*Corresponding author. Tel.: #46-31-773-3384; fax: #46-31-773-3330.E-mail address: sanjiv.kanagaraja@odontologi.gu.se (S. Kanagaraja).

Biomaterials 22 (2001) 1809}1818

Cellular reactions and bone apposition to titanium surfaces withdi!erent surface roughness and oxide thickness cleaned by oxidation

Sanjiv Kanagaraja��*, Ann Wennerberg�, Cecilia Eriksson�, Ha� kan Nygren��Department of Anatomy and Cell Biology, University of Go( teborg, P.O.B. 420, SE 405 30 Go( teborg, Sweden�Biomaterials group, Department of Handicap Research, University of Go( teborg, SE 413 12 Go( teborg, Sweden

Received 23 March 2000; accepted 17 October 2000

Abstract

Titanium surfaces with three di!erent surface characteristics were exposed to an intraperitoneal milieu in mouse or rat, or insertedinto rabbit bone. The cleaning regimen of the TiO

�surfaces in this study included oxidation by heat or acid and a "nal rinsing and

storage in water. Intraperitoneal exposure ranged from 1 to 64min and the healing period in bone was 6 weeks. Cell recruitment to thesurfaces was quanti"ed by acridine orange staining and speci"c antibodies directed against cell membrane antigens. Removal torque,bone-to-metal contact , total bone area and histological evaluations were used to evaluate "xture stability and the healing-in of theimplants. After the healing period of 6 weeks only a transient signi"cant di!erence was seen in the total number of cells adherent onthe surfaces. No signi"cant di!erences were observed between any of the surfaces for removal torque, bone-to-metal contact, or bonearea. The areas lacking bone-to-metal contact were "lled with normal vascularised connective tissue with no signs of "brous capsuleformation or giant cells. These "ndings di!er from "ndings published earlier of Ti implants that underwent a cleaning regimen withalcohol as the "nal rinsing step. The tissues around the implants were richly vascularised and there was continued bone growthtoward the surfaces. The bone-to-metal contact in this study was lower than that seen with alcohol-cleaned TiO

�. � 2001 Elsevier

Science Ltd. All rights reserved.

Keywords: Blood; Bone; Cell}surface reactions; In#ammation; Ti implants; TiO�

1. Introduction

Osseointegration is de"ned as direct contact, implyingstructural and functional connection between living boneand the surface of a load bearing alloplastic implant [1].The success rates for titanium implant supported pros-thesis have been reported to be as high as 100% [2,3].However, while the end point of the osseointegrationprocess is understood, the exact mechanisms leading tothis state is unclear.There is substantial documentation of how surface

properties of biomaterials in#uence the healing response,specially with regards to titanium surfaces. Bio-material}tissue interactions occur within a narrow zoneof less than 1 nm [4]. Authors such as Wennerberg andLarsson [5,6] have investigated quite extensively thehealing responses of titanium to soft and hard tissue after

implantation periods of 6 weeks and above. Otherauthors [7,8] have investigated the response of tissue#uids and blood components to Ti surfaces.Most bone anchored "xtures traverse both the cortical

and medullary bone. Medullary bone is richly vas-cularised, and therefore, an immediate contact betweenimplant and blood is almost inevitable. It is thereforereasonable to assume that the healing or chronic in#am-matory process should begin with reactions taking placebetween the foreign surface and blood. To our knowledgehowever, there are few studies so far done on the initialresponses between blood}surface contact, and correla-tions made to the bone healing response aroundimplants. In previous studies we have investigated theresponse of blood components to di!erently modi"edsurfaces for times ranging from 15 s and above [9]. Theaim of the current study was to investigate the tissue(blood and bone) responses to di!erently modi"ed surfa-ces, upto 6 weeks after implantation and to detect, ifpossible, factors that would give an indication ofa quicker and better healing response. In the cell

0142-9612/01/$ - see front matter � 2001 Elsevier Science Ltd. All rights reserved.PII: S 0 1 4 2 - 9 6 1 2 ( 0 0 ) 0 0 3 6 2 - 8

adhesion tests, emphasis was placed on macrophage de-tection on the surfaces, due to the pivotal role played bythese cells in wound healing [10].

