Author's personal copy - geosoft.ru 2010_Biomaterials_.pdf · Various differentin vitromodel systems for the study of osteo-clasts have been described. Besides the cultivation of

Author's personal copy

A comparison of osteoclast resorption pits on bone with titaniumand zirconia surfaces

Thomas Hefti a,b,c, Martina Frischherz a, Nicholas D. Spencer b,*, Heike Hall a, Falko Schlottig c

aCells and BioMaterials, Department of Materials, ETH Zurich, Wolfgang-Pauli-Strasse 10, CH-8093 Zürich, Switzerlandb Laboratory for Surface Science and Technology, Department of Materials, ETH Zurich, Wolfgang-Pauli-Strasse 10, CH-8093 Zürich, Switzerlandc Thommen Medical, Hauptstrasse 26 d, CH-4437 Waldenburg, Switzerland

a r t i c l e i n f o

Article history:Received 8 April 2010Accepted 1 June 2010Available online 7 July 2010

Keywords:OsteoclastOsseointegrationSurface roughnessTitaniumZirconiaBone-remodeling

a b s t r a c t

Osteoclasts resorb bone at surfaces, leaving behind pits and trails where both mineral and organic phasesof bone have been dissolved. Rough surface structures are deliberately imparted to synthetic implants, inorder to improve osseointegration. The aim of this study is to characterize osteoclastic resorption pits onnative bone surfaces and to compare these with state-of-the-art titanium and zirconia implant surfaces.The size (i.e. length, width and depth) of resorption pits was compared to the size of surface features ofsandblasted and etched titanium and zirconia surfaces. It was found that resorption pits from native boneand surface features of the sandblasted and etched titanium and zirconia surfaces were quite similar intheir dimensions. Most structures showed a length between 5 and 40 mm, a width between 2 and 20 mmand a depth between 1 and 8 mm. Additionally, the wavelength-dependent surface roughness wasmeasured, revealing an Sa value of 60 nm in the resorption pits, 86 nm on zirconia and between 127 and140 nm on titanium surfaces. The results of this study may provide some insight into structuralrequirements for the bone-remodeling cycle and help to improve the design of new implant surfaces forosseointegration applications.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

In healthy bone, remodeling is a continuous process that adjustsbone architecture to meet current mechanical needs as well asrepairing (micro-) damage [1]. Bone-remodeling can be roughlydivided into two opposing but highly balanced processes, boneresorption performed by osteoclasts and formation of new bone byosteoblasts [2].

When attached to bone, differentiated osteoclasts form a closedcompartment beneath them, in which the mineral phase of bone,mainly hydroxyapatite, is resorbed by means of acid, and theorganic phase, mostly collagen I, is solubilised by proteolyticenzymes, such as lysomal cysteine proteinases and matrix metalloproteinases (MMPs). This concerted activity leaves behind resorp-tion pits and trails named Howships’s lacunae [3]. In the secondphase of the bone-remodeling process, bone lining cells clean theresorption pits of bone matrix leftovers such as collagen fragmentsand deposit a thin layer of fibrillar collagen type I [4]. Thereafter,bone tissue is rebuilt by osteoblasts [5], initiated by bonemorphogenic proteins (BMPs). These osteoblasts secrete collagen

type I as well as osteocalcin and osteonectin. Through regulation ofthe local concentration of calcium and phosphate, the formation ofhydroxyapatite is promoted [6].

Various different in vitro model systems for the study of osteo-clasts have been described. Besides the cultivation of primary cells,which are most often isolated from bone marrow [7], the mousemacrophage cell line RAW 264.7 is an established in vitromodel [8].Because the main focus in osteoclast research is to understandsystemic bone diseases such as osteoporosis [9], little emphasis hasbeen placed on understanding how surface features similar to thoseproduced by osteoclasts might be helpful for improving theosseointegration of an implant surface. Characterizing the surfacetopography of osteoclastic resorption pits and trails could providevaluable information, since it is well known and accepted fromnumerous in vivo and in vitro studies that osseointegration, i.e. theattachment of bone tissue to an implant, can be improved by theuse of rough implant surfaces [10,11].

To measure and describe surface roughness and surface topog-raphy over a wide range of size scales, a variety of methods wasemployed, including white-light confocal microscopy, atomic forcemicroscopy (AFM) and scanning electron microscopy (SEM). Withnormal SEM, surface features can be reproduced with a high xeyresolution but without the ability to quantify the topography in

* Corresponding author. Tel.: þ41 44 632 58 50; fax: þ41 44 633 10 27.E-mail address: [email protected] (N.D. Spencer).

Contents lists available at ScienceDirect

Biomaterials

journal homepage: www.elsevier .com/locate/biomater ia ls

0142-9612/$ e see front matter � 2010 Elsevier Ltd. All rights reserved.doi:10.1016/j.biomaterials.2010.06.009

Biomaterials 31 (2010) 7321e7331


three dimensions. However by using stereo-SEM, where twoimages of the same region of interest (ROI) are produced witha relative eucentric tilt (around an axis in the focal plane), 3D datacan be calculated. From these data, obtained with a high spatialresolution, topographical information, such as the depth of surfacefeatures or the surface roughness, can be calculated [12]. Thistechnique has been used to characterize osteoclastic resorptiontrails on bone and dentine [13,14].

