The efficacy of cylindrical titanium mesh cage for the reconstruction of a critical-size canine segmental femoral diaphyseal defect

The Efficacy of Cylindrical Titanium Mesh Cage for theReconstruction of a Critical-Size Canine SegmentalFemoral Diaphyseal Defect

Ronald W. Lindsey,1 Zbigniew Gugala,1 Edward Milne,2 Michael Sun,3 Francis H. Gannon,4 Loren L. Latta2

1Department of Orthopaedic Surgery, Baylor College of Medicine, 6560 Fannin, Suite 1900, Houston, Texas 77030

2University of Miami, Miami, Florida

3Department of Orthopaedic Surgery, University of Chicago, Chicago, Illinois

4Department of Bone Pathology, Armed Forces Institute of Pathology, Washington, DC

Received 7 November 2005; accepted 28 December 2005

Published online 26 May 2006 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jor.20154

ABSTRACT: The authors developed anovel technique for the reconstruction of large segmental longbone defects using a cylindrical titanium mesh cage (CTMC). Although the initial clinical reportshave been favorable, theCTMCtechniquehas yet to be validated in a clinically relevant large animalmodel, which is the purpose of this study. Under general anesthesia, a unilateral, 3-cm mid-diaphyseal segmental defect was created in the femur of an adult canine. The defect reconstructiontechnique consisted of a CTMC that was packed and surrounded with a standard volume ofmorselized canine cancellous allograft and canine demineralized bone matrix. The limb wasstabilized with a reamed titanium intramedullary nail. Animals were distributed into fourexperimental groups: in Groups A ,B, and C (six dogs each), defects were CTMC reconstructed,and the animals euthanized at 6, 12, and 18 weeks, respectively; in Group D (three dogs), the samedefect reconstruction was performed but without a CTMC, and the animals were euthanized at 18weeks. The femurs were harvested and analyzed by gross inspection, plain radiography, computedtomography (CT), and single photon emission computed tomography (SPECT). The femurs weremechanically tested in axial torsion to failure; two randomly selected defect femurs from each groupwere analyzed histologically. GroupsA,B, andC specimens gross inspection, plain radiography, andCT, demonstrated bony restoration of the defect, and SPECT confirmed sustained biological activitythroughout the CTMC. Compared to the contralateral femur, the 6-, 12-, and 18-week mean defecttorsional stiffness was 44.4, 45.7, and 72.5%, respectively; the mean torsional strength was 51.0,73.6, and 83.4%, respectively. Histology documented new bone formation spanning the defect.Conversely, Group D specimens (without CTMC) demonstrated no meaningful bone formation,biologic activity, or mechanical integrity at 18 weeks. The CTMC technique facilitated healing of acanine femur segmental defect model, while the same technique without a cage did not. The CTMCtechnique may be a viable alternative for the treatment of segmental long bone defects. � 2006

Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 24:1438–1453,

2006

Keywords: titanium mesh cage; canine; femoral diaphyseal defect

INTRODUCTION

A large segmental defect in a long bone constitutesa challenging problem in orthopedic surgery.1

Although numerous surgical techniques have beenadvocated, none of them have reliably addressedall of the issues associated with this condition.Among considerations are restoration of skeletalcontinuity with biologically and biomechanically

sound bone; the early return to unrestricted limbfunction, minimal patient’s compro- mise, and/orcompliance throughout the course of the treat-ment, and applicability of the treatment withoutthe need for specialized surgical skills and/orequipment. Therefore, the optimal reconstructionmodality for large segmental defects in long boneshas yet to be established.

Major limitations exist in all currently utilizedtreatment modalities for large segmental bonedefects. Massive cancellous bone grafting is thestandard treatment for restoring continuity of longbone;2–4 however, it is not indicated for defects thatare greater than 6 cm in length, has difficultly

1438 JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006

Correspondence to: Ronald W. Lindsey (Telephone: 409-747-5757; Fax: 409-747-5745; E-mail: [email protected])

� 2006 Orthopaedic Research Society. Published by Wiley Periodicals,Inc.

maintaining graft in the defect, and does notaugment limb stability.3 Cortical allograft providesbetter structural support for the defect, butcomplete bony incorporation of this graft rarelyoccurs, and its susceptibility for infection is high.5

Vascularized cortical autografts have been advo-cated due to their more reliable incorporation andremodeling with the host;6–8 however, vascular-ized cortical grafting is a sophisticated, highlyspecialized procedure that requires prolongedprotected weightbearing, is associated with sig-nificant donor site morbidity, and has limitedavailability.9 Distraction osteogenesis has becomea popular alternative for the treatment oflarge bone defects.10–13 Although successful bonerestoration can be achieved with this modality,distraction osteogenesis can be protracted, painful,frequently complicated by pin site infections,fluctuates in quality and quantity of the newregenerate, and has healing at the docking sitewith bone transport.

