely from the plane of section. This means that many of the “apparent” trabecular termini in the two-dimensional (2D) histological image are artifactual. Analysis of the three-dimensional (3D) image evident in unsectioned material may be expected to overcome the difficulty; however, such images tend to consist of superimposed and complex trabecular networks that are not easy to evaluate. Moreover, removal of marrow tissue is invariably a requirement for a well-delineated image. This will be achieved at the cost of any skeletal elements that are detached from the main trabecular framework and which will disappear together with the eluted marrow cells. Although this is unlikely to affect the substantial interconnected framework characteristic of young healthy bone, it may affect the appraisal of elderly, disconnected, fragile, and osteopenic material (see, e.g., Kinney et al.12). A satisfactory compromise for analysis may lie somewhere between the thin histological section and the gross sawn segment favored by Atkinson5 and Amling et al.4 Their work serves to illustrate the potential a simple method might have, particularly if it was also inexpensive, nondisruptive to soft tissues, and applicable to small samples such as transiliac bone biopsies. Recently, such a method has been developed and which is not destructive for other histological purposes. On the contrary, it complements established histomorphometrical procedure by combining the 2D and 3D images and providing essential insight into the third dimension. It is based on the differential surface staining of thick, optically acceptable slices of cancellous bone.14
Key Words: Bone histology; Trabecular termini; Iliac crest biopsy; Two-dimensional (2D) and three-dimensional (3D) image analysis; Vertebral fracture; Osteoporosis.
Address for correspondence and reprints: Dr. Jean E. Aaron, School of Biomedical Sciences, Worsley Medical and Dental Building, University of Leeds, Leeds LS2 9JT, UK. E-mail: [email protected]@leeds.ac.uk © 2000 by Elsevier Science Inc. All rights reserved.
277
8756-3282/00/$20.00 PII S8756-3282(00)00328-8
278
J. E. Aaron et al. Three-dimensional histology and vertebral fracture
Figure 1. Diagram illustrating apparent (*) and real (⫹) trabecular termini and an “apparent” (ApIs) and a “real” (ReIs) island (isolated trabeculae) in a histological section (10 m thick; 2D image) and a slice (300 m; 3D image) of cancellous bone.
The present study applies this novel technique to the comparison of trabecular connectivity in transiliac biopsies from two previously analyzed groups of postmenopausal women11 with matched bone mass and with and without minimal trauma vertebral compression fractures. Earlier histomorphological examinations of 8-m-thick sections have been unable to demonstrate any significant structural difference between them. Methods The material was the same as that used in part I of this study.11 Transiliac bone biopsies, 8 mm in diameter, were taken from 31 postmenopausal white women (mean ⫾ SD: 66 ⫾ 9 years) with one or more nontraumatic vertebral compression fracture(s) and 22 similar women (64 ⫾ 9 years) without fractures. As reported in part I, no significant difference was recorded in dual-energy X-ray absorptiometry (DXA) of the spine between the two groups, and none of the subjects was on treatment likely to affect the bone. Biopsies from a group of 27 men with and without vertebral fracture (see part I) were similarly examined. After preservation in 70% alcohol, embedding in methylmethacrylate,3 and the removal of 8-m-thick sections for histomorphometry,2,11 the blocks were transferred to a Microslice 2 (Metals Research, Ltd., Cambridge, UK) that cuts by means of a watercooled, diamond-impregnated rotating disk. A single slice, 300 m thick, was collected from each specimen, such that the biopsy cylinder was cut in the usual longitudinal plane (illustrat-
Bone Vol. 27, No. 2 August 2000:277–282
