Leading Research PaperTissue Engineering

Int. J. Oral Maxillofac. Surg. 2006; 35: 928–934doi:10.1016/j.ijom.2006.03.024, available online at http://www.sciencedirect.com

A comparative analysis ofscaffold material modificationsfor load-bearing applications inbone tissue engineering§

H. Chim, D. W. Hutmacher, A. M. Chou, A. L. Oliveira, R. L. Reis, T. C. Lim, J. -T.Schantz: A comparative analysis of scaffold material modifications for load-bearingapplications in bone tissue engineering. Int. J. Oral Maxillofac. Surg. 2006; 35:928–934. # 2006 International Association of Oral and Maxillofacial Surgeons.Published by Elsevier Ltd. All rights reserved.

§ Presented in part at the First BiennialMeeting of the European Tissue EngineeringSociety (ETES) in November 2001.

0901-5027/100928 + 07 $30.00/0 # 2006 Interna

tional Association of Oral and Maxillofacial Surgeo

H. Chim1,2, D. W. Hutmacher2,A. M. Chou3, A. L. Oliveira4,R. L. Reis4, T. C. Lim1,J.-T. Schantz1,2

1Division of Plastic and ReconstructiveSurgery, National University Hospital, 5 LowerKent Ridge Road, Main Building, Level 2,Singapore 119074, Singapore; 2Division ofBioengineering, National University ofSingapore, 5 Lower Kent Ridge Road,Singapore 119074, Singapore; 3Departmentof Biological Sciences, S1A 05-36, NationalUniversity of Singapore, Singapore 119260,Singapore; 43B’s Research Group:Biomaterials, Biodegradables andBiomimetics, Department of PolymerEngineering, University of Minho, Campus deGualtar, 4710-057 Braga, Portugal

Abstract. To facilitate optimal application of appropriate scaffold architectures forclinical trials, there is a need to compare different scaffold modifications undersimilar experimental conditions. In this study was assessed the effectiveness ofpoly-e-caprolactone (PCL) scaffolds fabricated by fused deposition modelling(FDM), with varying material modifications, for the purposes of bone tissueengineering. The incorporation of hydroxyapatite (HA) in PCL scaffolds, as well asprecalcification through immersion in a simulated body fluid (SBF) to produce abiomimetic apatite coating on the scaffolds, was assessed. A series of in vitro studiesspanning 3 weeks as well as in vivo studies utilizing a subcutaneous nude mousemodel were carried out. PCL and HA–PCL scaffolds demonstrated increasing tissuegrowth extending throughout the implants, as well as superior mechanical strengthand mineralization, as evidenced by X-ray imaging after 14 weeks in vivo. Nosignificant difference was found between PCL and HA–PCL scaffolds.Precalcification with SBF did not result in increased osteoconductivity and cellproliferation as previously reported. Conversely, tensile forces exerted by tissuesheets bridging adjacent struts of the PCL scaffold caused flaking of the apatitecoating that resulted in impaired cell attachment, growth and mineralization. Theresults suggest that scaffolds fabricated by FDM may have load-bearingapplications.

Key words: bone tissue engineering; fuseddeposition modelling; polymer scaffolds; hydro-xyapatite; precalcification.

Accepted for publication 15 March 2006Available online 9 June 2006

There is a need for an alternative to auto-genous bone as graft material for thereconstruction of cranio-facial and orbitaldefects, or as filler material in orthopaedicsurgery and fracture fixation. Bone frac-

tures and trauma-related injuries accountfor more than 1.3 million surgical proce-dures per year in the United States8, andreconstructive surgery requiring bonegrafts is carried out every day in hospitals

ns. Published by Elsevier Ltd. All rights reserved.

