Age changes in bone

A. Boyde' and V.J. Kingsmill' ^

'Hard Ti.ssue Research lliiii. Depurtmcnl ol'Aniitomy and Developmenlal Biology, UiiiNcrsily College Liinddii, WCIIt 6BT. Deparlment of Conscrvalivi- ncniisiry, St Burlholoniew's and Ihe Koyal London School of MfilicinL- and Dcnlislry, I urner Street. London, Kl 2AD.

Abstract Changes in bone structure as a function of age have been studied by simple inspection, x-ray imaging, stereo-photography, deep field optical microscopy, circularly polarised light microscopy, and scanning electron microscopy (SEM), including both topographic and compositional backscattered electron (BSE) imaging modes. The study of bone as a three-dimensional object, rather than in thin sections, enables us to envisage modelling and remodelling processes in context. The study of ultra-flat block surfaces permits the acquisition of data from an effectively very thin layer in the block face, and to examine bone as a spectrum of tissue types varying in the degree of mineralisation. Particular attention has been paid in our earlier studies to the iliac crest, lumbar vertebral bodies, femoral mid-shaft, neck and head and parietal and frontal skull bones. Recently, we have compared findings from these sites with observations on the mandible. We conclude, from our new imaging data, that common generalisations about the changes in bone in ageing and osteoporosis are too simplified, and that the mandible differs sufficiently from postcranial skeletal sites that it would be unwise to extrapolate from findings in the jaw to the circumstances elsewhere.

Keywords: Osteoporosis, bone turnover, bone density, mineralisation, ageing

Commonly embraced views of bone turnover, remodelling, ageing, and osteoporosis accept the basic precept that bone is removed within a given area by osteoclastic action and that, after a relatively brief interval, bone is replaced in the same domain by osteoblastic activity. A net loss of bone mass will happen if the total amount of bone replaced is less than that removed. Significant changes in cancellous bone architecture will result if, for example, the randomly located resorption events happen upon both sides of a plate-like structure, thus causing the removal of a part of the plate so that it can no longer be reconstructed because the local template is missing. Whereas these basic propositions are, to a first approximation, true in the younger mature adult individual, they fall far short of providing an exact description of the processes in older humans and do not describe, and cannot account for, the development of the changes seen in ageing and osteoporosis. Osteoporosis critically affects certain skeletal

sites such as the bodies of the vertebrae - resulting eventually in crush fractures, and the neck of the femur - with often disastrous consequences following a fall. Other very well studied sites are the wrist - the distal radius - again because of fractures induced by falling, and the calcaneus, but only because it is accessible to measurement by ultrasound. Another site which has been very extensively studied is the iliac crest because it is considered safe and acceptable to remove cylindrical biopsy cores from this regions for histopathological study of ongoing bone disease in the living patient. The jaws must be the most imaged of all the bony structures which are studied radiographically in life, and it has often been proposed that such dental x-rays might be valuable in detecting generalised osteoporosis. This would require that the changes in the jaws would match those in other skeletal sites, but our studies ha\e demonstrated that this may not be true ' - -. The human bone histomorphometry literature is based mainly on the cancollous bone in the iliac