Surface roughness can in general be described as a combinationof profiles spanning several orders of magnitude in size, of partic-ular interest being the micrometer-to-nanometer range. It cantherefore be useful to describe the surface roughness as a functionof wavelength, the so called wavelength-dependent roughness orwindow-roughness [15]. This has the advantage that small surfacefeatures (micro-roughness) on a macroscale rough surface can bereadily investigated after filtering out the longer wavelengths(macro-roughness).

It has been shown that osteoblasts react differently to surfacefeatures and surface roughness on different size scales [16]. Byusing the wavelength-dependent roughness approach, featuresthat are too large to be recognized by cells can be filtered out toreveal the cell-relevant surface profiles and roughness.

To the best of our knowledge, characterisation of the surfacefeatures of osteoclastic resorption pits has not previously beenperformed with the intention of influencing design of syntheticimplant surfaces for osseointegration applications. The aim of thisstudy is to compare resorption pit surface topography with surfacefeatures found on selected synthetic implant surfaces. To describesurface topography with a high x, y, and z resolution, stereo-SEMwas coupled with wavelength-dependent surface roughness anal-ysis. Tailoring an implant surface with features resembling nativeresorption pits or trails may be a successful approach to improvingthe osseointegration behaviour of implants.

2. Materials and methods

2.1. Preparation of bone slices

Bovine femurs were obtained from a local butcher. All residual soft tissue wasmechanically removed. Bone slices were sawn with a thickness of approximately0.5 mm from cylinders with a diameter of approximately 10 mm. Slices were cut inlongitudinal and perpendicular directions from the cortical region of a long bone. Forsterilization, slices were immersed in 70% ethanol for 24 h (at 4 �C), then rinsed inphosphate buffered saline (PBS) (Sigma, Buchs, Switzerland) containing a 1% anti-bioticeantimycotic solution (ABAM) (Invitrogen, Carlsbad CA, USA) (at roomtemperature) and immersed for another 24 h in PBS and 1% ABAM and then stored at�80 �C until use [17]. Before use, thawed bone slices were rinsed once with PBScontaining 1% ABAM at room temperature.

2.2. Cell culture of RAW 264.7 cells

RAW 264.7 cells (TIB-71; ATCC) were cultured in 5% CO2 at 37 �C in a-MEMmedium (Invitrogen, Carlsbad CA, USA) supplemented with 10% heat-inactivatedFBS (SigmaeAldrich, Buchs, Switzerland) and 1% ABAM. Cells were seeded onto thebone slices at an initial density of 1500e3000 cells/cm2. To analyze only effects ofbone-adhered cells, i.e. to prevent cultivation of cells on tissue culture plastic (TCP)and to remove non-adherent cells, bone slices were rinsed once with PBS andtransferred to a new well 2 h after cell seeding. To initiate differentiation intoosteoclasts, 50 ng/ml RANKL (Invitrogen, Carlsbad CA, USA) was added with themedium. Medium and RANKL were changed every 2e3 days. To achieve a highnumber of resorption pits on the bone surfaces, cells were cultivated for 14 days,subsequently being removed from the bone slices by ultrasonication in 0.25 M

ammonia [18]. A number of samples were treated in 10 mg/ml collagenase Type 2(Worthington, Lakewood NJ, USA) in PBS for 1 h at 37 �C. The bone slices were thenincubated in 2% osmium tetroxide, dehydrated in an ascending ethanol series andcritical-point dried in CO2 (CPD 030, Baltec, Switzerland).

2.3. Titanium and zirconia surfaces

Commercially pure titanium disks and yttria-stabilized zirconia (Y-TZP)(Thommen Medical, Waldenburg, Switzerland) with a diameter of 15 mm weretreated according to the steps described in Table 1. Sandblasting with alumina

particles of the indicated size was performed, followed by hot acid etching inamixture of hydrochloric acid (14%) and sulphuric acid (34%) for titanium. Etching ofzirconia was performed in an alkaline bath of NaOH and KOH (1:1) at 210 �C for 30 h.By XPS analysis, no residue from the etching process could be identified on thezirconia surface [19].

2.4. SEM analysis, stereo-SEM

SEM analysis (Zeiss SUPRA 50 VP, Zeiss, Oberkochen, Germany) was performedon all samples, the same ROI being imaged twice with the stage eucentricly tilted by10� between the two images. From these pairs of images, 3D data were calculated bymeans of specialized software (MeX from Alicona, Graz Austria). By examining false-colour images (Fig. 1, bottom row), surface features could be easily identified andlength, width and depth measured by drawing profile lines. The length was definedas the maximal extension of the surface features in the xey plane, and consequentlythe width was defined as the dimension perpendicular to it. The aspect ratio wascalculated as the ratio of length to width. The depth was defined as the maximaldepth along the length or width profile. The contour length of resorption trails wasdefined and measured along the presumed migration path of an osteoclast.