2. Material and methods

2.1. Sample preparation

Commercially pure (c.p.), grade 1 (99.6%) titaniumsheets were cut in squares of 4�4 or 8�8mm and 96 c.p.titanium implants with a length of 7mm, an outer dia-meter of 3.7mm and a pitch-height of 0.6mm were used.The samples were washed in NH

�OH (25%) :H

�O :

H�O

�(30%), 1 : 5 : 1 at 803C for 10min and then rinsed

three times in boiled cooled distilled water and stored inthe distilled water. The samples were modi"ed in threedi!erent ways. The surface called polished was justcleaned and considered to have a smooth surface witha thin oxide layer. One surface was etched in 10%HF for3min and then oxidised in concentrated HNO

�for

20min and was called etched. This surface was rougherwith a thin oxide layer. The last surface, calledetched/annealed, was etched for 3min in 10% HF fol-lowed by annealing at 700}8003C to a purple interferencecolour, resulting in a rougher surface with a thick oxidelayer.The screws were divided into the three groups above,

i.e. polished, etched and etched/annealed.

Group A: PolishedGroup B: EtchedGroup C: Etched/annealed

After surface modi"cation all implants were immersed indistilled water and cleaned in ultrasonics (frequency50kHz) for 2min, rinsed in distilled water, and thenautoclaved at 1203C for 45min, and kept in the distilledwater they were autoclaved in until implantation.

2.2. Surface topography characterisation

Qualitative and quantitative characterisation of thesurface topography was made with a confocal laserscanning pro"lometer (TopScan 3D� Heidelberg Instru-ments, Heidelberg, Germany). This instrument has beendescribed and controlled in previously published studies[11,12]. Three screws from each surface modi"cation(polished, etched, etched/annealed), were selected at ran-dom and measured on three thread-tops, three thread-valleys and three #anks on each screw, that makes 27measurements for each modi"cation. The measuring areawas 245�245�m for all measurements. As required bythe de"nition of surface roughness parameters [13], a "l-ter was used to separate roughness from waviness andform. The "lter type was Gaussian and the "lter size wasset to 50�50�m�. To numerically describe the surface

roughness, three di!erent parameters recommended for3D characterisation of surface roughness were used. Theyare all mathematically described in the previous referredwork by Stout et al. [14].

S�is the arithmetic mean height deviation from a mean

plane, measured in �m.S��is the averaged mean spacing of pro"le heights in

spatial direction, measured in �m. S��is the ratio of the

measured surface and a corresponding perfect plane area.

2.3. Intraperitoneal experiments

All the animal work was approved and conductedaccording to the guidelines stipulated by the Gothenburganimal experiment ethical committee. All cell adhesiontests were done by intraperitoneal (IP) exposure of thesamples. IP experiments in mice were performed whereno suitable antibodies to rat antigens were available. Theexperimental procedure for mice was performed exactlyas described earlier [15], the duration of intraperitonealexposure being 1}64min.The rat intraperitoneal tests were performed using

Sprague Dawley male rats, approximately 250 g body wt(B&KUniversal AB, Sweden) which were provided withtap water and fed on R-34 rat-mouse pellets (Lactamine,Sweden). The animals were housed in meshed stainless-steel cages at a room temp of 21}223C with a 12 hlight-dark cycle. The animals were allowed at least 7 daysto acclimatise to our laboratory conditions. They wereanaesthetised by intramuscular administration of a mix-ture of approximately 0.5ml Pentobarbitalnatrium60mg/ml, NaCl 9mg/ml and Diazepam 5mg/ml mixedat a ratio 1 : 1 : 2, respectively. A midabdominal incisionof approximately 2.5}3 cm was made and three8�8mm� samples/rat were implanted intraperitoneally.The samples used had blunted edges in order to preventbleeding after implantation. The incision was closed with9mm stainless-steel surgical wound clips. Intraperitonealexposure times varied from 1 to 64min. The rats did notcome out of anaesthesia. The samples were either foundin the folds of the intestines or between the abdominalmusculature and viscera without detectable adhesions.They were carefully removed from the peritoneal cavity,gently washed in PBS in order to remove non-adherentcells, and were either used immediately for staining orstored in absolute alcohol at !203 overnight. On re-trieval, samples were discarded where bleeding could beobserved with the naked eye.

2.4. Cell staining and cell typing

Acridine orange staining was performed in order todetect nucleated cells on the surfaces as described earlier[16], and detection of an intact intracellular pool ofesterases with Propidium iodide/#uorescein diacetate(viability staining) was also performed as described

1810 S. Kanagaraja et al. / Biomaterials 22 (2001) 1809}1818

earlier [15,17]. Surface incubations were performed withspeci"c mouse anti rat CD11b/CD18, or rat anti mouseCD25 and CD74 (Serotec, Oxford, England).