ed in part I). The thick slices were superficially stained on both sides with the same dye, or by immersing one side in 1% Alizarin red for 5 min, followed by immersion of the other side in 1% light green for 5 min (Shore et al.14). The slices were viewed under the optical microscope in plain or partly polarized light either unmounted or after mounting in XAM organic medium (BDH, Poole, UK), which eliminated the parallel striations occasionally produced by the slicing process. Using a hand counter, the number of real (i.e., unstained) trabecular termini (Re.Tm) and the numbers of real isolated islands of bone (i.e., unstained throughout; Re.Tm-Re.Tm) were counted within the depth (or third dimension) of the slice (Figure 1). Also counted were the apparent isolated trabeculae, which in this case included the real unstained ones that were sparse in number, together with other single struts with one or both termini stained (an example is evident in the upper left corner of Figure 3a). The area of interest (i.e., the spongiosa between the two cortices) was defined within a window of recorded size prepared from a masked coverslip (Figure 2). Results were compared with the corresponding TAS analysis (trabecular analysis system2) of apparent termini (sometimes called free ends8) and islands measured together with other structural variables in an adjacent thin section (see part I for details). Intraobserver variation in the number of real termini was determined by one observer (P.A.S.) in a blinded manner, analyzing thick slices from 20 randomly selected biopsies on two occasions separated by a time interval of 3 months. Interobserver variation was also determined in a blinded manner on the same day by two independent observers (P.A.S. and J.E.A.) using the aforementioned 20 thick slices. Results As described in part I a similarity in the bone status of the fracture and nonfracture groups of women was indicated at the spine in terms of the bone mineral density (BMD) measured by absorptiometry and at the iliac crest in terms of the relative bone volume measured histologically (fracture 9.9 ⫾ 4.0% and nonfracture 10.2 ⫾ 3.7% using TAS). The same situation was repeated in the specimens from the men (fracture 10.7 ⫾ 4.5% and nonfracture 10.1 ⫾ 3.7% using TAS). When undecalcified 8-m-thick histological sections from each group were viewed under the optical microscope they presented the discontinuous cancellous image characteristic of idiopathic osteopenia in which trabecular termini were a common feature. Some of the trabecular termini, particularly at the periphery of the specimen, had arisen as a result of mechanical damage during the removal of the biopsy. These typically had sharp angular, and sometimes frag-
Figure 2. Low-power view of a typical 300-m-thick slice from an osteopenic transiliac bone biposy. The cancellous area of interest is outlined. Upper and lower surfaces stained with Alizarin red. (a) Plain light and (b) partly polarized light showing clearer 3D structure. Original magnification: ⫻10.
Bone Vol. 27, No. 2 August 2000:277–282
J. E. Aaron et al. Three-dimensional histology and vertebral fracture
279
Figure 3. “Apparent” and “real” trabecular termini demonstrated by (a) superficial staining on upper and lower surfaces with Alizarin red, or (b) Alizarin red and fast green when apparent termini (arrows) are colored and real termini (arrowheads) are unstained (i.e., white). Original magnification: ⫻20. (c) Discontinuity may arise from a microfracture (arrowed; original magnification: ⫻60) and may stimulate (d) callus repair (arrowed; original magnification: ⫻20) (e) A real terminus (single arrowhead) together with a real island (double arrowhead) are unstained. Arrows show a superficially stained island of bone that is clearly part of an interconnected system in the 3D image. This contrasts with the superficially stained apparent island of (a) (top left corner) that remains detached in the 3D image (although it may interconnect elsewhere). Original magnification: ⫻60. (f) Disconnections are rare in healthy cancellous bone. Original magnification: ⫻20.
