Irreversible Perforations in Vertebral Trabeculae?

X BANSE,1 JP DEVOGELAER,1 C DELLOYE,1 A LAFOSSE,1 D HOLMYARD,2 and M GRYNPAS2

ABSTRACT

In human cancellous bone, osteoclastic perforations resulting from normal remodeling were generally con-sidered irreversible. In human vertebral samples, examined by backscatter electron microscopy, there wasclear evidence of bridging of perforation defects by new bone formation. Hence trabecular perforations maynot be irreversible.

Introduction: Preservation of the trabecular bone microarchitecture is essential to maintain its load-bearing capacityand prevent fractures. However, during bone remodeling, the osteoclasts may perforate the platelike trabeculae anddisconnect the structure. Large perforations (�100 �m) are generally considered irreversible because there is nosurface on which new bone can be laid down. In this work, we investigated the outcome of these perforations onhuman vertebral cancellous bone.Materials and Methods: Using backscatter electron microscopy, we analyzed 264 vertebral bone samples from thethoracic and lumbar spine of nine subjects (44–88 years old). Nine fields (2 � 1.5 mm) were observed on each block.Several bone structural units (BSUs) were visible on a single trabecula, illustrating a dynamic, historical aspect ofbone remodeling. A bridge was defined as a single and recent BSU connecting two segments of trabeculae previouslyseparated by osteoclastic resorption. They were counted and measured (length and breadth, �m).Results and Conclusion: We observed 396 bridges over 2376 images. By comparison, we found only 15microcalluses on the same material. The median length of the bridge was 165 �m (range, 29–869 �m); 86% beinglonger than 100 �m and 35% longer than 200 �m. Their breadth was 56 �m (range, 6–255 �m), but the thinnestwere still in construction. Bridges were found in all nine subjects included in the study, suggesting that it is a commonfeature of normal vertebral bone remodeling. These observations support the hypothesis that perforation could berepaired by new bone formation. and hence, might not be systematically irreversible.J Bone Miner Res 2003;18:1247–1253

INTRODUCTION

BONE LOSS IN elderly patients is presently understood as aimbalance between bone resorption and new bone for-

mation, with progressive net loss over decades. In cancel-lous bone, this change in bone amount is associated withimportant structural changes. The key change in trabeculararchitecture seems to be rather a loss of entire structuralelements (drop of Tb.N) than a progressive thinning of allthe trabeculae (drop of Tb.Th). This has been observedinitially on iliac crest biopsy specimens(1,2) and confirmedon vertebral trabecular bone.(3,4) The remodeling cycle (nor-mal or pathological) is held responsible for that. Duringbone resorption, osteoclasts create cavities that may perfo-rate the trabecula.(5) Such disruption of the trabecular net-

work is also generally considered irreversible simply be-cause new bone apposition can only occur on a pre-existingbone surface, and if the surface is removed (in case ofperforation), the lost bone cannot be replaced by the normalremodeling process.(6) Consequently, perforations (i.e.,larger than 100 �m) are generally considered as irreversible.

The dynamic of bone remodeling can be observed usingvarious methods.(7,8) The backscatter electron microscopy(BSE) has proven effective in investigating local bone min-eralization. Different research groups have used this tech-nique to perform quantitative studies on the mineralizationprofile of bone mainly on iliac crest or proximal femur.(9–12)

In addition, BSE could also be used to study other aspectsof the cancellous bone remodeling. After its production byosteoblasts, the osteoid matrix undergoes a fast (primary)mineralization followed by a by a period of maturationduring which mineral is gradually deposited over yearsThe authors have no conflict of interest.

1Orthopaedic Research Laboratory and Arthritis Unit, Universite Catholique de Louvain, Brussels, Belgium.2Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, Canada.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 18, Number 7, 2003© 2003 American Society for Bone and Mineral Research

1247

(secondary mineralization). As a direct consequence of theprogressive and quite slow secondary mineralization, eachlocal wave of remodeling (or bone structural unit [BSU]) iseasy to identify on BSE image (Fig. 1). While bone lamel-lation or cement lines have been used to assess the size ofthe BSU (to measure the wall thickness),(8) the gray levelscan be used in BSE. A given trabecula looks like a patch-work of gray levels. Recent BSUs are relatively less min-eralized and appear as dark gray. Older BSUs are moremineralized and appear whiter.