2.5. Experiments in bone

Sixteen adult, female New Zealand White rabbits wereused in the present study. They were aged between 10 and12 months, and were allowed to run free in a largepurpose-designed room and had free access to pellets andtap water.At surgery, the animals hind legs were shaved and

washed with an iodine solution. Preoperative antibioticswere administered prophylactically (PenoVet, Boeringer,Denmark). The animals were anaesthetised with intra-muscular injections of Hypnorm Vet (Janssen Far-maceutica, Belgium) at a dose of 1ml and intraperitonealinjections of Stesolid Novum (a/s Dumex, Denmark) ata dose of 0.3ml per animal. Local analgesia (Xylocain,Astra, Sweden) was injected to the implant bed at a doseof 1ml per site.After 6 weeks the rabbits were euthanised with an

intravenous injection of Pentobarbitalum� (Apoteks-bolaget, Uppsala, Sweden).

2.6. Surgical technique and implant insertion

Six implants each were inserted in all of the 16 animals,one in each distal femoral metaphysis and two in eachproximal tibial metaphysis. In the femoral sites of eightrabbits one Gp A and one Gp C screw was inserted. Inthe remaining eight animals one Gp A and one Gp Bscrew was placed. In the tibia 2 Gp A screws wereinserted in one side and in the contra lateral side 1 Gp Band 1 Gp C screw was inserted. The screws were alter-nately inserted in the left or right leg. The tibial implantswere separated by at least 5mm. The implantation holeswere drilled with a low rotatory speed, profuse salineirrigation and with successively increasing diameters, and"nally tapped with a 3.7mm tap. The implants pen-etrated the "rst cortical layer only, with two threadsvisible above the cortex.

2.7. Torque measurement, preparation of specimensand histomorphometric evaluation

The tibial implants of eight animals were used forhistomorphometric evaluation i.e. 16 Gp A, 8 Gp B and8 Gp C screws. The femoral implants of these rabbits andthe femoral and tibial implants of the remaining eightanimals were evaluated with respect to the torque neededto loosen the screw from the bone tissue. Comparison fortorque as well as histomorphometrical evaluation wasmade only between Gp A and Gp B or Gp A and Gp C atthe same position (femur or in the distal or proximaltibia).

2.8. Torque measurement

The torque required to loosen the screws was mea-sured with an electrical torque transducer as described byJohansson et al. [18].

2.9. Histomorphometry

The screws with their surrounding tissue were "xed in4% neutral bu!ered formaldehyde and processed to beembedded in light-curing resin, (Technovit 7200 VLC�,Kultzer & Co, Germany). Cutting and grinding wasperformed ad modum Donath [19]. The specimens wereground to an approximate thickness of 10 �m andstained with toluidine blue. A PC connected to LeitzMicrovid equipment and a Leitz Aristoplan microscopewas used for calculating the percentage of bone-to-metalcontact and percentage of bone inside the screw threadsdirectly in the eyepiece of the microscope. All light micro-scopical calculations were made with a 10� objectiveand 10� eye pieces. Light microscopy of the histology oftissue adjacent to the implant surface was done witha Zeiss microscope connected to a digital camera.

2.10. Statistical analysis

Wilcoxon signed rank test was used to calculatep values for all the data in this study, for determination ofsigni"cant di!erences at the 5% level.

3. Results

3.1. Topographical evaluation

A similar surface topography was demonstrated(Fig. 1a and b) for the three di!erent surface modi"ca-tions, even though the two surfaces prepared with anexperimentally produced oxide layer was slightlyrougher. This was demonstrated with all of the threeparameters used. The polished surface showed an aver-age surface roughness (S

�) of 0.93�m, corresponding

value for the etched and etched/annealed surfaces were1.04 and 1.07�m, respectively. These "gures and thevalues for the space descriptive parameter S

��and the

hybrid parameter S��

are summarised in Table 1. Theoxide thickness for the polished and etched surfaces was4}5nm and for the etched/annealed surfaces 30 nm [20].

3.2. Cell adhesion

Detection of nucleated cells on the surfaces by acridineorange after 1, 4, 16 and 64min showed no signi"cantdi!erences (p'0.05) between the surfaces except at 4min(Fig. 2a).

S. Kanagaraja et al. / Biomaterials 22 (2001) 1809}1818 1811

Fig. 1. Computer-generated image of the surface topography for the (a) polished and (b) etched/annealed surfaces. The etched surface had a similarappearance to (b). Each of the red and white bars represent 10 �m.