Volume 15. No. I

Ml>\ill-'.mil V I

crosl hiopsy site. The usu;il lighl microscopic hisionioiphoinciiK doicnninallon ofbone volume traction lioin liuhi iiii(.n>scopiL' sections can give .iccuiatc rcsiilis, bul \cry sm;ill and unicprcscnlati\i.' \oluiiK'sortissue ;uv sampled by such pioccdmvs. Several authors have attempted to study features related to the intcivoniiectedness ot trabeculae. but h a \ e ignored the three dimensional aspects ol the beautiful, anisotropie architecture which is found at the site and have emploNed \ie\\s derived from unknown random rotations around the long axis of the cylindrical iliac crest biopsy. Every single apparently separate piece of bone in a section is connected to all the other bone, even if, in a minority of circumstances in osteoporotic bone, some trabeeulae have been cut through hy osteoclastic resorption and have unloaded ends. The best means of studying tissue organisation will be via complete 3D reconstructions provided by cutting as many phvsical sections as are needed to encompass whole bones^ or by reconstructions made by high resolution ('micro') computed tomography\ However, not all laboratories will be able to afford the capital outlay to invest in these methods, and they have their own limitations, particularly as regards resolution, though this problem is not to be compared with the serious lack of fine structure discrimination prevailing with clinical (in vivo) imaging modalities. Take, for example, the human lumbar vertebral bodies, which appear to have substantial cortices in plain radiography and in computed tomography imaging: infact their cortices are hardly thicker than their trabeeulae. Post mortem, direct inspection and optical imaging of macerated slices (i.e. the marrow has been removed^) remains the cheapest and most effective method to study bone architecture and porosity: for more detail, the depth of field and resolution of scanning electron microscopy (SEM) still make it the method of choice in this problem area^. Naked eye and magnifying glass magnifications of sectioned lumbar vertebrae show that the body portion is entirely cancellous, with a thin cortical shell. Taking as a standard a 4 mm thick slice, only a few pore spaces give a line of sight through a vertical section from a young mature individual, (Eig 1 see page 27). A similar slice from the same bone of any elderly individual appears to be 'porous'(Eig 2 see page 27). and this includes those bones without any signs of collapse or gross morphological change of the vertebral body and even when there is still a substantial volume fraction of bone. This reflects normal age-related

changes in the micro-architecture of the cancellous bone. Cancellous bone in the vertebral bodies of children has a foam-like structure, with more bone at the linear junctions of adjacent marrow chambers, forming short rod-like elements since the bony walls between are perforated*. In longitudinal sections, it may be difficult to recognise the load-bearing direction. However, this is recognised unambiguously in adult vertebral body trabecular bone architecture (Fig lj. In the younger mature adult, the tissue resembles a vertical honeycomb with holed walls between chambers and is evidently 'porous' in thick transverse sections. The cancellous bone in an ageing individual is a scaffolding of rods. These architectural changes are associated with loss of bone volume fraction, but the increase in the extent of line-of-sight, pore channels is greater in proportion than the loss of bone mass. In the older individual, the same plate-like cell walls of the honeycomb have been perforated and the structure remodelled and refined to make one of more rod-like elements. Those rods may be quite robust, with the vertical 'load-bearing' more so than the horizontal struts (Eig. 3 seepage 28). The clear visualisation of the 3D surface extent of new bone packets given by SEM shows a generalised deficit in their coupling to areas of resorption, so that large, unrepaired, resorbed areas become a dominant feature. Vertical load bearing struts may be fortified with a high proportion of new bone which, from its characteristic fine structure, can be seen to have been added initially as microcallus'^. Digital image analysis of high resolution x-ray images of plane-parallel slices is a good way of averaging local bone mass content, if we accept the usual premise that bone is a relatively uniform mixture at the microscopic scale. Eigure 4(a,b,c,d see page 29). shows the extent to which the net radiodensity varies within regions, even when these are rather substantial. Bone mineral density or volume fraction in sub-volumes of roughly ( 4 x 4 mm in area x 3 mm in depth) 48 mm-* varies by factors of 4:1 within the slice shown in Eigure 4(ad sec page 29). A subset of x-ray images of 2 mm thick cross sections of one partially dentate mandible from a 72 year old male, ranging from the retromolar region on the right to the first premolar region on the left, are shown in Eigure 5 (see page 32). As with Eigure 4 (see page 29), these images were acquired on high resolution film with a low kV (25kVp) source, and then digitised to 16 bits using a cooled CCD camera. The grey levels were

