From thNew South

SubmitThe Co

found at dReprint

Orthopaedversity of

© 20120883-5doi:10.1

The Journal of Arthroplasty Vol. 27 No. 8 2012

Influence of Electron Beam MeltingManufactured Implants on Ingrowth and Shear

Strength in an Ovine Model

Nicky Bertollo, PhD, Ruy Da Assuncao, MD, FRCS,Nicholas J. Hancock, MD, FRCS, Abe Lau, PhD, and William R. Walsh, PhD

Abstract: Arthroplasty has evolved with the application of electron beam melting (EBM) in themanufacture of porous mediums for uncemented fixation. Osseointegration of EBM and plasma-sprayed titanium (Ti PS) implant dowels in adult sheep was assessed in graduated cancellousdefects and under line-to-line fit in cortical bone. Shear strength and bony ingrowth (EBM) andongrowth (Ti PS) were assessed after 4 and 12 weeks. Shear strength of EBM exceeded that for TiPS at 12 weeks (P = .030). Ongrowth achieved by Ti PS in graduated cancellous defects followed adistinctive pattern that correlated to progressively decreasing radial distances between defect andimplant, whereas cancellous ingrowth values at 12 weeks for the EBM were not different.Osteoconductive porous structures manufactured using EBM present a viable alternative totraditional surface treatments. Keywords: electron beam melting, uncemented fixation, rapidprototyping, bone ingrowth, shear strength.© 2012 Elsevier Inc. All rights reserved.

Uncemented fixation in joint arthroplasty is reliantupon the establishment of a biologic and mechanicalinterlock at the implant-bone interface through osseoin-tegration. Roughened surfaces and porous mediums,such as sintered beads, wire meshes, and plasma-sprayed titanium (Ti PS), have been applied to metalsubstrates in various forms to provide for adequateanchorage via de novo cortical and cancellous boneongrowth and ingrowth [1].Rapid manufacturing technologies once exclusive to

the aeronautics industry are now being applied in thebiomedical sector for the manufacture of osteoconduc-tive porous mediums for tissue ingrowth in uncementedfixation [2-9]. Electron beammelting (EBM) is one suchmethod built on the fundamental tenet of additivemanufacturing. Complex meshes and macro-texturedmediums/shapes can be generated in a single processfrom powders of pure titanium and its alloys (ie, Ti-6Al-4V) that are sintered or melted together in a layer-by-

e Surgical and Orthopaedic Research Laboratories, University ofWales, Prince of Wales Clinical School, Sydney, Australia.ted October 31, 2011; accepted February 27, 2012.nflict of Interest statement associated with this article can beoi:10.1016/j.arth.2012.02.025.requests: William R. Walsh, PhD, Director, Surgical andic Research Laboratories, Prince of Wales Hospital, Uni-New South Wales, Sydney, Australia.Elsevier Inc. All rights reserved.

403/2708-0003$36.00/0016/j.arth.2012.02.025

142

layer fashion using electron beams. One of the mainadvantages of EBM technology is the ability to integratethe porous structure to the solid substrate instead oftraditional methods in which a coating is appliedseparately [2,10,11]. Porosity of these structures can betightly engineered such that the resulting constructsmimic the elastic modulus of human cancellous bone(∼0.5 GPa) and potentially ameliorate the effects ofstress shielding on bone resorption [2,10]. The primarymode of fixation for uncemented tibial trays, femoralcomponents, and acetabular cups is indeed via cancel-lous bone ongrowth and ingrowth.This study evaluates the osseointegration of a macro-

textured ingrowth structure manufactured using theEBM process and a control Ti PS medium after 4 and 12weeks in situ using an established ovine implantationmodel in the cortex of the tibia [12-15] and cancellousbone of the distal femur and proximal tibia [16]. The TiPS coating and EBM structure represented mediums forbone ongrowth and ingrowth, respectively. Our nullhypothesis was that there would be no differencesbetween implants with respect to interfacial shearstrength in cortical bone. We also evaluated the effectsof surgical interface (gap, line to line, and interference)on de novo ongrowth/ingrowth in the cancellous sitesof the distal femur and proximal tibia [16]. Ouradditional hypothesis was that implantation configura-tion would have no effect on osseointegration for eachof the 2 test mediums.