Table 1Three screws of each surface modi"cation were measured, and eachscrew was measured on nine areas. The mean value and standarddeviation of these 27 measurements on every surface modi"cation aresummarized below. Standard deviation within parentheses

S�

S��

S��

Polished surfaceScrew 1 0.95 8.53 1.33

(0.36) (1.45) (0.21)Screw 2 0.89 8.05 1.32

(0.20) (0.39) (0.12)Screw 3 0.96 8.24 1.35

(0.38) (0.76) (0.17)Mean value 0.93 8.24 1.33

(0.31) (0.76) (0.16)Etched surfaceScrew 1 1.01 9.52 1.35

(0.33) (0.90) (0.12)Screw 2 1.06 9.49 1.39

(0.30) (0.94) (0.11)Screw 3 1.03 9.39 1.39

(0.33) (0.64) (0.13)Mean value 1.04 9.47 1.38

(0.31) (0.80) (0.12)Etched/annealed surfaceScrew 1 0.99 9.38 1.34

(0.33) (0.74) (0.10)Screw 2 1.11 9.48 1.43

(0.34) (0.70) (0.13)Screw 3 1.10 9.43 1.39

(0.33) (0.76) (0.09)Mean value 1.07 9.43 1.39

(0.33) (0.76) (0.11)

3.3. Cell membrane permeability

Dead cells or cells undergoing apoptosis arepropidium iodide (PI) positive. These cells have tradi-tionally been considered as non-viable cells, but morerecent studies have shown that cell stress could transiently

cause cell membrane leakage to PI [21]. After 60min ofIP exposure approximately 80% of surface adherent cellson all three surfaces were permeable to PI (Fig. 2b) i.e. nosigni"cant di!erences could be observed between any ofthe surfaces (p'0.05).

3.4. Cell typing and morphology

Antibodies directed against cell membrane antigens ofrat neutrophils, CD11b/CD18, showed signi"cant di!er-ences (p(0.05) at 1, 16 and 64min between the etchedand polished surfaces only (Fig. 2c). The cells were foundsingly or in clusters evenly distributed over the surfaces.Adhesion and activation of macrophages were

detected by their expression of CD74 and CD25, respec-tively, after 60min IP (in mice). Quanti"cation of cells forCD25 was avoided due to the sparse and uneven distri-bution of the cell types on the surfaces. On all threesurfaces solitary CD25 positive cells were seen with noapparent variation between the surfaces. The surface-area/#uorescent cell was calculated for CD74 (Fig. 3).The results are summarised in Table 2. The etched surfa-ces had the highest surface area/cell while the polishedsurfaces had the lowest. Signi"cant di!erences were ob-served between both polished and etched and polishedand etched/annealed surfaces (p(0.05).

3.5. Torque measurement

Femur: In eight animals etched/annealed versuspolished (control) screws were compared. No statisticalsigni"cant di!erence was found. The controls (polishedscrews) required a torque of 21Ncm to be loosen fromthe bone bed while the value for the etched/annealedscrews was 27Ncm (p"0.18). In the remaining eightanimals a comparison was made between controls andetched screws. The values were 21Ncm for the controlsand 21Ncm for etched screws (p"0.83) (Fig. 4a).

1812 S. Kanagaraja et al. / Biomaterials 22 (2001) 1809}1818

Fig. 2. (a) Percentage of surface coverage of nucleated cells on the di!erent surfaces. A signi"cant di!erence is observed only at 4min (p(0.05); n"18.(b) Percentage of propidium iodide/#uorescein diacetate positive cells after 1 h IP exposure. No signi"cant di!erences were observed between any ofthe surfaces (p'0.05); n"18. (c) Normalised coverage, $standard deviation, CD11b/CD18 per adherent cell. Signi"cant di!erences were seen onlybetween the etched and polished surfaces at 1, 16 and 64min (p(0.05); n"18.

Tibia: In eight animals the tibial implants were evalu-ated with torque measurement, control versus etched/annealed and control versus etched. The removal torquewas 14Ncm for control screws and 15Ncm for theEtched screws inserted in proximal tibia (p"0.49). Cor-responding values for control versus Etched/annealedsituated in the distal tibia were 20 and 26Ncm, respec-tively (p"0.08) (Fig. 4b).

3.6. Histomorphometrical evaluation

No statistical di!erence was found when investigatingthe three surface modi"cations in terms of bone-to-metalcontact or the amount of bone inside the thread area. Theevaluation included calculation from all threads as wellas the best three consecutive threads representing thesituation in the cortical passage.Tibia: A comparison between the control (polished)

versus the etched/annealed screws demonstrated a bone-to-metal contact for the control screws of 3 and 5%

for the etched/annealed screws when calculating all avail-able threads. When calculating the best three consecutivethreads the corresponding values were 7 and 10%,respectively.Nor when the bone area inside the threads was evalu-

ated was there any signi"cant di!erence. There was 51%bone in all the control threads, 78% of bone in the threebest consecutive threads, corresponding "gures for theEtched/annealed side was 57 and 79% (Table 3, Fig. 5aand b).When the controls were compared with etched screws

a bone-to-metal contact of 5% was registered for thecontrol screws and 3% for the etched screws when thecalculation was performed over all threads, and 5 and6% when the three best threads were analysed.Bone area inside the threads was 49% for control and

41% for etched screws when all threads were calculated,and corresponding values for the 3 best consecutivethreads were 78 and 74%, respectively, (Table 4, Fig. 5aand c).