Gerodontology The Gerodoniology Association 1998

Age (

cs It) bone

27

k'J 1(1 S bit ivsolulion, subti.n. liiii' oil the grey \ ; i l i u - l o u n d l o r / c m b o n e , ; I I K I scllinL' ihc nuixiiiuim \;iluc (found in ihc iiiosi solul coilical hone) lo 2>.'^. 1 ho s;inic iiiuiiicsaiv liisl show ii with the grey sc;ilc (0 - 2^,^) modiliecl to show a pink l-i.K ki:n>iind w here \ui bone is piesenl: sueh leL-ions can ne\ei be idcnliHed h\ e\e in plain monochrome iiiiai:es Hee;iuse o f the slopes ol" eilges (partial \ o l u n i c effect), all substantive regions o f bone appear to be less dense at edges. Secondly, the images are shown w iih a colour look up table lo

iii.ihle Ihe idenlil K iilion ol legions with isoi;ulio(lensilies. Less dense central cortical ngions IKU- .III- cxpl;iincd partially by the non panillclity ol sdiiieol Iheseelions. However, ihe lowerdensity oT Ihe hone in alveolar ridge regions, even wheir no teeth were piesenl, is clearly demonslraled iHiiiherniore, extra-dense patches arc seen on the buccal side ol the cross seclion through the position of the missing lowcrriglii Ursi molar These patches may consist largely ol dead bone, as evidenced by more detailed study using quantitative

Figure I . Human fourth lumbar \ c r i c b r a l h.Hl> 30 year o l d female, photo ..I 4 m m thick vertical slice ajiainsi black paper ( g r i d at left in m m ) . Figure 2. H u m a n lourth lumbar vertebral body H9 year o l d female, photo ol 4 m m thick vertical s l u e

Volume !.'>. N o 1

A Ho\ili \ i i u l \ 1 Kiii)'smiir

Figure 3. 2.8 mm wide, deep and high volume of cancellous bone from the fourth lumbar vertebral body of an 89 year old female (cleaned with proteolytic enzyme and gold and carbon coated), imaged with secondary electrons (5 kV electron beam) to show surface detail and with primary transmitted electrons to discriminate line of sight pore space, here coded as uniform background colour. The vertical load bearing direction is left-right. Extensive areas of bone have been resorbed (e.g., the centra] and right hand regions of the uppermost trabecula). An area of consolidated microcallus increases the girth of the central trabecula just left of centre. Resorbed ends of (prior) rod shaped transverse trabeeulae can be seen in the central height domain (making it unlikely that they are so because of preparation damage). Extensive regions of the surface ha\ e the morphology typical of arrested mineralisation fronts which are common in elderly human bone.

backscattered electron (QBSE) imaging in SEM (vide infra). The bulk of the hard tissue in human bones is lamellar bone (mostly secondary-osteonal in nature, with extensive regions of circumferential lamellae in periosteal surfaces), but this bone is not a uniform material. Outwardly similar areas may differ on the basis of collagen fibre orientation

patterns, and, within such areas, neighbouring regions may have different levels of mineralisation. It is generally presumed that osteonal lamellae demonstrate a criss cross, plywood structure, the collagen bundles in alternating lamellae having contrasting orientations. However, extensive studies using quantitative circularly polarised light

Gerodontology
The Gerodontology Association 1998

Age

Ill lioiu'

2''

inieioseop\ ol l()()|jm thiek, plane parallel seelions ol long bone sluiKs have denionslraleil that there are important ditferenees in ihe si.iiistical aspect ol collagen orientation in regions subjecletl lo predominanlK tensile as against allei nali\ ely coni|-ircssi\e load'"" Siinilai detailed analyses h a \ e nol \et been pubhshed lor human jaw hone Mineral replaces water in bone, an inereasing eontent making il ineieasmgl\ still. Thus this, together with the eollagen w e a \ e , is also an important parameter in detertnining the local meehanieal properties ol the tissue. Si nee the earliest da\ s ol niicriiradiograplix. it has been k n o w n that the b o n e m i n e r a l c o n t e n t in a