Surface roughness was measured using the window-roughness method, asestablished by Wieland [20]. Briefly, the overall roughness Sa of the titanium andzirconia surfaces was calculated from SEM images with 1000� magnification andwith a cutoff wavelength, lc ¼ 580 mm. The roughness within the individual surfacestructures and resorption pits was calculated in defined areas from SEM images with5000� magnification and with a cutoff wavelength, lc ¼ 2 mm. By filtering outwavelengths above lc ¼ 2 mm, the overall curvature of the surface features wasexcluded and only the surface roughness below this wavelength was measured (seeFig. 2 for illustration).

2.5. Statistics

The dimensions of the different surfaces features of native and syntheticsurfaces were classified into 14 bins, each with a bandwidth of 5 mm, 2.5 mm and1 mm for the length, width and depth, respectively. The last bin included all valuesabove the lower limit of this last class, which was 65 mm, 32.5 mm and 13 mm for thelength, width and depth, respectively.

The binswere plotted in histograms, which showed the frequency of appearanceof structural features. Additionally the median and the lower and upper quartiles(Q1 and Q3) of the data were calculated. Statistically significant differences betweenthe roughness values were determined with a Student’s t-test, assuming normallydistributed values. Statistical significance was assumed for p < 0.01.

3. Results

3.1. Qualitative characterisation of resorption pits on cortical bone

After removing the osteoclasts from the bone surface, traces ofresorption could be morphologically identified. Three types ofresorption structures were distinguished: resorption pits showeda roundish shape and most probably resulted from the proteolyticactivity of one single, non-migrating osteoclast (Fig. 1, top left).Resorption trails refer to structures most probably produced by oneosteoclast that migrated over the bone surface, leaving behinda continuous trail (Fig. 3). A third type demonstrated extensiveareas of resorption, where no distinction of single resorption pitsand trails was possible. Therefore these very dense areas ofresorption were not analyzed. Individual and clearly distinguish-able resorption pits and trails were spread evenly over the bonesurface, the pits outnumbering the trails by approximately 20:1.

Both resorption pits and trails exhibited a round contour, witha sharp border between resorbed and non-resorbed areas. Thefalse-colour images illustrate the original smooth bone surfaceembedded with the round contours of the resorption pits (Fig. 1,

Table 1Description of titanium and zirconia surfaces and associated treatments.

Shortname

Substrate Sandblasting Etching

TA Titanium grade 4 None Hot acid etchedTSA Titanium grade 4 Small grit (64e71 mm) Hot acid etchedTLA Titanium grade 4 Large grit (126e150 mm) Hot acid etchedZLA Zirconia (Y-TZP) Large grit (126e150 mm) Hot alkaline etched

T. Hefti et al. / Biomaterials 31 (2010) 7321e73317322


bottom). In the resorbed area, exposed collagen fibres becamevisible, since the mineral phase seemed to be resorbed faster thanthe organic collagen matrix. The orientation of the resorption trailsappeared to be random, neither following scratches or cracks onthe bone surface nor the direction of the collagen fibres (Fig. 3).Collagenase treatment removed the exposed collagen fibres toa large extent, although intact collagen fibres on the bottom of thepits remained intact. After collagenase treatment, the structurewithin the resorption pits became more dominated by the mineralphase of the bone (Fig. 1, right top).

3.2. Qualitative characterisation of the structured titanium andzirconia surfaces

Fig. 4 shows SEM images and false-colour images of all types ofexamined surfaces. Each image shows one example of a structuralfeature that was further analyzed. The false-colour images nicelyrepresent the topographical features of the surfaces. All surfaces

showed a rough topography: on the three sandblasted surfaces(TSA: titanium sandblasted with small grit and acid etched; TLA:titanium sandblasted with large grit and acid etched; ZLA: zirconiasandblastedwith large grit and alkaline etched) the traces of impactfrom the sandblasting treatment could be clearly identified, furtherthe difference in grain size used for sandblasting was clearly visiblewhen comparing the TSAandTLA surfaces. ZLA surfaces showed lesspronounced structural features from the sandblasting due to thehigher hardness of the material. The etched-only titanium surfaces(TA) showed a relatively homogenous surface structure displayingonly small topographic features. The hot acid etching of the titaniumsurfaces revealed a fine, spiky topography whereas the alkalineetching of the zirconia surface exposed single grains of zirconia.

3.3. Quantification of the dimensions of resorption pits and trails

The dimensions of the resorption pits left on the bone surfaceswere determined and are presented as histograms, comparing

Fig. 1. SEM images of native resorption pits on bone surfaces produced by osteoclasts. Left: Untreated resorption pits after 14 days of culture with RAW 264.7 cells. Right: resorptionpits after collagenase treatment. Top row: SEM images; bottom row: false-colour images from stereo-SEM analysis. Scale bar 25 mm.