Scaffold material modifications for load-bearing applications in bone tissue engineering 929

Fig. 1. Scanning electron micrographs showing: (A) PCL scaffold with a lay-down pattern of08/608/1208 fabricated by FDM; (B) HA–PCL scaffolds have a fine apatite coating, obvious athigh magnification (C). (D) Precalcification of PCL scaffolds with simulated body fluidproduces a coarse, prominent apatite layer, obvious at high magnification (E). Originalmagnification: (A, B and D) �50 and (C and E) �750.

worldwide. Bone tissue engineering pro-mises such an alternative, and manygroups have developed strategies basedon seeding of cells onto 3-dimensionalbiodegradable polymeric scaffolds. Fabri-cation techniques that have been describedfor these scaffolds are myriad and includefused deposition modelling (FDM)6, sol-vent casting/salt leaching13, emulsionfreeze-drying22, phase separation10, 3Dprinting17 and gas foaming/salt leaching15,to name just a few. Two major structuralclasses of biodegradable polymer scaffoldmade using these techniques are the‘foams’ or those with a structure like aporous sponge24, and scaffolds fabricatedby fused deposition modelling with areproducible internal architecture muchlike that of a honeycomb6. Foams havebeen described to have a structure muchlike that of trabecular bone5, and were theearliest polymer structures explored foruse in bone and cartilage tissue engineer-ing. Many studies have demonstrated lim-itations with this architecture in bonetissue engineering, such as limited tissuegrowth confined to the surface of thefoam2,7, as well as poor structural proper-ties, in particular low strength and stiff-ness4,23.

Scaffolds made of poly-e-caprolactone(PCL) and fabricated by fused depositionmodelling have also been described26,with a highly reproducible internal struc-ture based on layer-by-layer deposition ofthermoplastic material (Fig. 1A). Suchscaffolds were designed to retain theirmechanical properties for 5–6 monthsand then gradually degrade over a periodof 2 years26. With a fully interconnectedpore network, these scaffolds were shownto support proliferation of cells throughoutthe scaffold6,4, as well as retain theirmechanical properties after implantationin rabbits20, pigs18 and mice19. This tissueengineering strategy may have a load-bearing application, such as in the recon-struction of long-bone defects or of largecritical sized defects in the calvarium ororbit.

There is a need for studies to comparethe effect of various material modifica-tions under similar experimental condi-tions before tissue engineering moves toclinical trials. This is essential to ensurethat the optimal scaffold architecture for aspecific application is used in patients. Inthis study, the effect of incorporation ofhydroxyapatite (HA) (Fig. 1B and C), aswell as precalcification in simulated bodyfluid (SBF) (Fig. 1D and E), was comparedwith untreated PCL FDM scaffolds for thepurposes of bone tissue engineering. Thegrowth of osteoblasts on these scaffolds

was evaluated in vitro and in a nude mousemodel, as well as comparing the degrada-tion characteristics in vivo. A subcuta-neous nude mouse model was ideal forthis experiment, due to the different envi-sioned applications of these scaffolds inthe clinic.

Previous studies have shown that blendsof HA and polymers as composite scaf-folds enhance osteoconductivity, as wellas improving mechanical properties, cellsurvival and proliferation12,9. Improvedcell survival has been postulated to berelated to the maintenance of a neutralpH, due to basic resorption by-productsof HA buffering acidic resorption by-pro-ducts of aliphatic polyesters4. Surfacemodification of polymer scaffolds by theformation of a bone-like apatite layerthrough immersion in SBF has also beenreported14,25, and MG63 cells were shownto attach to Bioglass-filled polylactidefoams after immersion in SBF2. This formof surface modification is used with thesame aim as composite scaffolds incorpor-ating HA: to enhance osteoconductivityand improve the mechanical properties ofthe scaffold. To the authors’ knowledge,

this is the first report of a direct compar-ison between different material modifica-tions in a similar experimental setting forthe purposes of bone tissue engineering.

Materials and methods

Fabrication of scaffolds by fused

deposition modelling

PCL scaffolds were fabricated by FDM aspreviously described by HUTMACHER et al.6

For HA–PCL scaffolds, PCL powder (cat-alog no. 44, 074-4, Aldrich Chemical Co.,Milwaukee, WI, USA), Mn of approxi-mately 80,000 Da (gel permeation chromo-graphy) and melt index of 1.0 g/10 min(125 8C/44 psi ASTM D1238-73), andHA (microemulsion-derived CaPO4 pow-der) were dried separately for 24 h in avacuum oven at 120 and 40 8C, respec-tively. Composite pellets of PCL–HA werethen formed by casting a solvent mixture of20 g PCL–HA blend with 15% HA byweight. The composite material was cutinto 5 mm � 5 mm pellets and stored in adesiccator until used. The PCL–HA pelletswere melt extruded to form monofilaments