niicii>seopie volume ol this ti^sue generally inereasrs with the age ol the tissue eleiiiint: in wly deposited bone paekets aie less well mmcialised. Thai t l u r e has to he an o p i i m n m level ol mineralisaiioii can be seen by the study ol diseases where this is nol aeliieved. In the lailure ol mineralisation in liekels and osleomalaeia. the bone IS, and the hones are, soil and bend(s) under load. In several different types ol osleogenesis imperlecta, the tissue reaches a higher level ol mi Derail sal ion than in normal aged-matched hone, which, together with its reduced bulk, contributes towards the hriltlcness characterising the bones in this class ol disorder'-. This was determined hv

i' 4. SM year old Icmale fourth lumbar vertebral budy. non-usteoporotic. Top left: (.ligilised \ ras of 3 mm section. Top right: pseudo-culour coded. Bottom left: he.i\y median Illteringof thelrabecular regions (50 repe.ilsol a 7 \ 7 iieighhmirhdnd. using a routine to avoid blurring aemss real edges) The localmn nl fi4 regions (each approxinuilely 3 \ 4 \ 4 mm) m which the mean pixel intensity in the trabccular bone legmns wa.s determined prior to image smoothing . Boiiom right: The local mean Iraetional bone volume on a percent scale from zero ino bone) to 100 (solid cortical tissue) for the trabecular liekK sh<l\^n al bottom left.Field is 32mm wide.

Volume 15. No.

,iiul VI

ihc relative dcjiivc of mineralisation of Itonc ;ii the microscopic scale by quanti(a(iiig the lunnbcr ol clccli\)Ms backscallcrcd Irom the (issue". We lui\e siiidicd large series of hiiiniin iliiic cresi iMopsN and autopsy samples, seeond and fourth lumbar vcrtcbtal body and femoral neck aut)psy samples usinu QBSh. The results document the normal range of variation of tissue composition and the normal proportions o\ volume fractions of bone having different compositions from these skeletal sites, fhey also show the generally greater levels of mineralisation in calcified hyaline and fibrocaililage within bone organs compared with the bone tissue proper. With more limited material, we v v ere able to compare tissue density distribution in autopsy samples of mandibles (at sites near the midline. near the mental foramen, and near the third molar), the parietal bone, femoral neck, iliac crest and fourth lumbar vertebral body sites from the same individuals^ To perform QBSE studies, it is necessary to fill marrow and cell space with a material of uniform composition which is both vacuum- and electronbeam-stable, since the specimen must not have surface-topographic features if we are to exploit the composition dependency of the backscattered electron (BSE) signal. For this purpose, we polymerise methylmethacrylate monomer in situ having first removed water and fat with ethanol. We use micro-milling (Reichert-Jung UltraMiller, Leica. UK) to finish the PMMA blocks, aiming to produce surfaces which are generally flat to 0, l|im. Using a very stable digital automated SEM (Zeiss DSM962 with Kontron IBAS external control computer and a KE Electronics Ltd annular solid state BSE detector), we can then compare signal brightness from one spot to another. This depends principally on variations in composition of the tissue, and in particular on the mineral content. We standardise our observations by referring BSE signal values to those obtained from suitable standard compounds. To avoid 'channelling' contrast, the standard sample must have no longrange crystallographic order, Eor this purpose, we have adopted the brt)minated and iodinated dimethacrylatc resins synthesised by KMW Davy'\ Eigure 6 (seepage 33) contrasts the findings from 2,7mm wide fields from the mandibular midline and the iliac crest: although these stem from a single individual (an 82 year old female), they are nevertheless representative of our fmdings'. We divide 'bone' into a number of sub-classes depending on the standardised, measured QBSE value. Using the same colours to show bone in