9

1430 The Journal of Arthroplasty Vol. 27 No. 8 September 2012

Materials and MethodsSix skeletally mature adult male sheep (crossbred

Merino Wethers, 18 months, 54.8 ± 1.4 kg) were usedwith ethical consent from our institutional animal careand ethics body. Animals underwent a bilateral proce-dure in which n = 10 pre-prepared implant dowels(6-mm diameter and 25-mm long) were implanted intosurgically created defects in the cancellous bone (n = 4)of the distal femur and proximal tibia [16] and corticalbone (n = 6) of the tibial diaphysis [12-16]. Theimplantation sites are summarized in Fig. 1. All implantswere sterilized using γ irradiation.Surfaces evaluated in this study included one manu-

factured using the process of EBM and another madeusing the process of plasma spraying (Ti PS) (Fig. 2),

Fig. 1. Implantation schedule for the cancellous (distal femurand proximal tibia) and cortical (tibial midshaft) sites in thisbilateral model. The sectioned illustrations provide an overviewof the cancellous gap and cortical implantation conditions.

with the latter serving as a comparison of ultimateinterfacial shear strength only. The EBM process is alayer-based additive manufacturing process in which abeam of electrons melt layers of titanium powder toform a 3-dimensional (3D) construct. The processpermits a unique porous-structured construct that isintegrated into the solid substrate, allowing a strongerporous-solid interface versus a porous coating that issintered to a solid surface [2,10,11]. Pore size andporosity of the EBM construct ranged from 130 to 370μm and 46% to 57%, respectively.The surface roughness of both implants was deter-

mined using a profilometer (Surfanalyzer EMD-5400;Federal Products Co). Six implants of each type weremeasured along the implant length in each of 4quadrants, and average (RA) and root mean square(RQ) average roughness values were computed.The surgical models used in this study have previously

been described in detail [16]. Briefly, a step drill wasused to create a single defect with graduated andconcentric 10-, 8-, 6-, and 4-mm diameter steps in thecancellous bone of the distal femur and proximal tibia.The length of each step was dictated by the drill steplength (eg, 6 mm). This series of holes was thenoverdrilled with a 5.5-mm diameter bit. A custom-builtimpactor was used to insert implants into the cancellousdefects with the 10-, 8-, 6-, and 5.5-mm steppeddiameter defects, producing 2-mm circumferential gap,1-mm circumferential gap, line-to-line, and interferencefits, respectively. The impactor ensured that each of theimplants was inserted to the correct depth, such thatequal amounts of the implant length were exposed toeach of the 4 simulated implantation scenarios. Anadditional step (12-mm diameter) was created in thecortex, the window through which the defect was drilledto position the contained gap defect deeper within thecancellous bone bed in an effort to further disjoin it fromany immediate cortical and periosteal influence.For cortical implantation, 3- and 6-mm diameter drills

were used in sequence to create 3 bicortical cylindricaldefects in the diaphysis of each tibia, as previouslydescribed [12-16]. Defects were at least 20 mm apart,and implants were inserted in a line-to-line fashion.Irrigation with sterile saline was applied during drillingof all bony defects. After implant placement theperiosteum, soft tissues and dermis were closed in layersusing 3-0 Vicryl (Ethicon Inc, Somerville, NJ) and 3-0 Polysorb (Covidien, Mansfield, Mass), respectively.Animals were returned to pasture after recovery andrandomly assigned for euthanasia at 4 or 12 weeks aftersurgery (n = 3 animals per time point).The femur and tibia from each aspect were harvested,

photographed, and radiographed (anteroposterior andlateral planes). Implants and surrounding cancellous orcortical bone were isolated using a handheld saw. Axialsections along the length of the tibia were made to

Fig. 2. Scanning electron microscope overview of the implants used in the current study. Top row is the EBM surface; bottomrow, Ti PS surface.