930 Chim et al.

using a one-shot extruder (Alex James &Associates, Greenville, SC, USA). HA–PCL scaffolds were then fabricated byFDM as described. PCL and HA–PCLscaffolds (PCL/HA = 85%/15%) with alay-down pattern of 0/60/120 and a porosityof 65% were used. All scaffolds had adimension of 10 mm � 10 mm � 8 mm.Prior to seeding they were sterilized bycentrifugation in 70% ethanol. Ethanolwas removed by washing 3 times for 1 heach in changes of phosphate-buffered sal-ine (PBS). Subsequently, the scaffolds weretransferred into 24-well plates, and dried inan incubator at 37 8C for 24 h.

Biomimetic coating using a SBF

To produce the biomimetic coatings thematerials were submitted to a previouslydeveloped procedure16. The PCL scaffoldswere ‘impregnated’ with a sodium silicategel. This treatment was aimed at generatingnucleating sites for the formation of theapatite layers. After the treatment the sam-ples were immersed in 50 ml of SBF (pH7.4, 37 8C) with an ion concentration nearlyequal to that of human blood plasma (SBF1�). After 7 days, the ion concentration ofthe SBF solution was raised to 1.5�, toinduce growth of apatite nuclei. SBF solu-tions were changed every 2 days. The coat-ing period was 30 days in duration. Aftercoating, samples were rinsed in de-ionizedwater and dried at room temperature.

Cell isolation, seeding and culture

Human calvarial osteoblasts were isolatedvia explant culture from a piece of calvar-ium obtained from an infant undergoingreconstructive surgery. Non-osseous tis-sue was mechanically removed and thespecimen was manually broken up intofragments measuring 5 mm � 5 mm.These fragments were extensively washedin PBS and seeded into T-75 culture flasks.The cells were cultivated in M199 med-ium (GIBCO, Rockville, MD, USA) sup-plemented with 10% foetal calf serum and1% penicillin–streptomycin (GIBCO).Osteogenic supplements (ascorbate dipho-sphate, b-glycerophosphate) were added.Medium was changed every 4 days.

Cell growth and differentiation in vitro in

combination with scaffolds

After 2 weeks of explant culture in T-75flasks, cells were detached with 0.25%trypsin–EDTA, centrifuged, washed withPBS, and seeded on polymer scaffolds.Three different groups of scaffolds weredefined: (1) PCL FDM scaffolds, (2) HA–

PCL FDM scaffolds and (3) PCL FDMscaffolds treated with SBF. Cells wereseeded at a density of 1 � 105 cells onscaffolds with dimensions of 10 mm �10 mm � 8 mm. The scaffolds were cul-tivated in M199 medium as described, andmedium was changed every 3–4 days. Tofacilitate osteogenic differentiation, cellswere stimulated with 10 mM dexametha-sone on days 3 and 10 of culture. Cell–scaffold constructs were maintained in 24-well plates. At various time points, super-natant was collected before fresh mediumwas added, and stored at �80 8C. Thesupernatant was subsequently analysedfor expression of alkaline phosphatase(ALP) and osteocalcin. The cell–scaffoldconstructs were serially observed under aphase-contrast light microscope.

Confocal laser microscopy was alsoused to assess cell morphology, prolifera-tion and attachment in vitro. Samples werefirst fixed in 3.7% formaldehyde at roomtemperature for 30 min. After rinsingtwice with PBS for 5 min each, RNaseA (200 mg/mL) was added and the sam-ples incubated at room temperature for30 min. Phalloidin (A12379 Alexa Fluo488 phalloidin; Molecular Probes,Eugene, OR, USA) was then added at1:200 dilution at room temperature andthe samples were kept in darkness for45 min. Samples were then counterstainedwith a 5 mg/mL concentration of propi-dium iodide (1 mg/mL; Molecular Probes)solution, dried and mounted for viewingunder a confocal laser microscope (IX70,Olympus). Cell viability and distributionwithin the scaffolds were assessed with thefluorescent dye fluorescein diacetate(Molecular Probes) and propidium iodideas the counterstain. Depth projectionimages were constructed from up to 50horizontal image sections (20 mm each)through the stained cell–scaffold con-structs.