T. Hefti et al. / Biomaterials 31 (2010) 7321e7331 7323


untreated and collagenase-treated resorption pits on bone surfaceswith synthetic titanium and zirconia surfaces (TA, TSA, TLA andZLA) (Figs. 5e7). Additionally the median and the first and thirdquartile (Q1 and Q3) were summarized in Table 2.

The length (maximal elongation in xey direction) of theresorption pits as shown in Fig. 5 was found to be between 5 and25 mm for the majority of the pits on native bone. The distributionshifted to larger values by approximately 5e10 mm for the colla-genase-treated bone samples, suggesting that the collagenasetreatment enlarged the perimeter of the pits. This is reflected inTable 2, where the first and third quartile (Q1 and Q3) and themedian shifted. The width of the resorption pits showed a distri-bution between 2.5 and 15 mm, againwith a slight increase inwidthof approximately 5e10 mm for the collagenase-treated samples. Asshown in Fig. 7, the values for depth of the resorption pits morethan doubled from the untreated bone to the collagenase-treatedbone surfaces. This is reflected in the median and the Q1 and Q3values shown in Table 2.

The median contour length of the resorption trails was 54.9 mm(n ¼ 21) with Q1 ¼ 34.7 mm and Q3 ¼ 104.53 mm, where the longesttrail measured had a length of 251 mm. The median width of thetrails was 15.6 mm with Q1 ¼ 12.9 mm and Q3 ¼ 22.3 mm and themedian depth was 1.5 mm with Q1 ¼ 1.3 mm and Q3 ¼ 2.2 mm.

3.4. Quantification of surface features of structured titanium andzirconia surfaces

On all surfaces, features between 5 and 100 mm were analyzed.In the histograms illustrating the distribution of the length (Fig. 5),the ZLA and TLA surfaces showed the widest distribution ofstructural features between 15 and 50 mm although the maximumnumber of structural features was found between 20 and 25 mm forZLA surfaces and between 40 and 45 mm for TLA surfaces. Thesecharacteristics in the distribution were not reflected in the mediannor in Q1 and Q3 values, which showed similar values for these twosurfaces. TSA and particularly TA surfaces showed a narrower sizedistribution around 25 mm and 10 mm, respectively. These findingsare reflected in the median and Q1 and Q3 values (Table 2).

The distribution of the width (Fig. 6) showed similar charac-teristics as described for the length with awide distribution for ZLAand TLA surfaces, however here the width on the ZLA surfaces washigher than on TLA surfaces visible in the histograms as well in themedian values (Table 2). TA and TSA surfaces showed a narrowerwidth distribution compared with TLA and ZLA. In particular, TSAsurfaces showed higher values than TA surfaces.

In Fig. 7 the distribution of the depth was found to be between 1and 4 mm for TA surfaces, increasing to 3e7 mm for TSA and ZLAsurfaces. TLA surfaces showed a wide distribution of values up to12 mm; this was reflected in the highest Q3 value of 9.7 mm.

The aspect ratios of the length and thewidth of TSA and TLA, thetwo sandblasted titanium surfaces were almost double those of theother surfaces (Table 2).

3.5. Surface roughness

The overall surface roughness Sa of all three titanium andzirconia surfaces was calculated from SEM images with a 1000�magnification and with a cutoff wavelength lc of 580 mm (Fig. 8).The surface roughness Sa was 1.10� 0.03 mm and 1.19� 0.40 mm forTA and ZLA, respectively. It increased to 1.86 � 0.11 mm for TSA andto 2.70 � 0.10 mm for TLA surfaces. The roughness of TLA surfaceswas significantly different (p < 0.01) from the other three surface

0 5 10 15 20 25

0

1

2

3

4

5

rela

tive

z [µ

m]

x [µm]

roughness with 2 µm cut-offroughness with100 µm cut-offoriginal profile

Fig. 2. Profile line through a native resorption pit (right) with corresponding profile and roughness profiles calculated for cutoff wavelengths of 2 mm and 100 mm, respectively (left).Scale bar 10 mm.

Fig. 3. SEM image of an osteoclastic resorption trail on a bone surface with randomchange in direction. Scale bar 50 mm.



types. TSA surfaces were significantly rougher than ZLA and TAsurfaces (p < 0.01). There was no statistically significant differencebetween the overall roughness of TA and ZLA surfaces.

The surface roughness within the resorption pits and in thestructural features of the titanium and zirconia surfaces was eval-uated from SEM images with a magnification of 5000� and wascalculated with a cutoff wavelength lc of 2 mm (for illustration seeFig. 2). As shown in Fig. 9, the surfaces were classified into threegroups according to their roughness (Group 1: contains both bonesurfaces displaying native resorption pits; Groups 2: contains alltitanium surfaces TA, TSA and TLA; Group 3: contains only zirconiasurfaces (ZLA)). The differences between the roughness values ofthe three groups were statistically significant (p < 0.001) whereasthey were not statistically different from each other within thegroup. The titanium surfaces showed the highest roughness values

with 127� 2.8 mm,128 � 9.8 mm and 140 � 12.8 mm for TA, TSA andTLA surfaces, respectively. The surface roughness Sa within thestructural features of zirconia surfaces ZLAwas 86� 4.8 mm and thesurface roughness values within the resorption pits on bone were60 � 11.8 mm and 67 � 14.8 mm for the untreated and collagenase-treated resorption pits, respectively.