In this study, we used BSE microscopy on a large scale,examining 264 samples from human vertebral cancellousbone. Looking at the precise location of recent BSUs withinthe trabecular network, our working hypothesis was thattwo-dimensional (2D) perforations are not irreversible.

MATERIALS AND METHODS

Samples preparation

Samples were obtained from nine autopsy subjects.(13–15)

In each subject, three vertebral bodies were obtained, onelumbar (L4), one from the thoraco-lumbar junction (T12 orL1), and one from the lower thoracic spine (T9). Twocylindrical cores (one anterior, one posterior) were extractedfrom each thoracic vertebral body, and three were extractedin the thoraco-lumbar or lumbar vertebral body because anadditional external core could be obtained. This lead to atotal of 68 cylindrical cores. These were fixed and embed-ded in spurr resin using a standard protocol.(16) The plasticblocks containing the whole cylindrical specimen were thencut at four levels (from bottom to top) using an Exactcutting grinding machine and re-embedded to obtain 268blocks. The same blocks were polished with a 1-�m dia-mond finish and carbon coated. Four blocks, broken duringthe polishing, were not used.

BSE image collection

We used a XL30 SEM (Philips/FEI, Eindhoven, TheNetherlands) with a BSE detector (Philips/FEI). The signalwas first calibrated using C (Z � 6) and Al (Z � 13), andthen the settings were changed to increase the contrast of thebone signal (Z � 10). During the acquisition session, wecontrolled the drift of the signal using SiO2 as a standard(only for slight adjustments). Beam intensity was 20 kV.

Digital images were collected at �100 magnification, eachfiled measuring 2 � 1.5 mm. Images were saved as TIFF files(645 � 484 pixels, pixel size � 3.21 �m). Nine adjacentimages were obtained on each block (3 rows � 3 columns).Thirteen images were lost because of operator error.

Image analysis

A total of 2363 BSE images were examined for the twospecific features described below. When a given featurecould be identified on two adjacent images, it was onlycounted once.

We searched for microcalluses. A microcallus was de-fined as a globular structure containing woven bone (vari-able degree of mineralization, large osteocyte lacunae, poresfor vessel).(5,17,18) Usually, the microcallus contained a ma-ture trabecula or joined two trabeculae, although this wasnot a restrictive criteria.

We also looked for bridges. Here, a bridge is arbitrarilydefined as follows (Fig. 2A): (1) a single BSU that connectstwo segments of trabeculae; (2) the BSU has a homoge-neous gray level (generally containing osteocyte lacunae)that is lower than the connected elements; and (3) contraryto the edges of the connected segments of trabeculae, whichusually show signs of osteoclastic resorption (irregular bor-der and interruption of the cement lines), the surface of thebridge is smooth.

Microcalluses and bridges were identified and counted oneach image. The length (�m) and the minimal breadth (�m)of each bridge was measured (Fig. 2B). In addition, wemeasured the bone surface and bone perimeter for eachblock, summing the values from nine images. Using simplemacros under Qwin Pro (Leica Imaging System, Cam-bridge, UK), we computed the local trabecular thickness(Tb.Th, �m)(19) to compare this parameter with the breadthof the bridge.

RESULTS

Bridges were commonly observed in human vertebralcancellous bone. We found 396 bridges. Examples ofbridges are shown in Fig. 3; some are very recent (Figs.3G–3I), and some are older (more mineralized; Figs. 3D and3E). Of 2363 images, there were 286 with one bridge, 48with two bridges, and 5 with three bridges. By comparison,there were only 15 microcalluses (5 in subject 2, 2 in subject3, and 4 in subjects 6 and 7).