Over the past decade a variety of osteoconduc-tive bonegrafts substitutes have been investigated,and they include ceramics or cements,14–16 andbioabsorbable polymers of synthetic or naturalorigin.17–20 Osteoconductive and osteoinductivesubstances such as extracted demineralized bonematrix (DBM), or pure osteoinductive substancessuch as recombinant growth factors (rhBMP), andcytokines have recently emerged.7,21–28 Treatingdefects with composites consisting of osteogeniccells (i.e., tissue engineering), osteoconductive, orosteoinductive substances has also been sug-gested.27,29–31 Although experimentally promis-ing, the efficacy of these approaches for largesegmental defects in long bones has not beenclinically determined.

Cylindrical titanium mesh cage implantsreceived clearance from the FDA in 1990 to bemarketed for osseous reinforcement and/or recon-struction. The successful use of the cylindricaltitaniummesh cage in combination with bone grafthas initially been reported for the reconstruction oflarge segmental long bone defects.32 Several sub-sequent reports have corroborated the efficacy ofthis treatmentmodality.33–35The biological advan-tages of this technique include its ability to utilizecancellous bone graft, maintain bone graft con-tinuitywith theadjacent boneand surrounding softtissue, and the potential for circumferential bonegraft reconstitution throughout the length of thedefect. Its mechanical advantages include thehollow and fenestrated nature of the implant,which provides themaximum strength for the leastamount of metal, thereby minimizing the risk for

stress shielding. Additionally, because the bonegraft can be both retained and loaded within thedefect, the functional advantage of this technique isimmediate limb stability, sufficient enough topermit early, active weight bearing.

Despite the encouraging preliminary clinicalexperience with this technique, its theoreticaladvantages have not been validated in a standar-dized experimental animal model. The objective ofthis study is to determine the biological andbiomechanical characteristics of bone healing in alarge femoral segmental defect reconstructed witha titanium cage in a canine model.

METHODS AND MATERIALS

Implants

A standard-size commercially available oval cylindricalmesh titanium cage (DePuyMotech, Johnson & Johnson;Warsaw, IN) with 22� 17-mm diameters and 50-mmlength was selected. The geometry of the oval cage wasmodified into a circular configuration with a diameter of20 mm prior to placement in all reconstructed limbs(Fig. 1A). A commercially available human titaniumhumeral intramedullary nail (DePuy Ace, Johnson &Johnson) with a nail length of 320 mm and diameter of7 mm was also selected. The nail was specificallymodified for this study to a length of 140 mm and asingle distal interlocking hole was created. Standardproximal and distal titanium interlocking screws werealso included.

Bone Graft

The bone graft composite used in each specimenconsisted of morselized commercially available fresh-frozen canine cancellous bone croutons (VeterinaryTransplant Services, Inc., Seattle, WA) and a canineDBM (Dynagraft; GenSci Regeneration, Ontario,Canada). Equal portions of cancellous bone allograft toDBM composite was used in the same volume of theallograft for each specimen.

Animals and Experimental Design

Fully quarantined and immunized adult dogs (age 2–3 years, body weight 22–26 kg) were used in this studyfollowing approval by the Institutional Review Board atBaylor College of Medicine (Houston, TX). Biplanarradiography was performed on all animals to excludefemur pathology and ensure adequate femur size andconfiguration (mean canine femur diameter was16.6 mm (range 15.2–18.0 mm). The basic experimentalmodel consisted of a unilateral cortical mid-diaphysealsegmental femoral defect (3 cm in length) created usingan oscillating saw. Defect reconstruction was providedby the standard volume of the allograft bone compositeplaced in the defect with or without containment by acylindrical titanium meshed cage.

RECONSTRUCTION OF A CRITICAL-SIZE CANINE SEGMENTAL FEMORAL DIAPHYSEAL DEFECT 1439

DOI 10.1002/jor JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006

Twenty-one dogs were distributed into four experi-mental groups: six dogs with a 3 cm-long unilateralsegmental femur defect reconstructed with the allograft-cage construct were euthanized at 6weeks (GroupA); sixdogs with a defect reconstructed with the allograft-cagewere euthanized at 12 weeks (Group B); six dogs with adefect reconstructed with the allograft-cage were eutha-nized at 18 weeks (Group C); and three dogs with a bonedefect reconstructed with allograft but without a cagewere euthanized at 18 weeks postsurgery (Group D). Inall groups the operated femurs were additionally stabi-lized with a statically locked titanium intramedullarynail.

Surgical Procedure

Prior to surgery, the animals received a dose ofamoxicillin (10 mg/kg) intramuscularly for infectionprophylaxis. General anesthesia was induced withpentobarbital sodium (30 mg/kg) and 1–4% halothane,and maintained with 0.5–2% halothane. The animalswere placed in a lateral decubitus position and theexperimental limb was prepped and draped in a sterilefashion.