S. Kanagaraja et al. / Biomaterials 22 (2001) 1809}1818 1813

Fig. 3. Immuno#uorescence micrograph of FITC-labelled monoclonal antibodies directed against the cell surface antigens, (a) antibody, CD74,directed against adherent macrophages exhibiting a round morphology and (b) spread morphology.

Table 2Surface area per CD74 positive cell on the three modi"ed titaniumsurfaces, after 1 h intraperitoneally in mice. SEM"standard error ofmean; n"36

Area/cell (�m�)

Mean SEM

Polished 130 8Etched 175 7Etched/annealed 163 9

Fig. 4. The torque required, $standard deviation, to loosen the implants from (a) Femur and (b) Tibia. No signi"cant di!erences were observedbetween any of the surfaces (p'0.05).

3.7. Histology

The areas adjacent to the implant surface were charac-terised by newly formed woven bone with a rich vascularsupply. Blood capillaries are seen traversing both thecortical and woven bone, and in close contact with the

implant surface (Fig. 6a). The bone in the areas wherethere was no bone-to-metal contact was separated bya space 40$40�m wide "lled with normal loose con-nective tissue with scattered leucocytes seen occasionally.Osteoblasts can be seen arranged in a row at the edge ofthe developing bone indicating normal bone growth to-wards the implant surface. No giant cells nor signs of"brous capsule formation could be observed in thetissues immediately adjacent to the implant surface(Fig. 6b).

4. Discussion

The permanent attachment of arti"cial devices(biomaterials) in bone are currently carried out by allow-ing bony ingrowth to the uneven/porous surface layerof the implant rather than by using bone cement.With regards to Ti, these surfaces have been subject to

1814 S. Kanagaraja et al. / Biomaterials 22 (2001) 1809}1818

Table 3Control (polished surface) versus etched/annealed surface (thick oxide)�

Polished surface Etched/annealed surface

All threads Best threeconsecutivethreads

All threads Best threeconsecutivethreads

BMC 3 7 5 10SD (min}max) 2 (1}7.3) 5 (1.6}18) 3.6 (0.4}10.5) 6.5 (1}19)Area 51 78 57 79SD (min}max) 8 (42}68) 2 (75}82) 9 (46}71) 3 (74}84)

�The percentage of bone-to-metal contact (BMC) and bone areacalculated for all threads available and the best three consecutivethreads representing the situation at the cortical passage after 6 weeksof insertion. SD"standard deviation, min}max values are within par-entheses.p values control versus thick oxide:

All threads Best threep-value BMC 0.31 0.32p-value area 0.16 0.61

Fig. 5. Ground section of implants inserted in rabbit bone (scale bar"200�m). (a) Polished surface (control), (b) etched/annealed surface, (c) etchedsurface.

numerous surface modi"cations in an attempt to speedup and optimise the healing process, i.e. to achieve higherbone-to-metal contact at shorter healing times [5,6,22].The ultimate success or failure of an implant is deter-

mined by how well it functions upon loading. Thus para-meters such as bone-to-metal contact, removal torqueand bone quality adjacent to implant surfaces have beenused to select suitable implant materials and even predictthe long-term functional life of an implant.In the present study, the intraperitoneal cell adhesion

tests clearly show only a transient signi"cant di!erence inthe total number of adherent nucleated cells, and nodi!erences in cell permeability to propidium iodidebetween the surfaces.Most previous studies concerning the in#uence of the

implant surface have pointed out the important roleplayed by a certain degree of roughness [5,23}27] andthat surface roughness also determines the tissue reactionat the interface and directly in#uences cell activities [27].Only the etched surfaces (rougher) showed signi"cantlyhigher values for CD11b/CD18 (an antibody directedagainst primed leukocytes), and also demonstrated thehighest surface area/cell for CD74 (an antibody directedagainst adherent macrophages), indicating more spreadthan round cells.The initial events of cell adhesion to substrata involve

extracellular proteins adsorbed onto the surface, speci"cmembrane receptors and cytoskeletal proteins.WaK livaara et al. [28] found that Ti surfaces with di!erent