Bridges were found in the cancellous bone of all the ninesubjects included in this study (Fig. 4). Most of the subjects(7 of the 9) had 39 or more bridges. Only 16 bridges werefound for subject 2 because the lumbar vertebra was ex-cluded from the study (hence, 20 blocks were observedinstead of 32).

FIG. 1. BSE image of a typical crescent shaped BSU (arrows).Osteocytes lacunae are clearly visible as well as bone lamellation. Oldbone matrix (center of the trabecula) has higher gray level and highermineralization. The recently formed bone matrix is darker becausesecondary mineralization is still in process.

1248 BANSE ET AL.

The size of the bridges varied from 29 to 869 �m, with amedian value of 172 �m (Fig. 5). A total of 339 of the 396bridges were longer than 100 �m, and 139 of the 396 werelonger than 200 �m.

On comparing the Tb.Th with the breadth of the bridgesfor each subjects (Fig. 6), it appeared that for most of them,the mean breadth was 50–60 �m, which was significantlythinner than their respective Tb.Th. However, this was notsystematic: subject 5 had quite thick bridges with a meanbreadth of 83 �m, identical to the mean Tb.Th.

DISCUSSION

On considering the data presented in this study, we pro-pose reconsideration of the principle stating that perfora-tions are irreversible.

Morphology and BSE

Boyde et al. showed few BSE images of vertebral can-cellous bone, enhancing that the structural information con-tained in BSE images (namely microcallus or hyperminer-alized woven bone) should not be overlooked.(10) Thepresent study confirms that BSE images can be used forother purposes than determination of mineralization profile.Here, it was used locate and delimit the BSU (Fig. 1). BSUscan be identified using other methods as well(7,8) (to mea-sure the wall thickness), but with BSE microscopy, one cansee which area in the trabecula is recent and which one isolder. Restated, there is a direct access to the historicalaspect of bone remodeling within the trabecula itself. Ac-tually, when running a preliminary test on a few blocks, weobserved an unexpected image of recent BSUs connectingtwo older segments of trabecula (this bridge is presented inFig. 3A). This intriguing feature did not match the conceptof irreversible perforation,(6) and consequently, we chose toinvestigate this issue on a large scale.

Normal remodeling?

The first issue regarding the interpretation of our data iswhether it is either a rare, accidental, and bizarre observa-tion or a normal feature of vertebral cancellous bone remod-eling. An argument supporting the latter hypothesis is theabsolute frequency of the bridge. In one of six images, wecould see a bridge (396/2363). This rate should be comparedwith the much lower frequency of microcalluses (15/2363).

It was also much easier to find a repaired perforation (abridge) than a nonrepaired perforation (i.e., see Fig. 3J).Note that we were unable to quantify nonrepaired—oropen—perforations because their identification is much toosubjective.

Atypical observation could also result from the origin ofthe material. We studied autopsy specimens, and one or twosubjects could have an unknown bone-related disease. How-ever, we found bridges in all nine subjects. It is unlikely thatall autopsy subjects were in that situation. Furthermore, wedid not find any morphological arguments for bone diseasein this investigation or in other studies on the same sam-ples.(14,15)

Consequently, we propose that these images of BSUsconnecting two segments of trabeculae is characteristic ofnormal remodeling in the human vertebral cancellous bone.

Perforations: the dogma

The classical view of bone remodeling cycle is that theosteoblasts refill the cavity created by the osteoclasts. Boneapposition happens in the concave surface of this cavityuntil partial or total refilling. When trabeculae are thickerthan the depth of this cavity, the remodeling cycle leaves thearchitecture of the network unchanged. There is a substan-tial scatter in the depth of the cavities(2,20,21) and in the localthickness of a trabecula,(22) making the occurrence of fen-estration or perforations likely to occur.(5) This is even morelikely in the vertebral bone where the trabeculae are verythin. Now, what about bone apposition on the perforatedtrabecular plate? Accepting the principle that bone has to bedeposited on a pre-existing bone surface, many authors haveconsidered the perforation or fenestration of the trabeculaeas irreversible.(2,20,22,23) Indeed, this was consistent with theprogressive age-related loss of entire structural elements.(24)

Overall, it is generally believed that large perforations(100–400 �m) are irreversible alterations of bone architec-ture.