Osseointegration of Porous Implants in an Ovine Model � Bertollo et al 1431

isolate the cortical specimens for mechanical testing.Cortical specimens were further sectioned perpendicularto the long axis of the implant (sagittal plane) midwayalong their length using a low-speed saw (BeuhlerIsoMet; Beuhler, Lake Bluff, Ill) fitted with a diamondwafering blade to generate 2 mechanical test samples(medial and lateral aspects) from each bicortical implan-tation site. All mechanical testings were performed onfresh samples on the day of euthanasia.A standard push-out test was used for evaluation of

the shear strength (megapascal) of the bone-implantinterface [12-16]. Specimens were displaced at 5 mm/min using a closed-loop servo-hydraulic testing machine(MTS MiniBionix, Minneapolis, Minn). Clearance be-tween the specimen and edge of the hole in the supportjig (approximately 1 mm) was greater than theminimum 0.7 mm specified by Dhert et al [17].Postprocessing of force-displacement data was per-formed using an in-house script written for MatlabR2009a (Mathworks Inc, Natick, Mass) in which themaximum push-out force (Newton), stiffness (Newtonper millimeter), energy to failure (Joules), and proofresilience (Joules) were computed. Cortical thicknessmeasurements were determined from the scanningelectron microscope (SEM) images taken after poly-methylmethacrylate embedding for the calculation ofultimate interfacial shear strength (megapascal).Cancellous and mechanically tested cortical samples

were fixed in phosphate-buffered formalin, dehydratedthrough increasing concentrations of alcohol (70%-100% ethanol), and embedded in polymethylmethacry-late resin. Cortical samples were bisected through the

long axis of the implant using a Buehler linear precisionsaw (Beuhler Isomet 5000; Beuhler). This cut was offsetfrom the center of the implant cross-section by anamount equivalent to the thickness of the diamond-coated wafering blade (635 μm) to ensure that thesurface to be analyzed with histomorphometric andhistologic end points was indeed coincident with thelong axis of the implant. Cancellous specimens weresectioned perpendicular to the long axis of the implantsuch that the individual sections were coincident withthe center of each implantation scenario (ie, step).The implant cross-section for each sample (cancellous

and cortical) was imaged using an environmental SEMoperating in back-scattering mode (Hitachi 1000; HitachiCo Ltd, Japan). Images for cortical specimens were taken(magnification ×40) to examine the cortical thicknessand bone-implant interface in addition to the mode offailure during push-out testing. Postprocessing of SEMimages was performed using an in-house script writtenfor Matlab. For images containing the textured surface, aperimeter was established for each image enveloping theporous substrate, denoting the region of interest in asemiautomated process. The collective number of pixelsoccupied by bone and implant void within this region ofinterest was calculated; this was equivalent to theporosity of the implant in each image or available voidfor bone infiltration. Bone in available void (BIAV) wasquantified for the EBM surface only, calculated bydividing the number of pixels occupied with bone by thenumber of pixels corresponding to the available void. Inthis way, bone infiltration was normalized to theamount of available void, which is a direct function of

Fig. 3. Histomorphometric analysis of the EBM poroustextured surface in the cortical site. In this example, thereis 62.1% BIAV. An automated process was used todemarcate the outer bound of the implant porosity in eachregion and was based on a minimum line length (in pixels)and radius of curvature.

Table. Mechanical Parameters for the Implants Used inthe Study

Parameter Ti PS EBM P

4 wk Energy tofailure (J)

136.2 (111.4) 150.1 (102.3) .981

Stiffness(N/mm)

4436.8 (1617.6) 5333 (1911.9) .455

Proofresilience (J)

80.7 (75.5) 75.3 (64.5) .987

Shear strength(MPa)

8.8 (5.4) 9.7 (4.5) .657

12 wk Energy tofailure (J)

600.1 (237.1) 900 (419.8) .146

Stiffness (N/mm) 9578.5 (3352.6) 11 034.7 (2634.8) .413Proof resilience (J) 211.2 (125.6) 263.4 (112.7) .894Shearstrength (MPa)

24.4 (6.8) 29.8 (10.3) .03

Significant differences are indicated in bold. Mean and SD arepresented in parentheses.