Summarizing, it was demonstrated that surface features andsurface roughness found on resorbed bone surfaces and synthetictitanium and zirconia implant surfaces could be quantified andcompared.

4. Discussion

It has been widely observed in animal experiments and inclinical studies that roughness improves the osseointegration

Fig. 4. SEM images of structured titanium and zirconia surfaces. SEM images in the top row, false-colour images indicating the surface topography in bottom row. From left TA, TSA,TLA, ZLA. Red circles indicate one of the surface features that was analyzed. Scale bar 100 mm.



0 10 20 30 40 50 60 >650

10

20

30

40

0 10 20 30 40 50 60 >650

10

20

30

40

0 10 20 30 40 50 60 >650

10

20

30

40

0 10 20 30 40 50 60 >650

10

20

30

40

0 10 20 30 40 50 60 >650

10

20

30

40

0 10 20 30 40 50 60 >650

10

20

30

40

bone

colla

gena

seTA

TSA

TLA

ZLA

length [µm]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

Fig. 5. Histograms showing the distribution of the length (maximal dimension) of untreated and collagenase-treated resorption pits on native bone and surface features on titanium(TA, TSA, TLA) and zirconia (ZLA) surfaces. Data were divided into 14 classes of 5 mm in size.



0 5 10 15 20 25 30 >32.50

10

20

30

0 5 10 15 20 25 30 >32.50

10

20

30

0 5 10 15 20 25 30 >32.50

10

20

30

0 5 10 15 20 25 30 >32.50

10

20

30

0 5 10 15 20 25 30 >32.50

10

20

30

0 5 10 15 20 25 30 >32.50

10

20

30

bone

colla

gena

seTA

TSA

TLA

ZLA

width [µm]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

Fig. 6. Histograms showing the distribution of the width of untreated and collagenase-treated resorption pits on native bone and surface features on titanium (TA, TSA, TLA) andzirconia (ZLA) surfaces. Data were divided into 14 classes of 2.5 mm in size.



2 4 6 8 10 12 >130

10203040

0 2 4 6 8 10 12 >130

10203040

0 2 4 6 8 10 12 >130

10203040

0 2 4 6 8 10 12 >130

10203040

0 2 4 6 8 10 12 >130

10203040

0 2 4 6 8 10 12 >130

10203040

bone

colla

gena

seTA

TSA

TLA

ZLA

depth [µm]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

frequ

ency

[%]

Fig. 7. Histograms showing the distribution of the depth of untreated and collagenase-treated resorption pits on bone and surface features on titanium (TA, TSA, TLA) and zirconia(ZLA) surfaces. Data were divided into 14 classes of 1 mm in size.



properties of implant surfaces [11]. A surface roughness Sa of1e2 mm (obtained by optical profilometry) has been stated as beingoptimal [21]. In this study we have tried to correlate these findingswith topographical features produced by osteoclasts and tocompare the size and surface roughness of osteoclastic resorptionpits with synthetic implant surfaces made out of titanium andzirconia.

The resorption pits generated by differentiated RAW 264.7 cellson cortical bone showed similar features to those described in otherSEM studies that employed primary cells [22,23]. In particular, theroundish shape with the clear border between resorbed and non-resorbed areas and the loose collagen fibres within the pits wereclearly observed (Fig. 1). We observed different types of morphol-ogies of resorption pits such as single pits, trails and extensive areasthat were resorbed, although the extensively resorbed areas werenot further analyzed as no distinction of single resorption structureswas possible. The difference between resorption pits and trails wasdiscussed in the context of continuous or non-continuous resorp-tion processes [23], although it is not possible to assign a specificresorption structure to one single or multiple osteoclasts. Howeverthe obtained values nicely correspond to the dimensions of a singleosteoclast or a small cluster of osteoclasts. Without quantifying thesize, Gentzsch et al. classified the different resorption patternsoccurringon autopsies of femoral heads into longitudinal resorptionpatterns and “reticulate patch resorptions”. These structural

features showed good correlation, respectively with the pits andtrails and the areas of extensive resorption that we observed in ourstudy. In similar samples, a differentiation between lacunarresorption and tunnelling perforation, which perforated an entiretrabecular structure, was made [24]. In our study only lacunarresorption was observed. Based on the work mentioned above weassume that our observations of resorption pits might be compa-rable with the in vivo situation while lacking the tunnelling perfo-rations. This leads to the hypothesis that the depth of resorptionpitsin vivo might include higher values compared to the measured invitro values presented here.