Morphology of the bridges

The measured length of the bridges (the size of therepaired perforation) precisely covers that 100- to 400-�mrange (Fig. 5), supporting the hypothesis that the BSUsdefined as bridges in this study are repaired perforations. Inaddition, signs of osteoclastic resorption on both segments

FIG. 2. (A) Typical presentation of the featurearbitrarily defined as a bridge (Br) in this paper.It is a BSU that connects two segments of tra-beculae (ST1 and ST2). These segments havebeen disconnected by osteoclastic resorption, asdemonstrated by the scalloped border of the in-terruption of the cement lines. Oppositely, theexternal surface of the bridge is smooth showingno signs of resorption. (B) The arrows indicatehow the length and breadth of the bridge weremeasured

1249IRREVERSIBLE PERFORATIONS IN VERTEBRAL TRABECULAE?

FIG. 3. Bridges (white arrows) were fairly easy to find in the vertebral trabeculae. (A) The first image represents the bridge that was fortuitouslyobserved during a preliminary test of the BSE-SEM. This single observation triggered the present study. A 1.2 � 1.2 mm field is selected fromthe original documents that served for the measurements. Note that variety in breadth, length, and degree of mineralization. All bridges respondto the selection criteria detailed in the Materials and Methods section.

1250 BANSE ET AL.

of reconnected trabeculae (Fig. 2A) were actually includedin the definition of the bridge. On comparing the meanbreadth of the bridges with the mean Tb.Th (Fig. 6), wefound that, for most subjects, the repaired segment wasthinner than the common trabecula. This means that eitherthe thinnest trabeculae have been selectively perforated orthat, for most subjects, the bridge is often thinner in its

central portion. Note that we always measured the breadthas the minimal diameter (often in the middle portion of thebridge) and that some bridges were very thin because theywere probably still in construction (low gray level, Figs.3F–3I).

Osteoid bridges

If the mineralized bridges, as seen on BSE images, rep-resent the new bone formed during a single wave of appo-sition, one should be able to find unmineralized bridgesformed of only osteoid bone. In a previous study, we havereported the assessment of new bone formation indices onthe same samples.(15) On reviewing these documents(trichrome Goldner-stained thin sections), we observed os-teoid bridges (Fig. 7). The frequency of osteoid bridges (onthe trichrome) was at least five times lower than that ofmineralized bridges (BSE). This is normal, because miner-alization of the osteoid takes place over few weeks, whilemineralized BSUs probably remain for years (before beingthemselves remodeled).

Orientation of the lamellae

Generally, new bone is deposited with lamellae parallel tothe eroded bone surface. In a perforated plate, new bonelamellae should be perpendicular to the big axis of theperforation (arrow for the length in Fig. 3B). However, onchecking the bridges at high magnification or using theorientation of the osteocytes lacunae (Figs. 3 and 7), wefound the lamellae axis to lie parallel to the big axis of theperforation.

Bridges and trabecular network

The problem of progressive disconnection of the trabec-ular network has been repeatedly described and illustrated

FIG. 4. Number of bridges for each of the nine subjects included inthe study. Most subjects had 39 or more bridges. *For subject 2, only20 blocks instead of the generally used 32 were examined, explainingthe relatively low number of bridges. The relatively high absolutefrequency of the bridge and the fact that they were found in all thesubjects supports the idea that it is a common feature of normal boneremodeling of human vertebral trabecular bone.

FIG. 5. Histogram of the length of the bridges. Small bridges werecommonly observed, but most bridges were longer than 100 �m.Dotted line is the mean trabecular thickness (85 �m).

FIG. 6. Comparison between the mean trabecular thickness (Tb.Th,filled circles) and the breadth of the bridges (open circles) for eachsubject included in the study. Error bar is � SEM. For most subjects,the mean breadth of the bridge was smaller than the mean trabecularthickness. However, subjects 2 and 5 made bridges with a breadthidentical to the calculated local trabecular thickness.