S. Kanagaraja et al. / Biomaterials 22 (2001) 1809}1818 1815

Table 4Control (polished surface) versus etched surface (thin oxide)�

Polished surface Etched surface

All threads Best threeconsecutivethreads

All threads Best threeconsecutivethreads

BMC 5 5 3 6SD (min}max) 7.4 (0.1}6) 3.5 (0.3}10.5) 1.8 (1.5}5.5) 3.6 (2}12)Area 49 78 41 74SD (min}max) 8 (42}62) 5 (72}84) 8 (30}56) 4 (69}79)

�The percentage of bone-to-metal contact (BMC) and bone area (area)calculated for all threads available and the best three consecutivethreads representing the situation at the cortical passage after 6 weeksof insertion. SD"standard deviation, min}max values are within par-entheses.p values control versus thin oxide:

All threads Best threep-value BMC 0.72 0.57p-value area 0.06 0.06

Fig. 6. Light micrographs of implants and surrounding tissue 6 weeks after insertion in rabbit bone. (a) Ti surface in the path of blood #ow betweenblood vessels (V). Blood and bone apparently in direct contact with the surface (arrows). Newly formed woven bone (WB) and mature lamellar bone(LB). Space (S) between surface and bone. Original magni"cation �400. (b) Loose connective tissue (CT) separating Ti surface and bone. Fibroblasts(F) and a few leucocytes (L) can be seen. Some cells appear to have a pyknotic appearance (P). Osteoblasts (O) on developing bone margin. Newlyformed woven bone (WB). Original magni"cation �1000.

oxide thickness (structures) bound signi"cant amount ofhuman serum albumin and "bronectin. Fibronectin isknown to be involved in cell adhesion processes andcontains the speci"c RGD sequence that is recognised byspeci"c cell membrane receptors such as integrins. Therougher surfaces in this study had similar surface topo-graphies but di!erent oxide thickness. The integrity ofthe oxide layer (thickness) was measured by Auguer elec-tron microscopy [20]. Before and after the mechanicalinsertion of the implants into cortical bone, and found tobe the same. Therefore, it appears that with Ti surfaces,the oxide thickness together with surface roughness playsan important role in the adhesion and spreading of cellsinitially arriving at the surfaces.Previous in vitro studies indicate that surface rough-

ness in#uences the adhesion, spreading and proliferation

of "broblasts and osteoblast-like cells on titanium sur-faces [29,30]. However, Sennerby et al. reported thatosteoblasts did not adhere to the surface and that boneformation was not initiated at the surface [31,32]. Thisobservation suggests that studies on osteoblast}surfaceinteractions in vitro is of lesser relevance compared toother potential and complex modulating factors presentin the vicinity of cells and surfaces in a complex in vivosystem.In this study, no signi"cant di!erences could be ob-

served between any of the surfaces for the torque measure-ment, bone-to-metal contact or bone area measurements.Tendencies indicate however, a higher removal torqueand bone-to-metal contact for the etched/annealed surfa-ces than for the annealed surfaces. Thus we see an inverserelation between initial cell adhesion and spreading andimplant "xture in bone. It has to be noted that thecomparison between IP and bone implant evaluationsmade here have di!erent animal species.It is also noteworthy that the bone-to-metal contact

and even total bone area values obtained in this study aresubstantially lower than in those of previous studies[23}25]. It may be argued that comparisons betweenresults from di!erent studies may be di$cult since animalspecies, experimental model, observation period, surfaceproperties, implant design andmethod of evaluationmaybe di!erent. Other factors such as oxide thickness, "xturemobility after implantation, sex and age of test animalsalso play an important role in the healing process. How-ever, results obtained after 6 weeks in bone are compara-ble despite variations in experimental protocol. Forinstance, in this study and another by Thomsen et al.[33], the animals were allowed free movement after im-plant surgery, while in another study by Wennerberget al. [25] the animals were caged. Further, the implantsurface preparation procedures in these two studies weredi!erent. The implants were retrieved in the above-men-

1816 S. Kanagaraja et al. / Biomaterials 22 (2001) 1809}1818

tioned studies [25,33] after only 4 weeks and despite thevariations in experimental protocol between these twostudies, similar results were obtained both for bone-to-metal contact and total bone area. Therefore, despite thevariation in implantation time between this study (6weeks) and the others (4 weeks) and the variations orsimilarities in animal movement post operatively, it issafe to conclude that the total bone-to-metal contact wasconsiderably lower in this study.The titanium surfaces used in the present study were

etched or heated to produce novel oxide, and then storedunder water until use in order to avoid the formation ofan organic layer at the surfaces. Most studies reportinghigher levels of bone contact [23}25] have used alcoholas the "nal rinsing step. Alcohol may add to the TiO

�surface and make it hydrophobic [34]. The alcoholtreated surfaces and the etched surfaces (hydrophilic)stored in water have di!erent reactions with blood [34]which may re#ect di!erences in bone integrating capacity.The water-stored TiO

�surfaces did not induce formation

of giant cells seen along the alcohol-rinsed TiO�surfaces

[33], and no "brous capsule formation was seen evenwhen the implants reached the bone marrow. Instead,normal vascularised connective tissue with ongoingbone growth toward the surfaces was seen around theimplants used in the present study. We suggest thereforethat these di!erences are due to the di!erent cleaningregimens employed between this study and the others,which results in a di!erent surface chemistry of thesurfaces.