The defect was created through mid-lateral skinincision that was made along the thigh and soft tissuedissection performed using a cutting cautery. The lateralextensor muscles were split in line with their musclefibers, and subperiosteal exposure of themid-diaphysis ofthe femur was achieved. A 3 cm-long segmental defectwas created in the mid-diaphysis of the femur using anoscillating saw under continuous saline irrigation. Thestandard size cylindrical mesh titanium cage wastrimmed to a length of 40 mm and packed with thestandard volume of allograft bone composite (Fig. 1B). Aguide wire was concentrically placed through the longaxis the packed cage, and an 8-mm reamer passed overthe guide wire to create a central channel through thegraft to facilitate the placement of the intramedullarynail.

The nail was placed through a mid-lateral incisionthat was made over the proximal end of the femurbeginning at the tip of the greater trochanter andextended proximally. The medullary canal was enteredthrough the tip of the greater trochanter using an awl,and a guide wire was placed through the proximalfemoral fragment to the level of themid-diaphysis defect.In experimental Groups A, B, and C, the prepared

Figure 1. The commercially available oval cylindrical titaniummeshcage (A)hasbeenmodified into a circular configuration, trimmed to a length of 40 mm, and packed with astandard volume of morselized commercially available fresh-frozen canine cancellousallograft in combination with canine demineralized bone matrix (B).

1440 LINDSEY ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006 DOI 10.1002/jor

titanium cage was then fitted into the defect with theends of the cage overlapping the cortex at both ends ofthe femur by 5 mm. The guide wire was passed throughthe proximal femurmedullary canal, through the centralchannel in the cage, and into the distal femoralmedullary canal. The portion of the standard volume ofallograft bone composite that had been dislodged fromthe reamed cage was collected and placed about theperiphery of the cage. In an attempt to contain the graftat the defect site, the lateral thigh wound was carefullyclosed with absorbable sutures for the deep fascia, andnylon sutures for the skin. Themedullary canal was thensequentially reamed proximally and distally through thecageup toadiameter of 8mm.Thestandard size titaniumnail was inserted into the proximal femur, through thecage, into the distal femur, and statically locked withproximal anddistal screws. In experimentalGroupD, thesame 3 cm-long defect was also created in the mid-diaphysis of the femur as previously noted. The proximaland distal femur fragments were reamed and thenstabilized with the standard size, statically lockedtitanium intramedullary nail without a cage. Thestabilized defect was then packed with the standardvolume of the allograft bone composite to decrease thelikelihood of graft dislodgement.

Postoperatively biplanar radiography was obtainedfor all operated femurs to ensure adequate implantplacement and alignment. The postoperative analgesiain all animals consisted of butorfanal (0.5mg/kg, 3 tid) for3 days for pain management. The animals werepermitted to freely weight bear immediately followingsurgery, and allowed to eat ad lib.

Euthanasia and Radioactive Labeling

The animals were injected with 99-meta-technitiummethylenediphosphonate (Tc99m MDP) with a radio-activity of 4 mCi and 6 h prior to euthanasia (a lethaldose of buthansia 2 mL/10 kg). Both femurs were thenharvested and stripped of their soft tissues. The radio-active Tc99m MDP injection was performed to quantita-tively and qualitatively assess the biologic activity of thespecimens with Single Photon Emission ComputedTomography (SPECT). Radioactivity counts were calcu-lated from biplanar (anterior–posterior and oblique)quantitative measurements. The defect was divided intothree equal sections along the femur’s longitudinal axis(proximal, middle, and distal), and mean counts �SDwere calculated. The calculated radioactivity counts ofthe proximal, middle, and distal sections of the defectwere calculated and compared to the correspondingregions of the contralateral intact femurs. Subse-quently, the specimens were stored frozen at �208Cpending further evaluation.

Gross Inspection

Postmortem, all specimens in Groups A, B, C, and Dwere visually inspected to determine the presence anddistribution of bone formation at the defect. Addition-

ally, the alignment of the defect femur was comparedwith the contralateral intact femur. All femurs werepalpated and subjected to manual nondestructiveassessment of torsion and bending stability.

Plain Radiography

Biplanar plain radiography was obtained of all har-vested femurs postmortem. A high-resolution (AgfaGevaert AG; Leverkusen, Germany) radiographic ana-lysis was performed using a self-contained faxitron(Faxitron 804; Wheeling, IL) (operating voltage 60 kV,exposure time 3 min).

Computed Tomography

All harvested femurs were scanned using computedtomography (CT). Scans were performed at 3-mmintervals with 1.5 mm-thick slices. Image processingand three-dimensional (3D) reconstructions wereaccomplished using the Vitrea Software Package (VitalImages, Inc; Plymouth, MN).