1432 The Journal of Arthroplasty Vol. 27 No. 8 September 2012

the porosity of the 3D structure. An example of ananalyzed image is presented in Fig. 3. In a separateprocess, bone ongrowth onto the Ti PS implant was definedas the fraction of the implant surface in each image indirect contact with bone using a purpose-written scriptin Matlab.After SEM analysis, 20-μm sections were obtained

from the implant cross-section of cortical and cancelloussamples using a saw microtome (SP 1600; LeicaMicrosystems, Germany) and stained with methyleneblue and basic fuchsin for examination under a lightmicroscope (Olympus, Japan), and the images werequalitatively assessed for bone morphology.A 2-way analysis of variance was used in the analysis

of both cancellous histomorphometric (time and im-plantation scenario) and mechanical (implant and time)data. Post hoc, a Tukey's HSD test was applied, and allstatistical analyses were performed using SPSS 17.0(SPSS Inc, Chicago, Ill). Differences were considered tobe significant where P b .05.

ResultsBoth RA and RQ surface roughness values for the EBM

material were significantly higher than the Ti PS coating(P = .002 and P = .003, respectively). The RA and RQ

values for the EBM and Ti PS coatings were 50.52 ± 8.54μm and 62.15 ± 10.03 μm and 43.45 ± 5.90 μm and53.95 ± 7.70 μm, respectively.All animals recovered uneventfully after surgery and

reached their allocated time point with no infections orpostoperative morbidities noted.Interfacial shear strength (megapascal) from the tibial

cortical implantation sites increased significantly (P b.001) from 4 to 12 weeks postoperatively for both

implant types. Although there were no significantdifferences between the implant types at 4 weeks, themean ultimate shear strength of the EBM porousimplant was significantly greater than that of the Ti PS(P = .030) after 12 weeks in situ. The EBM porousstructure was associated with the higher mean ultimateshear strength at both 4 and 12 weeks, 9.7 ± 4.5MPa and29.8 ± 10.3 MPa, respectively, as compared with the TiPS implant at 4 and 12 weeks, 8.8 ± 5.4 MPa and 24.4 ±6.8 MPa, respectively. Mode of failure for the push-outtesting from the cortical bone of the tibia was repeatedlycoincident with the margin of the de novo bone and hostcortex, with no delamination or failure of either porousstructure observed. Stiffness (Newton per millimeter),energy to failure (Joules), and proof resilience data(Joules) are presented in the Table. No significantdifferences in mechanical parameters, other than shearstrength at 12 weeks, were observed.Cortical ingrowth increased significantly (P b .001)

from4 to 12weeks for the EBMmaterial, 27.3%±17.2%and 67.6% ± 9.9%, respectively. Similarly, fraction boneongrowth for the Ti PS implant also increased signifi-cantly with time in the cortical site from 38.1% ± 25.6%to 45.3% ± 13.2% at 4 and 12 weeks (P b .05).Ongrowth data for the Ti PS implant in the cancellous

sites (distal femur and proximal tibia) is contained inFig. 4. A trend of progressively increasing meanongrowth was identified for the 2-mm gap, 1-mmgap, line-to-line, and press fits at 4 weeks, and therewas significantly more ongrowth associated with thepress fit than both the 2-mm (P b .001) and 1-mm gap(P b .001) fits. The increased mean ongrowth at thepress fit was not significantly more than the line-to-linefit (P = .265) at 4 weeks. A similar trend was identified12 weeks postoperatively, with the exception of the 2-mm gap fit that had greater mean ongrowth than the1-mm gap, although it was not statistically significant

Time(weeks)

412

50

40

30

20

10

02mm Gap 1mm Gap L2L Press

Implantation Scenario

Per

cen

tag

e C

ance

llou

s O

ng

row

th

Fig. 4. Percentage of bone ongrowth in the cancellous sitesassociated with the Ti PS implant as a function of implantationscenario and time in situ (mean ± SE).

Fig. 6. Cortical ingrowth associated with the EBM coating.Infiltration of bone to the implant substrate is seen, as isintimate contact of cortical bone with the EBM porous domain.This image is representative of the general findings that theEBM material is osteoconductive in this ovine model.