1251IRREVERSIBLE PERFORATIONS IN VERTEBRAL TRABECULAE?

on 2D documents.(25) Before the rise of three-dimensional(3D) morphology (with �CT for example), all connectivityparameters (Tb.N, Nd/Fe, Euler number, etc.) were obtainedfrom 2D images. Each bridge reported in this paper is likelyto increase the connectivity, as measured by these 2D pa-rameters. For example, in Figs.3A, 3G, 3H, 3I, 3K, and 3L,if there was no bridge, there would be two extra free-ends(Fe). In Figs. 3B–3F and 3J, no bridge would mean the lossof one node (Nd) and one extra Fe. Therefore, we canconclude that 2D disconnections seem to be reparable inhuman vertebral cancellous bone.

In 3D, interpretation is much more hazardous. Schemat-ically, aggressive resorption on thin trabeculae may lead totwo types of network “lesions”: perforations of the platelikeelements and true disconnection of the thin (i.e., horizontal)bars.(5,26) Throughout this paper, we deliberately chose areasonably conservative approach, considering the bridgesas “reparation of perforations.” For the bridge to correspondto a true 3D reconnection (namely a bridge between twomechanically disconnected parts of an horizontal bar), itshould be demonstrated that there was no bone above andbelow the level of the section, a point that will certainlyrequire further investigation.

Conclusion

Whatever the 3D interpretation, it becomes difficult toconsider the teams of osteoblasts as “cavity fillers” or “sur-face painters,” laboriously adding layers on an existing

surface. They obviously have the capacity to lay downpackets of bone that have a given shape and are going in agiven direction; therefore, they should be considered as“active builders,” probably capable of protecting the trabec-ular network against excessive structural damage.

ACKNOWLEDGMENTS

This work was supported by the National Funds forScientific Research (Belgium).

REFERENCES

1. Parfitt AM 1992 Implications of architecture for the pathogenesisand prevention of vertebral fracture. Bone 13(Suppl 2):S41–S47.

2. Parfitt AM 1983 The physiological and clinical significance ofbone histomorphometric data. In: Recker RR (ed.) Bone Histomor-phometry: Techniques and Interpretations. CRC Press, Boca Ra-ton, FL, USA, pp. 143–223.

3. Mosekilde L 1988 Age-related changes in vertebral trabecularbone architecture–assessed by a new method. Bone 9:247–250.

4. Thomsen JS, Ebbesen EN, Mosekilde L 2002 Zone-dependentchanges in human vertebral trabecular bone: Clinical implications.Bone 30:664–669.

5. Dempster DW, Shane E, Horbert W, Lindsay R 1986 A simplemethod for correlative light and scanning electron microscopy ofhuman iliac crest bone biopsies: Qualitative observations in normaland osteoporotic subjects. J Bone Miner Res 1:15–21.

6. Parfitt AM 1984 Age-related structural changes in trabecular andcortical bone: Cellular mechanisms and biomechanical conse-quences. Calcif Tissue Int 36(Suppl 1):S123–S128.

7. Eriksen EF, Gundersen HJ, Melsen F, Mosekilde L 1984 Recon-struction of the formative site in iliac trabecular bone in 20 normalindividuals employing a kinetic model for matrix and mineralapposition. Metab Bone Dis Relat Res 5:243–252.

8. Villanueva AR, Ming-Cai Q, Parfitt AM 1985 Mean wall thicknessin gallocyanine stained sections: Comparison with ultraviolet light(UVL) and polarized light (PL) microscopy. Bone 6:412.

9. Boyce TM, Bloebaum RD 1993 Cortical aging differences andfracture implications for the human femoral neck. Bone 14:769–778.

10. Boyde A, Elliott JC, Jones SJ 1993 Stereology and histogramanalysis of backscattered electron images: Age changes in bone.Bone 14:205–210.

11. Grynpas MD, Holmyard DP, Pritzker KPH 1994 Bone mineraliza-tion and histomorphometry in biopsies of osteoporotic patientstreated with fluoride. Cell Mater 4:287–297.