mented, edges, and generally occurred in compacted groups.14 They contrasted with others that had smoother rounded contours and were surrounded by intact marrow tissue. Only the latter were generally included in the present quantitation, because, although a proportion of the former may have existed in vivo1 as a result of localized microfissuring, their identification using small biopsies is generally unreliable. When 300-m-thick slices of undecalcified bone were examined under the light microscope the trabecular termini within the intact spongiosa could be divided into two groups. “Apparent” termini stained red or green were clearly distinguishable from the “real” termini, which were unstained (Figure 3a,b). Occasionally, microfractures were observed (Figure 3c), their occurrence confirmed by the presence of sites of callus repair (Figure 3d), as
described previously by our group1 and others.4,15 Also evident were unstained (i.e., real) isolated islands of bone (Figure 3e), although these were uncommon. In some biopsies, trabecular disconnection was rare with no evidence of discontinuity in the regularly anastomosing system (Figure 3f). A comparison of cancellous termini, as demonstrated in women by 2D and 3D imaging, is shown in Figure 4. In the 2D image measured by TAS (Figure 4a,b) apparent isolated trabeculae (mean 14 per 30 mm2 field) and apparent trabecular termini (mean 50 per 30 mm2 field) appeared with similar frequency in both fracture and nonfracture subjects. In the 3D image, on the other hand, the number of apparent isolated trabeculae (i.e., single unbranched struts) stained at one or both termini was reduced (collective mean 2.5 per 30 mm2 field), with the majority (i.e., 85%) now
280
J. E. Aaron et al. Three-dimensional histology and vertebral fracture
Bone Vol. 27, No. 2 August 2000:277–282
Figure 4. Histograms of apparent and real trabecular disconnections in fractured (F) and nonfractured (NF) women with the same percent bone volume. (a,b) Two-dimensional image: Cancellous termini. In (a) and (b), thin sections (which do not differentiate real from apparent) showed no significant difference in either the number of isolated trabeculae (bone islands) or termini. (c,d) Three-dimensional image: Cancellous termini. In (c), using thick slices, most of the islands of bone observed in the 2D image above were interconnected in the 3D image and, although more apparent islands (single unconnected struts both with as well as without terminal staining) remained detached in F than NF, the difference did not reach significance. (The proportion of these that were demonstrably real isolated trabeculae [i.e., unstained throughout] was too small for comparison, amounting to only five instances in the combined groups.) Significance was reached, however, in (d) where the number of real (unstained) trabecular termini was higher in fracture than in nonfracture cases. *p ⬍ 0.05; data expressed as mean ⫾ SE. (The field size of 30 mm2 is arbitrary and was selected because it fits within the cancellous area of a typical bone biopsy section.) Filled bars (NF), nonfractured; hatched bars (F), fractured.
being demonstrated as part of the major network, particularly in the nonfracture group, although the difference remained nonsignificant (Figure 4c). At the same time, the number of real termini was invariably low, such that fewer than 2 in 50 apparent termini were real. In other words, 96% of apparent trabecular termini in thin sections of the aforementioned material were not real. Similarly, only a fraction of the apparent trabecular islands (five throughout the entire collection of samples and too few for statistical analysis) was confirmed as real (i.e., unstained at both termini). However, there was a significant difference in the number of real trabecular termini between the two groups of women, with the fracture subjects having four times the number found in the nonfracture group (Figure 4d). Comparison with the samples from the men suggests that they tended to have fewer real termini than did the women, and although the mean number was greater in those with fractures (0.69 ⫾ 0.30/30 mm2) than
those without (0.18 ⫾ 0.18/30 mm2) the difference was not quite significant. That the reproducibility of the novel method applied was acceptable, and the results reliable, is supported by the intraobserver and interobserver coefficients of variation (CV%), which were calculated to be 1.70 and 1.95, respectively. Discussion Trabecular disconnection may occur histologically for a number of reasons. It may result from localized osteoclastic activity leading to trabecular perforation, from the progressive attenuation of a bony bar due to inadequate apposition, or from the proliferation of microfissures into fatigue fractures and localized sites of dissolution and dispersion that are sometimes followed by callus repair.1,15 Disconnection may also arise as a prepara-
Bone Vol. 27, No. 2 August 2000:277–282