Osseointegration of Porous Implants in an Ovine Model � Bertollo et al 1433

(P = .177). The increase in mean ongrowth with timewas not significant for either of the implantationsimulations in cancellous bone (P N .109).Cancellous ingrowth data for the EBM implant is

presented in Fig. 5, where BIAV increased significantlyunder each implantation scenario from 4 to 12 weekspostoperatively. Line-to-line implantation was associated

Time(weeks)

412

50

60

40

30

20

10

02mm Gap 1mm Gap L2L Press

Implantation Scenario

Can

cello

us

Ing

row

th (

%)

*

*

*

*

P = .026

P < .001

Fig. 5. Cancellous ingrowth for the EBM implant as a functionof implantation scenario and time in situ (mean ± SE). Asteriskdenotes significant (P b .05) increase in cancellous ingrowthwithin each implantation scenario.

with the highest mean ingrowth at 4 weeks, which wassignificantly more than that associated with the 2-mmgap. Twelve weeks postoperatively, there was signifi-cantly more ingrowth associated with the interference fitthan the 2-mm gap. No other differences in ingrowthwere observed.Histology confirmed the absence of an intervening

fibrous layer in both the cortical and cancellous sites forboth implant types, with intimate contact betweenimplant and bone present at both time points. Abundantbone formation could be seen in the porous domain ofthe EBM porous structure, as evidenced by thepenetration of bone to the level of the substrate inboth the cortical and (Fig. 6) cancellous (Fig. 7) sites. Denovo–woven bone occupied the porous implant domainat 4 weeks postoperatively, which had remodeled andmatured by 12 weeks, as indicated by the presence ofhaversian systems. Similarly, intimate bony contact withthe Ti PS implants in both bony sites was noted, with denovo bone formation extending deep into the regionsbetween surface asperities (Fig. 8).

DiscussionIn this study, 2 uncemented fixation surfaces (EBM

and Ti PS) were assessed for mechanical and biologicfixation in cortical bone [12-15] as well as the responsein cancellous bone [16] using well-defined ovinemodels. Interfacial shear strength increased significantlywith time for both implant types (P b .001). Twelveweeks postoperatively, the EBM achieved a value of29.8 MPa, which was the higher of the 2 test surfaces.The Ti PS implant was incorporated into this study as acomparison for interfacial shear strength on the basisthat it represents an implant surface technology oftenused in uncemented joint arthroplasty [1].

Fig. 7. Histologic sections at 12 weeks depicting cancellous ingrowth into the EBM coating as a function of implantation scenario;sections taken perpendicular to the implant long axis. Intimate contact between bone and coating is evident.

1434 The Journal of Arthroplasty Vol. 27 No. 8 September 2012

Graduated cancellous defects in the present studyenabled an assessment of osseointegration underdifferent implantation scenarios. Clinically, 3 basictypes of surgical fit (gap, line to line, and interference)are permissible at any bone-implant interface inuncemented fixation and have implications for theextent of osseointegration [16]. In the absence ofaugmentation from an osteoconductive calcium phos-phate or similar coating, one would reasonably surmisefor a plain cylindrical dowel the 2-mm gap, 1-mm gap,and line-to-line fits to be associated with progressivelyincreasing ongrowth at any given snapshot in time. Thisassumption is made on the basis that each type of fitrepresents progressively smaller radial distances fromthe defect margin to the implant surface. A meanongrowth trend of this sort was indeed identified for theTi PS implant after both 4 and 12 weeks in situ, with theexception of the 2-mm gap, which overtook the 1-mmgap at the latter time point (P N .9). Interestingly,although the mean value for each implantation scenarioincreased with time, it had failed to reach significancefor either fit type by 12 weeks (P N .109).

Fig. 8. Histology at 12 weeks depicting cortical bone growthonto the Ti PS surface. The limited inherent porosity of thiscoating is evident in this image.

Whether a press fit in cancellous bone is associatedwith greater ongrowth than a line-to-line implantationis less intuitive, although radial compaction (using aradial expansion tool for defect preparation) beforeimplantation of HA-coated porous titanium dowels in acanine cancellous gap model has been shown to increasepercentage bone ingrowth [18]. Conceivably, the pressfitting of cylindrical implants in the current study wasassociated with some minor level of compaction becauseof the physical overlap but did not translate into astatistically significant improvement in bone ongrowth.For the Ti PS coating, the interference fit was associatedwith higher mean ongrowth than for the line-to-line fitat both 4 and 12 weeks, but the disparity between thesefits at either time point was not significant (P N .99).The performance of the EBM and Ti PS surfaces in the