5. Conclusions

Although the results indicate an initial transient di!er-ence in cell recruitment to the di!erent surfaces, no directcorrelation could be made between cell recruitment andbone formation around the implants. i.e. di!erent surfaceroughness and oxide thickness within the range investi-gated in this study did not result in di!erent cell activity,bone}implant contact or removal torque. However,the cleaned and water-stored TiO

�induced less

in#ammatory reactions, less or lower rate of bone forma-tion and richly vascularised loose connective tissuearound the implant, compared to other studies wherealcohol was used as the "nal rinsing step in the cleaningprocedure.

Acknowledgements

This study has been supported by grants from Sylvans'foundation, the Wilhelm and Martina Lundgren ScienceFoundation, the Swedish Medical Research Council andthe Hjalmar Svensson Research Foundation.

References

[1] Bra� nemark PI, Zarb GA. Introduction to osseointegration. In:Bra� nemark PI, Zarb GA, Albrektsson T, editors. Tissue integ-rated prostheses: osseointegration in clinical dentistry. 1st ed.London: Quintesence Publishing Co. Inc., 1985. p. 11}63.

[2] Albrektsson T, Arnebrant T, Larsson K, et al. E!ect of aglycoprotein monomolecular layer on the integration oftitanium implants in bone. Biol Biomed Perform Biomater1986;15:349}54.

[3] Albrektsson T, Dahl E, Enbom L, et al. Osseointegrated oralimplants a Swedish multicentre study of 8139 consecutively in-serted Nobelpharma implants. J Periodontol 1988;5:287}96.

[4] Kasemo B, Lausmaa J. Surface science aspects on inorganicbiomaterials. CRC Crit Rev Biocompat 1986;2:335}80.

[5] Wennerberg Ann. On surface roughness and implant incorpora-tion. Thesis. Department of Biomaterials/handicap research,GoK teborg University, GoK teborg, Sweden, 1996.

[6] Larsson Cecilia. The interface between bone and metalswith di!erent surface properties. Thesis. Institute of Anatomyand Cell Biology, GoK teborg University, GoK teborg, Sweden,1997.

[7] Rosengren A, Johansson BR, Danielsen N, Thomsen P, EricsonLE. Immunohistochemical studies on the distribution of albumin,"brinogen, "bronectin, IgG and Collagen around PTFE andTitanium implants. Biomaterials 1996;17(18):1779}86.

[8] WaK livaara B, Askendal A, LundstroK m I, Tengvall P. Blood pro-tein interactions with titanium surfaces. J Biomed Sci Polym Ed1996;8(1):41}8.

[9] Kanagaraja S, LundstroK m I, Nygren H, Tengvall P. Plateletbinding and protein adsorption to titanium and gold after shorttime exposure to heparinised plasma and whole blood. Bio-materials 1996;17:2225}32.

[10] DiPietro LA. Wound healing: the role of the macrophage andother immune cells. Shock 1995;4(4):233}40.

[11] Wennerberg A, Albrektsson T, Ulrich H, Krol J. An optical 3Dmethod for topographical description of surgical implants. J Bio-med Engng 1992;14:412}8.

[12] Wennerberg A, Albrektsson T, Andersson B. Bone tissueresponse to c.p. titanium implants blasted with "ne and coarseparticles of aluminium oxide. J Oral Maxillofac Implants1996;11:38}45.

[13] Bs 1134: Part 1 (1988) British standard assessment of surfacetexture. Part 1. Methods and instrumentation. British StandardsInstitution, London, England.

[14] Stout KJ, Davis EJ, Sullivan PJ, Dong WP, Mainsah E, Luo N,Mathia T, Zahouani H. The development of methods for thecharacterisation of roughness in three dimensions. EUR 15178EN of commission of the European Communities, University ofBirmingham, Birmingham, England 1993.

[15] Nygren H, Kanagaraja S, Braide M, Eriksson C, LundstroK m I.Characterisation of cellular response to thiol modi"ed gold surfa-ces implanted in mouse peritoneal cavity. J Biomed Mater Res1999;45:117}24.