tive artifact related to peripheral trephine damage, an inadequate histology technique, or an inherent fragility of the material. Criteria used in defining and eliminating areas of artifactual damage have been described elsewhere.14 Whatever the etiology, the aforementioned evidence suggests that a separate role for trabecular disconnection independent of bone mass is likely in vertebral osteoporotic fracture.10 This conclusion is contrary to that of Flautre and Hardouin7 whose fracture subjects (bone volume 15%) had fewer trabeculae than nonfracture subjects (bone volume 19%), yet, paradoxically, had better connectivity (in terms of the trabecular bone pattern factor, although not in terms of the trabecular plate density—a measurement now called trabecular number). In the present study, subjects were matched skeletally in terms of BMD at the spine as well as histologically at the ilium, suggesting similar biological behavior at each site. Microstructural differences that were independent of bone mass and consistent with those just noted have been reported previously by Kleerekoper et al.13 using histological sections and variables derived indirectly from area and perimeter measurements. Both of the aforementioned groups of investigators acknowledged the limitations imposed by 2D histological sections and the need for a rigorous measurement of connectivity in the skeletal context (whether defined topologically or morphologically). In contrast to the observations of Kleerekoper et al.,13 the trabecular plate density (now known as the trabecular number), together with several other established variables previously applied to the material in part I (e.g., the trabecular node:terminus ratio,8 the trabecular pattern factor,9 the marrow star volume16), showed no more than a common trend that failed to reach significance.11 That the “real” trabecular terminus number succeeded in demonstrating a significant difference between the same groups suggests the direct method applied here is more sensitive. It follows that, as a measure of trabecular disconnection, the number of real trabecular termini in a defined field is apparently a better discriminator of fracture than the bone volume alone. Moreover, the information can be obtained at a modest cost using standard hard tissue laboratory equipment, in contrast to the technologically complex and expensive alternatives described in the literature.6,12 Young healthy bone is less usefully examined in this way, because in a thick slice it presents an image in which the trabeculae are too dense and superimposed. Furthermore, while in the atrophied material noted earlier some 96% of the apparent trabecular termini were not real, in normal young adult spongiosa the evidence suggests that this figure is probably 100%, with any disconnection being a transient rarity. As is the case for all architectural variables, accuracy in determining the number of real trabecular termini is dependent upon the integrity of the bone biopsy, and the trauma of extraction may create fractured trabeculae and displaced fragments. This led to the exclusion of a small number of specimens from otherwise suitable subjects. The retention of the marrow tissue throughout was considered an essential aspect of the method in ensuring that naturally detached components of the cancellous network and isolated trabeculae were not lost. Isolated trabeculae were uncommon in the relatively small sample provided by the typical iliac bone biopsy. At the same time, any detached elements exceeding the 300-m-thick dimension of the slice would not be perceived as such as the termini would lie outside the window of observation, hence the need for the thickest slice possible.14 In the osteopenic cancellous skeleton as a whole, however, discrete islands of bone may persist for longer than is generally supposed and together with disconnected systems will contribute to the bone mass measurements but not to the bone strength. For example, slices of bone 30 mm2 (area) ⫻ 0.3 mm (thick) from 53 female subjects represents a total cancellous
J. E. Aaron et al. Three-dimensional histology and vertebral fracture
281
volume of approximately 0.50 cm3, contained within which were 75 “real” termini and only 5 “real” islands. If the average skeleton weighs 20 kg, 20% of which is cancellous bone with a density of about 2 g/cm3, the cancellous volume of an average skeleton is 2000 cm3 and, based on the evidence given, would be predicted to contain 300,000 real trabecular termini and 20,000 real islands. This estimate can be compared with a similar evaluation of a large ilial autopsy specimen from an 80-year-old man without fractures, which was derived using a video reconstruction of the surface of a serially sectioned block (M. F. Wilson and J. E. Aaron, unpublished data). The projected cancellous skeleton was found to contain 80,000 real termini and 8000 real islands. The difference may be due in part to a more effectively sustained connectivity in men than in women.3 The presence of small isolated trabecular elements supported by marrow tissue has been described in the proximal tibia by Kinney et al.12 in ovariectomized rats 5 weeks after operation, with these amounting to about 1.5% of the total trabecular bone volume. The relative variations in connectivity just described may explain why some subjects with an apparently adequate bone mass suffer fractures, whereas differences in neuromuscular function, aspects of lifestyle, or other features of bone “quality” as yet too ill-defined for quantitation may explain why some subjects remain structurally intact despite an impoverished bone mass. An upward trend in trabecular thickness13 may be a compensatory mechanism after disconnection, although this was not confirmed by Flautre and Hardouin.7 Finally, despite a compelling trend, no significant difference in histological connectivity was found in the smaller group of men11 examined, possibly because numbers of suitable biopsies were too few and possibly because male osteoporosis is complicated by secondary factors. In conclusion, a simple histological method has been applied to the evaluation of trabecular termini in osteopenic women and men with and without vertebral compression fractures. This method demonstrated a mass-independent difference in women in terms of more numerous real trabecular termini in those with fractures. Comparison with established indirect methods suggests that the direct method explained herein is a more sensitive predictor of fracture predisposition and may provide a useful adjunct in the appraisal of bone fragility.