The length, width and depth of the surface features on thetitanium and zirconia surfaces increased with increasing size of thegrit used for sandblasting, especially in the case of the two surfacesthat were sandblasted with large grit sizes (TLA and ZLA), whichalso showed the broadest size distribution. The non-sandblastedtitanium surface TA showed the smallest distribution and absolutesize of surface features (Figs. 5e7).

Resorption trails showed a median contour length of 54.9 mmwith a large standard deviation. The width of the trails was largerthan that of the resorption pits, whereas the depth showed a verysimilar distribution. This may indicate that trails originate frommature, and hence large, osteoclasts.

Table 2Statistical evaluation (n, median, first and third quartile Q1 and Q3 and mean aspect ratio) of all examined surfaces.

n Length [mm] Width [mm] Depth [mm] Aspect ratio

Median (Q2) Q1 Median (Q2) Q1 Median (Q2) Q1 Length/width

Q3 Q3 Q3

Bone 134 13.1 8.9 8.1 6.1 1.4 1.0 1.6719.8 11.3 2.0

Collagenase 103 21.7 15.6 15.5 11.5 3.7 2.9 1.4327.3 19.3 4.5

TA 99 11.4 8.7 7.1 6.0 2.5 2.1 1.8115.1 8.8 3.1

TSA 100 23.9 18.8 10.3 9.3 5.4 4.5 2.3528.1 13.0 6.5

TLA 105 32.8 23.5 16.1 12.0 7.7 6.1 2.2941.8 19.1 9.7

ZLA 105 29.0 23.1 20.7 16.1 4.8 3.8 1.5638.1 26.2 5.7

TA TSA TLA ZLA0

500

1000

1500

2000

2500

3000

S a [n

m]

a

a

b

c

Fig. 8. Surface roughness Sa calculated from stereo-SEM images. SEM images wereproduced with a magnification of 1000� and evaluated with a cutoff wavelength oflc ¼ 580 mm. Statistical significances (p < 0.01) between the groups are indicated by a,b, c.

bonecollagenase TA TSA TLA ZLA0

20

40

60

80

100

120

140

160

S a [n

m] a a

b bb

c

Fig. 9. Surface roughness Sa within individual bone resorption pits and within thestructural features of titanium (TA, TSA, TLA) and zirconia (ZLA) surfaces calculatedfrom stereo-SEM images. SEM images were obtained with a magnification of 5000�and evaluated with a cutoff wavelength of lc ¼ 2 mm. Statistical significances(p < 0.001) between the groups are indicated by a, b, c.



The loose ends of the collagen fibres in the resorption pits wereremoved with collagenase treatment (Fig. 1, top right). This led toan increase in the length and width of the resorption pits, whichcould be explained by collagen removal from the edges of the pits.The depth of the resorption pits more than doubled during colla-genase treatment, and the increase of the median depth by 2.3 mmcould be related to the volume of the loose collagen fibres at thebottom of the resorption pits. As collagenase treatment couldpotentially be correlated with the function of bone lining cells,which remove remaining collagen fibres from the resorption pits[4], collagenase-treated resorption pits might be more relevantwhen analysing the dimensions of resorption pits.

The untreated resorption pits showed a median length andwidth that was roughly half the size of sandblasted surfacesalthough in the same range as TA surfaces. Assuming that the in vivosituation might rather be represented by collagenase-treatedresorption pits than resorption pits that still contain residuals ofcollagen fibres, values for length and width correlate well withvalues determined for titanium and zirconia surfaces. This simi-larity was even more pronounced when considering the largerstructures of the resorption trails and tunnelling resorption pitsthat have been described by other authors [24].

The surface roughness within the structural features of the threetitanium surfaces did not differ significantly, since the roughness inthe range below lc ¼ 2 mm was mainly governed by the etchingprocess. The roughness in the bone pits before and after thecollagenase treatment did not differ either, although the structurewithin the pits appeared to be different in the SEM images (Fig. 1).The topography of the loose collagen fibres and the structure of themineral phase after removal of the organic phase (collagenasetreatment) remained similar; however the mechanical propertiesof the soft collagen fibres and the hard hydroxylapatite mightinduce different cellular responses. Overall the observed differ-ences might be important for osteoblastic cell adhesion anddifferentiation considering that single cells react differently tostructures with different configurations in the nanometer scale [25]and in the micrometer scale [26].

In vitro studies using osteoblast cell lines or primary bone cellsshowed increased differentiation on rough surfaces compared tosmooth or machined surfaces [27e29]. Previous work by Boyanet al. showed that osteoblasts cultured on bone surfaces displayingosteoclastic resorption pits showed similar differentiation behav-iour compared to rough titanium surfaces and behaved differentlycompared to smooth surfaces [30].