cancellous gap model differs by virtue of manufacturingmethod, making direct comparison of them difficult.Interestingly, by 12 weeks, ingrowth into the EBMimplants was similar for 1-mm, line-to-line, and pressfits, whereas a 2-mm gap remained inferior with respectto bone ingrowth. In contrast, bone ongrowth for the TiPS surface remained inferior for the 1- and 2-mm-gapscenarios compared with the line-to-line and press fitconditions. The porous nature of the EBM mediumappears to allow bone integration in all planes comparedwith 2-dimensional planar ongrowth, as is the case forthe Ti PS medium. This accounts for increased shearstrength values in cortical implantation sites for EBMcompared with Ti PS. Surface roughness of the EBMwasalso greater compared with the Ti PS, which may alsocontribute to superior fixation [19].Several investigators have assessed both the in vitro

and in vivo performances of porous domains for hardtissue ingrowth manufactured using rapid manufactur-ing technologies. Recently, Ponader et al [8] usedcalvarial (temporal) implantation in the porcine modelto demonstrate the bone ingrowth into 3D porousscaffolds manufactured using selective EBM. Ingrowthpercentages of 30% and 46% were found at 30 and 60days postoperatively, respectively. Biemond et al [19]recently reported the in vivo results of EBM implants in

Osseointegration of Porous Implants in an Ovine Model � Bertollo et al 1435

cancellous bone using a goat model. Bone ingrowth wasevaluated at 4 and 6 weeks using routine histology andfluorochrome labels, whereas no mechanical testing wasperformed. Cancellous ingrowth was reported to be upto 18.1% at 6 weeks. This cancellous data compares wellwith the performance of the EBM implants in thecurrent study under line-to-line implantation, also incancellous bone (Fig. 5).As with all preclinical models, the limitations of the

current study must be recognized. The sheep chosenwere young adults (18-month-oldWethers) and providea well-healing bone bed because of the young age.However, this allowed the comparison of data with thatobtained from a previous analysis of different porousand surface-treated implants [12-14,16]. Mean percent-age ingrowth values at 12 weeks of 32%, 46%, 37%,and 47% for the 2-mm gap, 1-mm gap, line-to-line, andinterference fits in the current study are comparablewith 25%, 34%, 45%, and 38% obtained by poroustitanium dowels (Regenerex; Biomet, Warsaw, Ind)using the same model [16]. An important distinctionbetween these 2 studies is that the said titanium dowelswere completely porous, whereas the EBM porouscoating was confined to the implant periphery. Withrespect to shear strength, the ultimate mean value of29.8 MPa achieved by the EBM material at 12 weeks isless than the mean 35 MPa obtained by a porous-beadedconstruct (Porocoat; Johnson & Johnson, DePuy, Leeds,United Kingdom) [14] but more than the 26.1 MPaencountered for porous titanium (Regenerex; Biomet)[16]. With respect to other ongrowth surface technol-ogies evaluated in these models, a mean shear strengthof 10 MPa for a grit-blasted titanium surface at 12 weeks[14] was substantially inferior to the 24.4 MPa for the TiPS implant in the current study.The results of this study demonstrate comparable

ingrowth values that create a stable ingrowth constructwhile also maintaining the mechanical strength due tothe integration of the porous structure and solidsubstrate because of the EBM manufacturing process.Similar bone ingrowth can be achieved with poroussubstrates, for example, porous tantalum (Hedrocel, alsoknown as Trabecular Metal; Zimmer, Warsaw, Ind)[20,21] or porous titanium (Regenerex; Biomet) [16],but they have the limitation with respect to prematurecompressive failure attributed to the large porosity andlack of a solid metal core before failure at the implant-bone interface [20].

References1. Ryan G, Pandit A, Apatsidis DP. Fabrication methods of

porous metals for use in orthopaedic applications. Bio-materials 2006;27:2651.

2. Heinl P, Muller L, Korner C, et al. Cellular Ti-6Al-4Vstructures with interconnected macro porosity for bone

implants fabricated by selective electron beam melting.Acta Biomaterialia 2008;4:1536.