Metabolic assays (in vitro)

To assess phenotype of osteoblasts seededon polymer scaffolds in vitro, ALP andosteocalcin assays were performed. Super-natant from different time points on days1–20 of incubation was used. This wasstored at�80 8C until assayed. For assess-ment of ALP (n = 3), the production of p-nitrophenol in the presence of ALP wasmeasured by a bone-specific immunoas-say (Metra BAP EIA kit; Quidel, SanDiego, CA, USA). The concentration ofthe major N-terminal fragment of osteo-calcin (n = 1) was assayed by an enzyme-linked immunoassay kit (Metra Osteocal-cin; Quidel).

Animal surgery

Cell–scaffold constructs were implantedsubcutaneously in the back of Balb C nudemice (n = 3) after 21 days in culture. The 3groups of constructs (1–3) describedabove were implanted together in the backof a single mouse. Animals tolerated sur-gery well. The mice were killed after 6 and14 weeks, and the specimen retrieved en-bloc for histology, mechanical testing andanalysis of degradation of scaffolds. Hous-ing and feeding of the animals was accord-ing to standard animal care protocols. Thestudy was approved by the Animal Wel-fare Committee, National University ofSingapore, and licensed by the NationalInstitute of Health’s guide for care and useof laboratory animals.

Mechanical testing and image analysis

Samples retrieved after 14 weeks in vivowere compared to matched unseeded scaf-folds of similar dimension and fabricationfor compression testing. Six differentpoints on each construct were tested onan Instron 4502 uniaxial testing systemwith a 10-N load cell (Canton, MA, USA)using an indenter with a diameter of0.625 mm. The specimens were firmlysecured onto a vise to ensure immobiliza-tion and a flat testing surface. The com-pression test was divided into 2 steps. Inthe first loading phase, a load of up to 10 Nwas applied at 0.1 mm/min. Results of thisphase were analysed and the stiffness(loading between 8 and 10 N) of eachreconstructed area was calculated. Thesecond phase included a holding phaseat 10 N for 60 s, for deformation ratemeasurements (creep). Afterwards theindenter was unloaded at 2.5 mm/min.The explanted en-bloc specimen consist-ing of all 4 scaffolds implanted in the micewas imaged with an X-ray scanning ana-lytical microscope (XSAM) (XGT-2000;Horiba, Tokyo, Japan).

Statistical analysis

Results of metabolic assays reported aremeans of at least 3 different samples ofsupernatant. Differences in mean values ofALP activity at different time points wereanalysed by unpaired, 2-tailed t-tests; Pvalues less than 0.05 were consideredsignificant.

Results

Cell growth and differentiation in vitro

Calvarial osteoblasts were successfullyisolated and seeded onto the 3 different

Scaffold material modifications for load-bearing applications in bone tissue engineering 931

Fig. 2. Phase-contrast light microscopy of scaffolds 11 days after seeding. (A) PCL and (B) HA–PCL scaffolds show a tissue sheet bridgingadjacent struts. (C) PCL scaffolds treated with SBF showing flaking of the induced apatite layer and subsequent detachment of cells from thescaffold bars. Original magnification: (A–C) �100.

groups of polymer scaffolds, after initialexpansion in 2-dimensional culture for 2weeks. When seeded onto all 3 types ofscaffold, cells attached, spread and prolif-erated, adopting a stellate morphologytypical of attached cells, with numerousfilopodia and cell-to-cell contacts visible.The PCL FDM and HA–PCL FDM scaf-folds demonstrated a constant rate of cellproliferation, with cells initially attachingto the scaffold bars and later formingsheets of cells bridging adjacent scaffoldstruts (Fig. 2A and B). Cells penetrateddeep into the centre of the PCL scaffolds,where tissue sheets were even observed.Confocal laser microscopy confirmed theuniform distribution of cells throughoutthe PCL (Fig. 3A) and HA–PCL(Fig. 3B) scaffolds. Grossly, there wasno difference in cell morphology or pro-liferation between the PCL and HA–PCLcell–scaffold constructs.