Surfaces comparable to TLA and ZLA surfaces were used for invivo studies to evaluate their osseointegration performance. Itcould be shown that TLA (sandblasted and etched titaniumsurfaces) showed superior mechanical anchorage in the pelvis ofsheep and in pig mandibles compared to ZLA (sandblasted andalkaline etched zirconia surfaces) after different healing times[19,31]. However, histologically the bone-to-implant contact (BIC)did in general not show any significant differences between thesesurfaces. At the 2-week time point in the sheep’s pelvis zirconiashowed a trend towards higher BIC compared to titanium surfaces[19,32]. Based on the in-depth comparison of these implantsurfaces with osteoclastic resorption pits it can be assumed that thehigh bone-to-implant contact of titanium and zirconia surfacesmay be explained by the fact that the structures of the implantsurfaces remain in the same order of magnitude as compared tosurface structures produced by osteoclastic resorption. The slightlybetter histological results found in vivo for zirconia surfaces may beexplained with roughness within the surface features in thesubmicron range (lc ¼ 2 mm). These values for zirconia are closer tothe roughness values found in resorption pits than for titaniumsurfaces (Fig. 9). The difference in mechanical anchorage in the

bone may be explained by the increased overall surface roughnessof the titanium surfaces, which improves mechanical interlocking(Fig. 8). However comparing a very rough titanium plasma sprayedsurface (TPS) to a medium rough sandblasted and hot acid etchedtitanium surface (SLA) it could be shown histologically that surfaceroughness can also be too high, leading to less bone-to-implantcontact with the TPS surface in vivo for unloaded and 12-monthloaded implants [33].

5. Conclusions

Our study presents surface roughness and surface features ofosteoclastic resorption pits on native bone and compares themwithstate-of-the-art synthetic implant surfaces. The comparison of thedimensions of surface features and the wavelength-dependentroughness of the different surfaces revealed similarities betweennative resorption pits and surface features of the examined tita-nium and zirconia surfaces. In particular the size of the collagenase-treated resorption pits resembled the sandblasted and etchedtitanium and zirconia surfaces. Collagenase treatment of theresorption pits is presented to better reflect the in vivo situation.The results presented here help to interpret in vitro and in vivoresults on osseointegration and may provide some insight into thebone-remodeling cycle on a structural basis. Further work wouldemploy osteoblast cultures on these surfaces to directly correlatethe presented results to in vitro data.

Acknowledgements

The authors would like to thank Dr. Ute Hempel (TechnicalUniversity Dresden) for providing the RAW 264.7 cells, Peter Zim-mermann (University of Basle) for cutting the bone slices, Dr. AnneGreet Bittermann (ZMB, University of Zurich) for SEM support, theCentre for ElectronMicroscopy at ETH Zurich for providing theMeXsoftware and Thommen Medical for a scientific fellowship for TH.

Appendix

Figures with essential colour discrimination. Figs. 1, 2 and 4 inthis article are difficult to interpret in black and white. The fullcolour images can be found in the on-line version, at doi:10.1016/j.biomaterials.2010.06.009.

References

[1] HadjidakisDJ,Androulakis II .Boneremodeling.AnnNYAcadSci2006;1092:385e96.[2] Robling AG, Castillo AB, Turner CH. Biomechanical and molecular regulation of

bone remodeling. Annu Rev Biomed Eng 2006;8(1):455e98.[3] Boyle WJ, Simonet WS, Lacey DL. Osteoclast differentiation and activation.

Nature 2003;423(6937):337e42.[4] Everts V, Delaisse JM, Korper W, Jansen DC, Tigchelaar-Gutter W, Saftig P, et al.

The bone lining cell: its role in cleaning howships lacunae and initiating boneformation. J Bone Miner Res 2002;17(1):77e90.

[5] Abe E, Yamamoto M, Taguchi Y, Lecka-Czernik B, O’Brien CA, Economides AN,et al. Essential requirement of BMPs-2/4 for broth osteoblast and osteoclastformation in murine bone marrow cultures from adult mice: antagonism bynoggin. J Bone Miner Res 2000;15(4):663e73.

[6] Boskey AL. Matrix proteins and mineralization: an overview. Connect TissueRes 1996;35(1):357e63.

[7] Arnett TR, Dempster DW. Effect of pH on bone resorption by rat osteoclasts invitro. Endocrinology 1986;119(1):119e24.

[8] Valverde P, Tu Q, Chen J. BSP and RANKL induce osteoclastogenesis and boneresorption synergistically. J Bone Miner Res 2005;20(9):1669e79.

[9] Teitelbaum SL. Bone resorption by osteoclasts. Science 2000;289(5484):1504.[10] Wennerberg A, Albrektsson T. Effects of titanium surface topography on bone

integration: a systematic review. Clin Oral Implants Res 2009;20:172e84.[11] Junker R, Dimakis A, Thoneick M, Jansen JA. Effects of implant surface coatings

and composition on bone integration: a systematic review. Clin Oral ImplantsRes 2009;20:185e206.



[12] Richards RG, Wieland M, Textor M. Advantages of stereo imaging of metallicsurfaces with low voltage backscattered electrons in a field emission scanningelectron microscope. J Microsc 2000;199(2):115e23.