3. Li X, Wang CT, Zhang WG, et al. Properties of a porousTi-6Al-4V implant with a low stiffness for biomedicalapplication. Proc Inst Mechan Eng Part H-J Eng Med2009;223:173.

4. Marin E, Fusi S, Pressacco M, et al. Characterization ofcellular solids in Ti6Al4V for orthopaedic implant applica-tions: trabecular titanium. J Mechanic Behav BiomedMater 2010;3:373.

5. Murr LE, Gaytan SM, Medina F, et al. Next-generationbiomedical implants using additive manufacturing ofcomplex, cellular and functional mesh arrays. PhilosTrans Royal Soc London, Series a (Mathematical, Physical& Engineering Sciences) 2010;368:1999.

6. Murr LE, Quinones SA, Gaytan SM, et al. Microstruc-ture and mechanical behavior of Ti-6Al-4V producedby rapid-layer manufacturing, for biomedical applica-tions. J Mechan Behav Biomed Mater 2009;2:20.

7. Ponader S, Vairaktaris E, Heinl P, et al. Effects oftopographical surface modifications of electron beammelted Ti-6Al-4V titanium on human fetal osteoblasts.J Biomed Mater Res Part A 2008;84:1111.

8. Ponader S, von Wilmowsky C, Widenmayer M, et al.In vivo performance of selective electron beam–meltedTi-6Al-4V structures. J Biomed Mater Res Part A2010;92:56.

9. Murr LE, Amato KN, Li SJ, et al. Microstructure andmechanical properties of open-cellular biomaterials pro-totypes for total knee replacement implants fabricated byelectron beam melting. J Mech Behav Biomed Mater2011;4:1396.

10. Parthasarathy J, Starly B, Raman S, et al. Mechanicalevaluation of porous titanium (Ti6Al4V) structures withelectron beammelting (EBM). J Mechan Behav of BiomedMater 2010;3:249.

11. Thomsen P, Malmstrom J, Emanuelsson L, et al. Electronbeam-melted, free-form-fabricated titanium alloy im-plants: material surface characterization and early boneresponse in rabbits. J Biomed Mater Res B Appl Biomater2009;90:35.

12. Svehla M, Morberg P, Bruce W, et al. No effect of atype I collagen gel coating in uncemented implantfixation. J Biomed Mater Res Part B, Appl Biomater2005;74:423.

13. Svehla M, Morberg P, Bruce W, et al. The effect ofsubstrate roughness and hydroxyapatite coating thick-ness on implant shear strength. J Arthroplasty 2002;17:304.

14. Svehla M, Morberg P, Zicat B, et al. Morphometric andmechanical evaluation of titanium implant integration:comparison of five surface structures. J Biomed Mater Res2000;51:15.

15. Chen D, Bertollo N, Lau A, et al. Osseointegration ofporous titanium implants with and without electrochem-ically deposited DCPD coating in an ovine model. J OrthopSurg Res 2011;6.

16. Bertollo N, Matsubara M, Shinodo T, et al. Effect ofsurgical fit on integration of cancellous bone and implantcortical bone shear strength for a porous titanium.J Arthroplasty 2011;26:1000.

1436 The Journal of Arthroplasty Vol. 27 No. 8 September 2012

17. Dhert WJ, Verheyen CC, Braak LH, et al. A finite elementanalysis of the push-out test: influence of test conditions.J Biomed Mater Res 1992;26:119.

18. Kold S, Rahbek O, Zippor B, et al. Bone compactionenhances fixation of hydroxyapatite-coated implants in acanine gap model. J Biomed Mater Res Part B, ApplBiomater 2005;75:49.

19. Biemond JE, Aquarius R, Verdonschot N, et al. Frictionaland bone ingrowth properties of engineered surface

topographies produced by electron beam technology.Arch Orthop Trauma Surg 2011;131:711.

20. Bobyn JD, Stackpool GJ, Hacking SA, et al. Character-istics of bone ingrowth and interface mechanics of anew porous tantalum biomaterial. J Bone Joint Surg Br1999;81:907.

21. Bobyn JD, Toh KK, Hacking SA, et al. Tissue response toporous tantalum acetabular cups: a canine model. JArthroplasty 1999;14:347.