The PCL scaffolds treated with SBF didnot support cell growth as well as the PCLor HA–PCL scaffolds. In the initial phase,cells attached to the bars of the coated PCL

Fig. 3. Confocal laser microscopy with depth proof cells within the scaffolds. (A) PCL and (B) HAcentre of the construct. (C) Few cells are seen oncontact, unlike cells on the other 3 scaffolds. O

scaffolds. It was observed subsequentlythat, with formation of tissue sheets brid-ging adjacent struts, sheets of cellsdetached from the bars of the scaffolds(Fig. 2C). Grossly, flaking of the apatitelayer was seen on the bars of the scaffolds.This was probably a result of the apatitelayer being to thick, due to an overlyprolonged period of SBF immersion, suchthat the structural integrity of the coatingon the PCL fibres was affected. Furtherstudies to determine the optimum time forSBF immersion, so as to achieve a stable,continuous adherent apatite layer arerequired. Confocal laser microscopyshowed that while there were still cellsattached to the scaffold, these did not forma confluent layer, and were at a lowerdensity than in the other 3 scaffold groups(Fig. 3C).

Analysis of ALP and osteocalcinexpression by osteoblasts revealed a trendconsistent with that expected of calvarialosteoblasts in culture for the PCL and HA–PCL scaffolds. Previously, it was shownthat during culture of rat calvarial osteo-

jection images reconstructed from multiple horiz–PCL scaffolds show even distribution of cells alothe PCL scaffolds treated with SBF, and these doriginal magnification: (A–C) �50.

blasts, expression of osteocalcin synthesisoccurred subsequent to initiation of ALPactivity and together with the formation ofmineralized nodules1. ALP has beendescribed as an early marker in osteogen-esis, while osteocalcin is a late marker thatis produced synchronously with produc-tion of extracellular matrix1,4. ALP pro-duction peaked immediately after seedingin all scaffolds (Fig. 4A), as measured onday 1, falling to a trough and subsequentlyreaching a peak again between days 17and 20 after seeding. During the course ofthe in vitro study, ALP production wasconsistently lower for the SBF-treatedPCL scaffold than the PCL and HA–PCL scaffolds, the difference increasingsubstantially from day 10 after seedingand reaching statistical significance ondays 17 and 20 (P < 0.05). The lack ofALP expression for the apatite-coatedPCL is in concordance with that observedby light and confocal laser microscopy,and was probably due to an increasingnumber of non-viable cells as the apatitelayer sloughed off the bars of the scaffold.

ontal images shows 3-dimensional distributionng the bars of the scaffolds, extending into thenot form a confluent network with cell-to-cell

932 Chim et al.

Fig. 4. Metabolic assays were performed to assess expression of osteogenic markers during a21-day period in vitro. (A) ALP expression peaked immediately after seeding in all scaffolds,then slowly fell and peaked again between days 17 and 20 for the PCL and HA–PCL scaffolds.From day 10 onwards, ALP expression for the SBF-treated PCL scaffolds was consistentlylower, reaching statistical significance on days 17 and 20 (P < 0.05). (B) Osteocalcin (OC)expression peaked at day 10 for the PCL and HA–PCL scaffolds.

Similarly, for osteocalcin expression(Fig. 4B), while a peak was observedfor the PCL and HA–PCL scaffolds 10days after seeding, no peak was visible forthe SBF-treated PCL scaffolds. Formationof extracellular matrix was observedthroughout the PCL and HA–PCL scaffoldarchitecture, increasing gradually and fill-ing the whole scaffold by 3 weeks (datanot shown), as previously demon-strated4,21. Production of extracellularmatrix accompanied osteocalcin expres-sion in these scaffolds. For the SBF-trea-ted PCL scaffolds, the lack of osteocalcinexpression can be explained by an inabil-ity of cells to attach properly to the scaf-fold bars, and subsequently being unableto secrete and produce extracellularmatrix.

Fig. 5. Cell-seeded scaffolds were implanted subcutaneously in the back of Balb C nude miceafter 21 days in culture (A). Explantation of the implant 14 weeks later showed evidence ofvasculogenesis (B). In this specimen, 4 vessels can be seen emanating from the PCL construct,and 1 from the HA–PCL construct. (C) An X-ray taken of the explanted specimen showedevidence of mineralization (bright white areas) in PCL and HA–PCL scaffolds. This was lesspronounced in the SBF PCL scaffold.