equivalent portions of the .spectrum of possible mineralisation densities, it is seen that the iliac crest (top left image with histogram at bottom left) has a generally broader range, with far more 'normal' medium density bone in range 5 of an 8 range scheme. The mandible (right two figures in Figure 6 (sec page 33)). has a much higher fraction of dense and very dense bone. We emphasise that these estimations derive from very large numbers of repeated measurements from very small volumes. At about one cubic micron, the sampled points are generally within or outside bone tissue proper, and the estimate is not seriously affected by partial volume effects. The volumetric density of these bones may, of course, also differ. In this particular example, the reduced area under the histogram of the iliac crest image reflects the lower volume fraction of mineralised bone tissue in this field. Thus the QBSE studies show that there is an increased proportion of highly mineralised bone in the elderly human mandible compared with the sites (spine, hip, wrist) most commonly symptomatic in clinical osteoporosis. Radiographic density data from the elderly mandible will be biased by this shift in tissue composition compared with other skeletal sites. In addition, the mandible differs in terms of apparent density (gms/ml) and volume density (mV ml)'. To emphasise this point, Eigures 7 and 8 (see page 34). show slices of the same thickness of the end plate of a fourth lumbar vertebra and the edentulous mandible from the same 91 year old male. Other interesting features of geriatric human mandibular bone include extensive areas of dead tissue. Elderly human mandibular bone may contain significant tracts carrying mineralised osteocytic lacunae and canaliculi (Eig 9 se^ page 34). Calcified contents of former Haversian canals are not uncommon'. Conclusions The QBSE density data suggest that the stiffness of the bone tissue fabric in the ageing human mandible will be comparatively high. The stiffness of the whole organ will be modulated by the anatomy of the bone, including the high apparent density and volume fraction. Any changes in bone mineralisation density and microarchitectural arrangement of edentulous regions of the mandible will have an obvious bearing on the practice of dental implantology. It should also be bome in mind that the dense, relatively avascular, aged mandible may not have the same healing potential

Gerodontology
The GcriKiontology Associalion 1998

Age changes jn bone

as other sites at other ages. In the femur, where orthopaedie replacements of the head with ii stem prosthesis are exceedingly common, a resulting local porosity of formerly compact cortical bone is the rule rather than the exception. This is the consequence of stress shielding in the bone segment by the stiff metal stems ofthe hip implant. It is unlikely that a similar consequence will be encountered for single dental implants. Increasing gross porosity of bone is a common feature at those skeletal sites which are subject to osteoporotic fracture. Similar changes are not the norm for the mandible. In addition, there is a trend towards an increase in the number of defective joins between old and new bone packets in lumbar vertebral body

bone in the ageing human, with some incompletely mineralised tissue incorporated at the level of erstwhile cement lines". Far Irom this bciriy the case in the mandible, impressive accumulation ol den.se resting cement lines are common. Acknowledgements The facility for quantitative analysis of bone mineralisation density at the microscopic scale was established with MRC funding. The micromilling equipment was provided by the Wellcome Trust. We also thank the Horserace Betting Levy Board for funding teehnical support, and Roy Radcliffe and Maureen Arora for their technical assistance.