Acknowledgments: The project was funded by a grant from Action Research and was part of the master’s of science dissertation of P. A. Shore. We are grateful to V. Hawkins of the Division of Oral Biology, Dental Institute, University of Leeds, for valuable technical advice.
References 1. Aaron, J. E. Autoclasis—a mechanism of bone resorption and an alternative explanation for osteoporosis. Calcif Tissue Res 22(Suppl.):247–254; 1977. 2. Aaron, J. E., Johnson, D. R., Kanis, J. A., Oakley, B. A., and Paxton, S. K. An automated method for the analysis of trabecular bone structure. Comput Biomed Res 25:1–16; 1992. 3. Aaron, J. E., Makins, N. B., and Sagreiya, K. The microanatomy of trabecular bone loss in normal aging men and women. Clin Orthop 215:260 –271; 1987. 4. Amling, M., Hahn, M., Wening, V. J., Grote, H. J., and Delling, G. The microarchitecture of the axis as the predisposing factor for fracture of the base of the odontoid process. J Bone Jt Surg 76-A:1840 –1846; 1994. 5. Atkinson, P. Variation in trabecular structure of vertebrae with age. Calcif Tissue Res 1:24 –37; 1967. 6. Feldkamp, L. A., Goldstein, S. A., Parfitt, M., Jesion, G., and Kleerekoper, M. The direct examination of three-dimensional bone architecture in vitro by computed tomography. J Bone Miner Res 4:3–11; 1989. 7. Flautre, B. and Hardouin, P. Microradiographic aspect on iliac bone tissue in
282
8.
9.
10. 11.
12.
J. E. Aaron et al. Three-dimensional histology and vertebral fracture postmenopausal women with and without vertebral crush fractures. Bone 15:477– 481; 1994. Garrahan, N. J., Mellish, W. E., and Compston, J. E. A new method for the two-dimensional analysis of bone structure in human iliac crest bone biopsies. J Microsc 142:341–349; 1986. Hahn, M., Vogel, M., Pompesius-Kempa, M., and Delling, G. Trabecular bone pattern factor: A new parameter for simple quantification of bone microarchitecture. Bone 13:327–330; 1990. Heaney, R. P. The natural history of vertebral osteoporosis. Is low bone mass an epiphenomenon? Bone 13(Suppl.):S23–S26; 1992. Hordon, L., Raisi, M., Aaron, J. E., Paxton, S. K., Beneton, M., and Kanis, J. A. Trabecular architecture in women and men of similar bone mass, with and without vertebral fracture: I. Two-dimensional histology. Bone 27:271–276; 2000. Kinney, J. H., Lane, N. E., and Haupt, D. L. In vivo three-dimensional microscopy of trabecular bone. J Bone Miner Res 10:264 –270; 1995.
Bone Vol. 27, No. 2 August 2000:277–282 13. Kleerekoper, M., Villanueva, A. R., Stanciu, J., Rao, S. D., and Parfitt, M. The role of three-dimensional trabecular microstructure in the pathogenesis of vertebral compression fractures. Calcif Tissue Int 37:594 –597; 1985. 14. Shore, P., Shore, R. C., and Aaron, J. E. A histological method to directly determine the number of trabecular termini in cancellous bone. Biotechn Histochem. In press. 15. Todd, R. C., Freeman, A. R., and Pirie, C. J. Isolated trabecular fatigue fractures in the femoral head. J Bone Jt Surg 54-B:723–728; 1972. 16. Vesterby, A. Star volume of marrow space and trabeculae in iliac crest. Bone 11:149 –155; 1990.
Date Received: August 13, 1999 Date Revised: March 22, 2000 Date Accepted: March 22, 2000