Comparative analysis of different

scaffolds in a nude mouse model

The different scaffolds loaded with cellsafter a 3-week period of in vitro culturewere implanted simultaneously into theback of Balb C nude mice (Fig. 5A).

The implants were well integrated intothe surrounding tissue and there was noevidence of encapsulation or excessivefibrosis. Gross examination of theexplanted specimen 14 weeks afterimplantation revealed evidence of vascu-lar ingrowth, particularly pronounced forthe PCL and HA–PCL scaffolds. In therepresentative specimen shown (Fig. 5B),3 blood vessels can be seen emanatingfrom the PCL scaffold, and 1 from theHA–PCL scaffold. Gross evidence of vas-culogenesis was absent for the SBF-trea-ted PCL and poly(DL-lactic-co-glycolicacid) (PLGA) foams. Increased vasculo-genesis in the PCL and HA–PCL scaffoldswould be expected, considering increasedcell proliferation and expression of osteo-genic markers, as shown in the in vitrostudy. An X-ray micrograph of theexplanted specimen (Fig. 5C) showed thatwhile mineralization (bright white areas)was prominent in the PCL and HA–PCLscaffolds, it was less pronounced in theSBF-treated PCL scaffold and PLGAfoams. This is to be expected on the basisof the findings of the in vitro studydescribed above.

Mechanical testing was performed toassess the structural integrity of unseededcontrol scaffolds and specimens explantedafter 14 weeks in a nude mouse model.The mean stiffness (Fig. 6A) of the PCL,HA–PCL and apatite-coated PCL was

Scaffold material modifications for load-bearing applications in bone tissue engineering 933

Fig. 6. Mechanical testing was performed on explanted scaffolds after 14 weeks in vivo withmatched control unseeded scaffolds. (A) Mean stiffness is similar for PCL scaffolds. Stiffnessincreased after in vivo culture in the PCL and HA–PCL scaffolds, but decreased for the SBF PCLscaffolds. (B) Maximum load that could be supported before indentation of the scaffolds wasgreatest in the PCL and least in the SBF PCL scaffolds.

similar, as would be expected in scaffoldswith the same architecture and composi-tion, but the stiffness of the apatite-coatedPCL scaffold after 14 weeks implantationwas less than that of a control scaffold.Conversely, the stiffness of the PCL andHA–PCL specimens increased after theperiod of in vivo implantation; this wasexpected to occur due to ingrowth of tissueand formation of extracellular matrix. Thedecreased stiffness of the SBF-treatedPCL scaffolds after implantation couldbe explained by poor attachment of cellsto the unstable apatite layer in vivo, andtherefore decreased mechanical strength.The maximum load that could be sup-ported before indentation of the scaffold(Fig. 6B) was in the order of PCL, HA–PCL, SBF-treated PCL; again, this was asexpected.

Discussion

For tissue-engineering strategies to pro-gress to clinic trials, adequate studies mustbe performed to compare different scaf-fold modifications and cells under similarexperimental conditions. Here, it was

sought to compare different modificationson a scaffold designed for load-bearingapplications in bone tissue engineering.Scaffolds fabricated by FDM have ade-quate mechanical strength, and allowingrowth of tissue throughout the implant,as demonstrated. The fabrication processallows customized shaped implants to begenerated from computed tomography(CT) data6.

The effect was investigated of incor-poration of HA into the scaffold mate-rial, and precalcification with SBF toproduce a biomimetic apatite coatingin PCL scaffolds made by FBM. Whileprevious studies reported beneficialeffects of HA in composite scaffolds9,12,a significant difference was not foundbetween PCL and HA–PCL scaffolds inthe present experiments. It is postulatedthat, because of the fully interconnectednetwork and large porous spaces of theFDM architecture, at baseline osteocon-ductivity is already relatively high, andtherefore, HA incorporation did notresult in a significant increase. In addi-tion, the enhanced cell proliferation andmechanical properties attributed to HA

incorporation were reported from experi-ments conducted on ‘foams’, withdecreased cell penetration and poorermechanical properties compared to theFDM scaffolds.