12. Roschger P, Fratzl P, Eschberger J, Klaushofer K 1998 Validationof quantitative backscattered electron imaging for the measurementof mineral density distribution in human bone biopsies. Bone23:319–326.

13. Banse X, Sims TJ, Bailey AJ 2002 Mechanical properties of adultvertebral cancellous bone: Correlation with collagen intermolecu-lar crosslinks. J Bone Miner Res 17:1621–1628.

14. Banse X, Devogelaer JP, Grynpas M 2002 Patient specific micro-architecture of the vertebral cancellous bone: A pQCT and histo-logical study. Bone 30:829–835.

15. Banse X, Devogelaer JP, Lafosse A, Sims TJ, Grynpas M, BaileyAJ 2002 The crosslink profile of bone collagen correlates with thestructural organization of the trabeculae. Bone 31:70–76.

16. Schenk RK, Olah AJ, Herrmann W 1984 Preparation of calcifiedtissues for light microscopy. In: Dickson GR (ed.) Methods inCalcified Tissue Preparation. Elsevier, New York, NY, USA, pp.1–56.

17. Hahn M, Vogel M, Amling M, Ritzel H, Delling G 1995 Micro-callus formations of the cancellous bone: A quantitative analysis ofthe human spine. J Bone Miner Res 10:1410–1416.

18. Cheng XG, Nicholson PH, Lowet G, Boonen S, Sun Y, Ruegseg-ger P, Muller R, Dequeker J 1997 Prevalence of trabecular micro-callus formation in the vertebral body and the femoral neck. CalcifTissue Int 60:479–484.

19. Parfitt AM, Mathews CH, Villanueva AR, Kleerekoper M, FrameB, Rao DS 1983 Relationships between surface, volume, andthickness of iliac trabecular bone in aging and in osteoporosis.

FIG. 7. Example of unmineralized bridge made of osteoid bone(trichrome Goldner thin section, bar � 100 �m). At this stage of bridgeformation, the bridge could not be seen with the BSE microscopebecause it has the same density as the plastic resin. *An osteoclasticresorption cavity and **another osteoid seam are visible.

1252 BANSE ET AL.

Implications for the microanatomic and cellular mechanisms ofbone loss. J Clin Invest 72:1396–1409.

20. Thomsen JS, Mosekilde L, Mosekilde E 1996 Quantification ofremodeling parameter sensitivity–assessed by a computer simula-tion model. Bone 19:505–511.

21. Lips P, Courpron P, Meunier PJ 1978 Mean wall thickness oftrabecular bone packets in the human iliac crest: Changes with age.Calcif Tissue Res 26:13–17.

22. Compston JE, Mellish RW, Croucher P, Newcombe R, GarrahanNJ 1989 Structural mechanisms of trabecular bone loss in man.Bone Miner 6:339–350.

23. Reeve J 1986 A stochastic analysis of iliac trabecular bone dy-namics. Clin Orthop 213:264–278.

24. Thomsen JS, Ebbesen EN, Mosekilde L 2002 Age-related differ-ences between thinning of horizontal and vertical trabeculae inhuman lumbar bone as assessed by a new computerized method.Bone 31:136–142.

25. Compston JE 1994 Connectivity of cancellous bone: Assessmentand mechanical implications. Bone 15:463–466.

26. Mosekilde L 1990 Consequences of the remodelling process forvertebral trabecular bone structure: A scanning electron micros-copy study (uncoupling of unloaded structures). Bone Miner 10:13–35.

Address reprint requests to:Xavier Banse, MD, PhD

Universite Catholique de LouvainOrthopaedic Research Laboratory

Tour Pasteur 5388Avenue E. Mounier 53

B-1200 Bruxelles, BelgiumE-mail: xavier.banse@orto.ucl.ac.be

Received in original form August 13, 2002; in revised form No-vember 18, 2002; accepted January 21, 2003.

1253IRREVERSIBLE PERFORATIONS IN VERTEBRAL TRABECULAE?