References 1. Kingsmill V J, Boyde A. Variation in the apparent density of human mandibular bone with age and dental status. JAnat 1998a; 192: 233-244. 2. Kingsmill V J, Boyde A. Mineralisation density of human mandibular bone: quantitative backscattered eleetron image analysis. J Anat 1998b; 192: 245-256. 3. Kingsmili V J, Jones S J, Boyde A. Mineralisation density of aged human mandibular bone. In: Biological Mechanisms Of Tooth Eruption, Resorption And Replacement By Implants, eds. Davidovitch Z and Mah J, Boston Mass: Harvard Society for the Advancement of Orthodontics, 1998; pp 167-171. 4. Odgaard A. Three dimensional methods for quantification of cancellous bone architecture. 1998; 20: 315-328. 5. Muller R, Van-Campenhout H, Van Damme B, etal. Morphometric analysis of human bone biopsies: a quantitative structural comparison of histological seetions and miero-computed tomography. Bone 1998; 23: 59-66. 6. Jayasinghe JAP, Jones S J, Boyde A. 3D photographic study of cancellous bone in human fourth lumbar vertebral bodies. Anat Emhryol 1994; 189: 259-274. 7. Boyde A, Jones S J. Scanning electron microscopy of bone: instrument, speeimen and issues. Microsc Res & Techn 1996; 33: 93-120. 8. Kneissel M, Boyde A, Hahn M, et al. Ageand sex-dependent cancellous bone changes in a 4000y BP population. Bone 1994; 15: 539-545. 9. Hahn M, Vogel M, Amiing M, et al. Microcallus formations of the eancellous bone: a quantitative analysis of the human spine. J Bone Miner Res 1995; 10: 1410-1416. 10. Riggs C M, Lanyon L E, Boyde A. Funetional associations between collagen fibre orientation and loeomotor strain direction in cortical bone of the equine radius. Anat Embryol 1993a; 187: 231-238. 11. Riggs C M, Vaughan L C, Evans G P, Lanyon L E, Boyd A. (1993b) Mechanieal implieation of collagen fibre orientation in cortical bone of the equine radius. Anat Embryol 1993b; 187: 239-248. 12. Boyde A, Travers R, Glorieux F H, Jones S J, The mineralisation density of iliac crest bone from children with osteogenesis imperfecta. Calcified Tissue Intemational (in press). 13. Boyde A, Davy K W M, Jones S J. Standards for mineral quantitation of human bone by analysis of backscattered electron images. Scanning 1995; 17 Suppl. V:6-7. 14. Jayasinghe J A P , Jones S J, Boyde A. Seanning eleetron microscopy of human vertebral trabecular bone surfaces. Vircho\\s.\nhi\A Puthol Anat 1993; 422: 25-34.

Volume 15, No. 1

'^

\ UoNili

.mil \ .1 Kinjjsmill

47m

46d

Figure 5. Dignised and prdLesscil x-ray imajies ol 2 mm thiek LTUSS seeiinns of mandible frum ,i 72 \eai old male I n si three are from right retronmlar region, other locations are labelled aeemdmi; in standard scheme (-) is lower right 8 is third molar e k ) Suffix m = mesuil. d = disial. c = central. Further explanatidn in main text.

The (icrii(Junliili)gy Associanon I'^'

'i III bone

^ 1

Figure 6. Top left: iliac crest cortex, QBSE image ( a dense stereological grid of 512x512 measurements) scaled from 0. being the value found for a monobrominated dimethacrylate standard to 255 for a monoiodinated standard. 20 kV beam, field width = 2,7 mm. Top right: mandible, near midline QBSE image, 20 kV beam, field width = 2,7 mm. Bottom left: histogram of iliac crest image, showing the relative frequency of bone mineralisation density. Most points fall in range 5, Bottom right: histogram of mandible image. Most points fall in ranges 6 and 7, high and very high densities, Eurther explanation in main text.

Volume 15, No, 1

\ liovdc ;inil \

Figure 7. Wh^lc mandible cruss-scciion cleaned hy rtxim temperature plasma ashing, gold sputter coated. IN ing on lead background which gives higher BSE signal and appears white in this 20 kV BSE-SEM image. Field is 11.4 mm w ide.

Figure 8. Vertical section of end plate region of L4 body of 91y male, trom same individual imaged at same magnification and under the same conditions. Original magnification was lOx.

Figure 9. Alveolar crest region in the cross section of a PMMA embedded, micromilled. carbon eoated cross section of the mandible of an 86 year old male, showing extensive regions of high density (whiter) bone with many osteocytic lacunae (and canahculi, though these are not discriminated at this resolution) filled with mineral. Scale bar for 2(X)mn in image

Gerodontology
The Gerodontology Association 1998