While precalcification with SBF pro-mises similar benefits to incorporation ofHA into polymer scaffolds, it was foundthat, conversely, this was detrimental inthe present experiments. Studies havereported attachment of MG63 cells toBioglass-filled polylactide foams2, aswell as enhanced growth of human osteo-blast-like cells on polymer–glass compo-sites11 and glass fibres3 over 14 days inculture. While these findings are dis-puted, the 3-dimensional porous archi-tecture of the PCL FDM scaffoldsdictates different mechanical constraintson cells attaching and growing on thebars of the scaffold. Cells seeded onscaffolds such as those used in this studywill naturally form a tissue sheet brid-ging adjacent bars, as was shown hereand previously4,6,20,21. The tensile forceexerted by this free tissue sheet on thescaffold bars is greater than would beexpected for cells grown on foams orfibres, which attach individually to thesupporting material. This would explainthe poor attachment of cells to the SBF-treated PCL scaffold in vitro, and theobserved flaking of the apatite layer.While precalcification with SBF mightbe a viable form of surface treatment forscaffolds of a ‘foam’ architecture or for2-dimensional surfaces, the method ascurrently described may not be suitablefor 3-dimensional scaffolds made byFDM. A thinner apatite layer, achievedthrough shorter immersion times in SBFand lower SBF concentrations, may yieldmore promising results.

Acknowledgement. Authors would like tothank Ms Gouk Sok Siam for much invalu-able assistance with this project.

References

1. Aronow MA, Gerstenfeld LC, Owen

TA, Tassinari MS, Stein GS, Lian JB.Factors that promote progressive devel-opment of the osteoblast phenotype incultured fetal rat calvarial cells. J CellPhysiol 1990: 143: 213–221.

2. Blaker JJ, Gough JE, Maquet I,Notingher I, Boccaccini AR. In vitroevaluation of novel bioactive compositesbased on Bioglass-filled polylactidefoams for bone tissue engineering scaf-folds. J Biomed Mater Res 2003: 67A:1401–1411.

3. Clupper DC, Gough JE, Hall MM,Clare AG, LaCourse WC, Hench

934 Chim et al.

LL. In vitro bioactivity of S520 glassfibers and initial assessment of osteoblastattachment. J Biomed Mater Res A 2003:67A: 285–294.

4. Endres M, Hutmacher DW, Salgado

AJ, Kaps C, Ringe J, Reis RL, Sittinger

M, Brandwood A, Schantz JT. Osteo-genic induction of human bone marrow-derived mesenchymal progenitor cells innovel synthetic polymer–hydrogelmatrices. Tissue Eng 2003: 9: 689–702.

5. Holy CE, Fialkov JA, Davies JE, Shoi-

chet MS. Use of a biomimetic strategy toengineer bone. J Biomed Mater Res 2003:65A: 447–453.

6. Hutmacher DW, Schantz T, Zein I,Ng KW, Teoh SH, Tan KC. Mechanicalproperties and cell cultural response ofpolycaprolactone scaffolds designed andfabricated via fused deposition modeling.J Biomed Mater Res 2001: 55: 203–216.

7. Ishaug SL, Crane GM, Miller MJ,Yasko AW, Yaszemski MJ, Mikos

AG. Bone formation by three-dimen-sional stromal osteoblast culture in bio-degradable polymer scaffolds. J BiomedMater Res 1997: 36: 17–28.

8. Langer R, Vacanti JP. Tissue engineer-ing. Science 1993: 260: 920–926.

9. Laurencin CT, Attawia MA,Elgendy HE, Herbert KM. Tissueengineered bone-regeneration usingdegradable polymers: the formation ofmineralized matrices. Bone 1996:19(Suppl 1):93S–99S.

10. Lo H, Kadiyala S, Guggino SE,Leong KW. Poly(L-lactic acid) foamswith cell seeding and controlled-releasecapacity. J Biomed Mater Res 1996: 30:475–484.