[16] Kanagaraja S, Alaeddine S, Eriksson C, Lausmaa J, Tengvall P,Wennerberg A, Nygren H. Surface characterisation, protein ad-sorption, and initial cell surface reactions on glutathione and3-mercapto-1,2,-propanediol immobilized to gold. J Biomed Ma-ter Res 1999;46:582}91.

[17] Jones KH, Senft JA. An improved method to determine cellviability by simultaneous staining with #uorescein diacetate}propidium iodide. J Histochem Cytochem 1985;33:77}9.

[18] Johansson CB, Wennerberg A, Albrektsson T. A quantitativecomparison of machined commercially pure titanium andtitanium}aluminium}vanadium implants in rabbit bone. IntJ Oral maxillofac Implants 1998;13:315}21.

S. Kanagaraja et al. / Biomaterials 22 (2001) 1809}1818 1817

[19] Donath K. (1988) Die Trenn-DuK nnschlif-Technik zur Herstellunghistologischer PraK parate von nicht schneidbaren Geweben undMaterialen. Der PraK parator 1988;34:197}206.

[20] Nygren H, Eriksson C. Experimental approach to a biologicalcharacterisation of materials. J Vac Sci Technol A 1997;15(3):768}72.

[21] Lee S, Andersson T, Zhang H, Flotte TJ, Doukas AG. Alterationof cell membrane by stress waves in vitro. Ultrasound Med Biol1996;22:1285}93.

[22] Hazan R, Brener R, Uron U. Bone growth to metal implants isregulated by their surface chemical properties. Biomaterials1993;14(8):570}4.

[23] Larsson C, Thomsen P, Aronsson BO, Rodahl M, Lausmaa J,Kasemo B, Ericson LE. Bone response to surface modi"ed tita-nium implants: Studies on the early tissue response to machinedand electropolished implants with di!erent oxide thicknesses.Biomaterials 1996;17:605}16.

[24] Larsson C, Thomsen P, Lausmaa J, Rodahl M, Kasemo B,Ericson LE. Bone response to surface modi"ed titanium implants:studies on electropolished implants with di!erent oxide thick-nesses and morphology. Biomaterials 1994;15:1062}74.

[25] Wenneberg A, Albrektsson T, Andersson B. Bone tissue responseto commercially pure titanium implants blasted with "ne andcoarse particles of aluminium oxide. Int J Oral Maxillofac Im-plants 1996;11:38}45.

[26] Bowers KT, Keller JC, Randolph BA, Wick DG, Michaels CM.Optimisation of surface micromorphology for enhanced osteoblastresponses in vitro. Int J Oral Maxillofac Implants 1992;7:302}9.

[27] Ungersbock A, Pohler O, Perren S. Evaluation of soft tissuereactions at the interface of titanium limited contact dynamiccompression plate implants with di!erent surface treatments. Bio-materials 1996;17:797}806.

[28] WaK livaara B, Aronssob BO, Rodhal M, Lausmaa J, Tengvall P.Titanium with di!erent oxides: in vitro studies of protein adsorp-tion and contact activation. Biomaterials 1993;15:827}34.

[29] Eisenbarth E, Meyle J, Nachtigall W, Breme J. In#uence of thesurface structure of titanium materials on the adhesion of "bro-blasts. Biomaterials 1996;17:1399}403.

[30] Degasne I, Basle MF, Demail V, Hure G, Lesourdf M, GrolleauB, Mercier L, Chappard D. E!ect of roughness, "bronectin andvitronectin on attachment, spreading and proliferation of humanosteoblast-like cells (Saos-2) on titanium surfaces. Calci"ed TissueInt 1999;64:499}507.

[31] Sennerby L, Thomsen P, Ericson LE. Early tissue response toTitanium implants inserted in rabbit cortical bone. Part I. Lightmicroscopic observations. J Mater Sci: Mater Med 1993; 4:240}50.

[32] Sennerby L, Thomsen P, Ericson LE. Early tissue response toTitanium implants inserted in rabbit cortical bone. Part II. Ultra-structural observations. J Mater Sci: Mater Med 1993;4:494}502.

[33] Thomsen P, Sennerby L, Larsson C, Lausmaa J, Kasemo B,Ericson LE. Structure of the interface between rabbit corticalbone and implants of gold, zirconium and Titanium. J Mater Sci:Mater Med 1997;8(11):653}65.

[34] Nygren H. Initial reactions of whole blood with hydrophilic andhydrophobic titanium surfaces. Coll Surf B: Biointerfaces1996;6:329}33.

1818 S. Kanagaraja et al. / Biomaterials 22 (2001) 1809}1818