[13] Jones SJ, Boyde A. Some morphological observations on osteoclasts. Cell TissueRes 1977;185(3):387e97.

[14] Fuller K, Thong JT, Breton BC, Chambers TJ. Automated three-dimensionalcharacterization of osteoclastic resorption lacunae by stereoscopic scanningelectron microscopy. J Bone Miner Res 1994;9(1):17e23.

[15] Wieland M, Hänggi P, Hotz W, Textor M, Keller BA, Spencer ND. Wavelength-dependent measurement and evaluation of surface topographies: applicationof a new concept of window roughness and surface transfer function. Wear2000;237(2):231e52.

[16] Brunette DM. Principles of cell behavior on titanium surfaces and theirapplication to implanted devices. In: Brunette DM, editor. Titanium inmedicine. Berlin: Springer; 2001.

[17] Flanagan AM, Massey HM. Generating human osoteoclasts in vitro from bonemarrow and peripheral blood. In: Helfrich MH, Ralston SH, editors. Boneresearch protocols. Humana Press; 2003.

[18] Breuil V, Cosman F, Stein L, Horbert W, Nieves J, Shen V, et al. Human oste-oclast formation and activity in vitro: effects of alendronate. J Bone Miner Res1998;13:1721e9.

[19] Schliephake H, Hefti T, Schlottig F, Gédet P, Staedt H. Mechanical anchorageand peri-implant bone formation of surface-modified zirconia in minipigs.J Clin Periodontol, in press. doi:10.1111/j.1600-051x.2010.01549.

[20] Wieland M. Experimental determination and quantitative evaluation of thesurface composition and topography of medical implant surfaces and theirinfluence on osteoblastic cell-surface interactions. PhD Thesis. ETH Zürich;1999.

[21] Albrektsson T, Wennerberg A. Oral implant surfaces: part 1-review focusingon topographic and chemical properties of different surfaces and in vivoresponses to them. Int J Prosthodont 2004;17(5):536e43.

[22] Sasaki T, Debari K, Hasemi M. Measurement of howships resorption lacunaeby a scanning probe microscope system. J Electron Microsc 1993;42(5):356e9.

[23] Chambers TJ, Revell PA, Fuller K, Athanasou NA. Resorption of bone by isolatedrabbit osteoclasts. J Cell Sci 1984;66(1):383.

[24] Gentzsch C, Delling G, Kaiser E. Microstructural classification of resorptionlacunae and perforations in human proximal femora. Calcif Tissue Int 2003;72(6):698e709.

[25] Arnold M, Cavalcanti-Adam EA, Glass R, Blummel J, EckW, Kantlehner M, et al.Activation of integrin function by nanopatterned adhesive interfaces. Chem-PhysChem 2004;5(3):383e8.

[26] Lovmand J, Justesen E, Foss M, Lauridsen RH, Lovmand M, Modin C, et al. Theuse of combinatorial topographical libraries for the screening of enhancedosteogenic expression and mineralization. Biomaterials 2009;30(11):2015e22.

[27] Kieswetter K, Schwartz Z, Dean DD, Boyan BD. The role of implant surfacecharacteristics in the healing of bone. Crit Rev Oral Biol Med 1996;7(4):329.

[28] Kunzler TP, Drobek T, Schuler M, Spencer ND. Systematic study of osteoblastand fibroblast response to roughness by means of surface-morphologygradients. Biomaterials 2007;28(13):2175e82.

[29] Hempel U, Hefti T, Kalbacova M, Wolf-Brandstetter C, Dieter P, Schlottig F.Response of osteoblast-like SAOS-2 cells to zirconia ceramics with differentsurface topographies. Clin Oral Implants Res 2010;21(2):174e81.

[30] Boyan BD, Schwartz Z, Lohmann CH, Sylvia VL, Cochran DL, Dean DD, et al.Pretreatment of bone with osteoclasts affects phenotypic expression ofosteoblast-like cells. J Orthop Res 2003;21(4):638e47.

[31] Ferguson SJ, Langhoff JD, Voelter K, von Rechenberg B, Scharnweber D,Bierbaum S, et al. Biomechanical comparison of different surface modifica-tions for dental implants. Int J Oral Maxillofac Implants 2008;23(6):1037.

[32] Langhoff JD, Voelter K, Scharnweber D, Schnabelrauch M, Schlottig F, Hefti T,et al. Comparison of chemically and pharmaceutically modified titanium andzirconia implant surfaces in dentistry: a study in sheep. Int J Oral MaxillofacSurg 2008;37(12):1125e32.

[33] Cochran DL, Schenk RK, Lussi A, Higginbottom FL, Buser D. Bone response tounloaded and loaded titanium implants with a sandblasted and acid-etchedsurface: a histometric study in the canine mandible. J Biomed Mater Res1998;40(1):1e11.


Documents

Author's personal copy - geosoft.ru 2010_Biomaterials_.pdf · Various differentin vitromodel systems for the study of osteo-clasts have been described. Besides the cultivation of