11. Lu HH, El-Amin SF, Scott KD,Laurencin CT. Three-dimensional,bioactive, biodegradable polymer-bioactive, glass composite scaffoldswith improved mechanical propertiessupport collagen synthesis and miner-alization of human osteoblast-like cellsin vitro. J Biomed Mater Res 2003:64A: 465–474.

12. Ma PX, Zhang R, Xiao G, Franceschi

R. Engineering new bone tissue in vitroon highly porous poly (a-hydroxyl acids)/hydroxyapatite composite scaffolds. JBiomed Mater Res 2001: 54: 284–293.

13. Mikos AG, Sarakinos G, Leite SM,Vacanti JP, Langer R. Laminatedthree-dimensional biodegradable foamsfor use in tissue engineering. Biomater-ials 1993: 14: 323–330.

14. Murphy WL, Kohn DH, Mooney DJ.Growth of continuous bonelike mineralwithin porous poly(lactide-co-glycolide)scaffolds in vitro. J Biomed Mater Res2000: 50: 50–58.

15. Nam YS, Yoon JJ, Park TG. A novelfabrication method for macroporous scaf-folds using gas foaming salt as porogenadditive. J Biomed Mater Res B ApplBiomater 2000: 53: 1–7.

16. Oliveira AL, Malafaya PB, Reis RL.Sodium silicate gel as a precursor for thein vitro nucleation and growth of a bone-like apatite coating in compact and por-ous polymeric structures. Biomaterials2003: 24: 2575–2584.

17. Park A, Wu B, Griffith LG. Integra-tion of surface modification and 3Dfabrication techniques to prepare pat-terned poly(L-lactide) substrates allow-ing regionally selective cell adhesion. JBiomater Sci Polym Ed 1998: 9: 89–110.

18. Rohner D, Hutmacher DW, Tan KC,Oberholzer M, Hammer B. In vivoefficacy of bone-marrow-coated polyca-prolactone scaffolds for the reconstruc-tion of orbital defects in the pig. J BiomedMater Res Part B Appl Biomater 2003:66B: 574–580.

19. Schantz JT, Hutmacher DW, Chim H,Ng KW, Lim TC, Teoh SH. Induction ofectopic bone formation by using humanperiosteal cells in combination with anovel scaffold technology. Cell Trans-plant 2002: 11: 125–138.

20. Schantz JT, Hutmacher DW, Lam

CXF, Brinkmann M, Wong KM, Lim

TC, Chou N, Guldberg RE, Teoh SH.

Repair of calvarial defects with custo-mized tissue-engineered bone grafts II.Evaluation of cellular efficiency and effi-cacy in vivo. Tissue Eng 2003: 9(Suppl1):S127–S139.

21. Schantz JT, Teoh SH, Lim TC, Endres

M, Lam CXF, Hutmacher DW. Repairof calvarial defects with customized tis-sue-engineered bone grafts I. Evaluationof osteogenesis in a three-dimensionalculture system. Tissue Eng 2003: 9(Suppl1):S113–S126.

22. Whang K, Thomas CH, Healy KE,Nuber GA. Novel methods to fabricatebioabsorbable scaffolds. Polymer 1995:36: 837–849.

23. Yoon JJ, Kim JH, Park TG. Dexametha-sone-releasing biodegradable polymerscaffolds fabricated by a gas-foaming/salt-leaching method. Biomaterials2003: 24: 2323–2329.

24. Yoon JJ, Park TG. Degradation beha-viors of biodegradable macroporous scaf-folds prepared by gas foaming ofeffervescent salts. J Biomed Mater Res2001: 55: 401–408.

25. Yuan X, Mak AF, Li J. Formation ofbone-like apatite on poly(L-lactic acid)fibers by a biomimetic process. J BiomedMater Res 2001: 57: 140–150.

26. Zein I, Hutmacher DW, Tan KC, Teoh

SH. Fused deposition modeling of novelscaffold architectures for tissue engineer-ing applications. Biomaterials 2002: 23:1169–1185.

Address:Jan-Thorsten SchantzDivision of Plastic and Reconstructive

SurgeryNational University Hospital5 Lower Kent Ridge RoadMain BuildingLevel 2Singapore 119074SingaporeTel: +65 6 772 4226Fax: +65 6 772 8427E-mail: surjts@nus.edu.sg