Mechanical Testing

Four cage-reconstructed paired canine femurs wererandomly selected from Groups A, B, and C forbiomechanical analysis. Two paired femurs from theGroup D specimens reconstructed without a cage wereselected for mechanical testing. A power analysis wasperformed to determine the number of specimens tostudy from each group. The defect specimens werethawed and the distal interlocking screws removed. Theproximal and distal ends of the femurs were potted incement (Smooth Cast 300; Smooth-On, Inc; Easton, PA)perpendicular to their long axis. The specimens werealigned in two perpendicular planes to a plumb lineoriented to the center of rotation of the testing system(MTS 858MiniBionix; MTS Systems Corp; Eden Prairie,MN). The axis of rotation was along the long axis of theintramedullary nail. The specimens were tested inrotation by applying axial torsion to failure at a rate of208/min maintaining an axial compression load of 10 N.Stiffness and load-to-failure strength were calculatedand compared. Stiffness was calculated from the bestlinear portion of the load–deflection curve, and load-to-failure was defined as the ultimate torque.

Histological Evaluation

Two cage-reconstructed canine femurs were randomlyselected from Groups A, B, C, and D for histologicalanalysis. A 6 cm-long bone segment (the 3-cm con-structed defect with 1.5-cm bone margins) was resected,and the orthopedic hardware (cage and intramedullarynail) retained. The resected segments were fixed with4% buffered formalin, dehydrated in ethanol withsequential concentrations, and embedded in polymethl-metacrylate (PMMA). Three sections were obtainedfrom each specimen using a low-speed oscillating sawunder constant irrigation (Leitz model 1600; Germany).

RECONSTRUCTION OF A CRITICAL-SIZE CANINE SEGMENTAL FEMORAL DIAPHYSEAL DEFECT 1441

DOI 10.1002/jor JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006

Coronal sections were obtained through the middle ofthe defect at both, the proximal and distal bone–defectjunctions; an axial section was obtained through themiddle of the defect. The sections were analyzed usingmacroradiography and a bright-field light microscopy.

Statistical Methodology

The quantitative results were presented as averagevalues� standard deviations. This data were analyzed,reduced to the controls, and crosscompared to determinewithin and between the experimental groups. Thecorrelation tests between data obtained from varioustesting modalities was also done. The experimentalgroups A, B, and C and the control group D werecompared using analysis of variance (ANOVA) with apost hoc Bonferroni test. Statistically significant differ-ences were defined as p¼ 0.05.

RESULTS

Surgical Follow-up

The 21 dogs in the study tolerated the surgicalprocedures well, and fully recovered withoutinfections, wound dehiscence, fractures about thefixation, or the need for additional surgical pro-cedures. All dogs were immediately able to fullyweightbear on the defect limb.

Gross Inspection

In Groups A, B, and C, there was visible boneabout most of the external circumference of thecage bridging the entire defect and integrated withthe proximal and distal host bone–cage junctions.The new bone formed at the defect site wasapproximately twice the diameter of the contral-ateral intact femur. In the sparse areas where thecage fenestrations were visible, bone could bevisualized within the cage. Conversely, in GroupD, all specimens demonstrated fibrous tissuewithin the defect. Although sparse bone formationwas present at the host bone–defect junctions, itdid not exceed the diameter of the contralateralintact femur, nor did it bridge the defect. Further-more, the distal interlocking screw site demon-strated exuberant callus formation compared tothe corresponding region in the contralateralintact femur.

Plain Radiography

In Groups A, B, and C, the cage-nail constructremained stable in all femurs without evidence ofloosening at the bone–cage interface, hardwarefailure, or cortical hypertrophy at the interlockingscrew sites. Marked progression of new boneformation was present about the cage bridging

the defect and at the host bone–defect junctions.New bone, however, could not be accuratelydetected within the cage with plain radiography(Fig. 2). In Group D, anteroposterior and lateralradiographs revealed sparse new bone formationat the host bone–defect junctions and a prominentlucency in the center of the defect without evi-dence of bony bridging at 18 weeks. Distal inter-locking screw failure was present in all specimens,and these femurs demonstrated focal corticalhypertrophy at the site (Fig. 3).

CT

The Group A, B, and C femurs demonstratedmarked progression of new bone formation withinand around the cage throughout the defect andat the host bone–defect junctions. This well-mineralized new bone exhibited a homogeneouspattern (Fig. 4). In the Group D specimens, boneformation was limited to sparse fragments withoutbridging of the defect (Fig. 5).

SPECT

SPECT evaluation of the specimens from GroupsA, B, and C demonstrated considerable greateruptake of Tc99m MDP in the defect femurscompared to their matched contralateral intactcounterparts (Fig. 6). At 6 weeks, the uptake of theradioactive Tc isotope was uniformly distributedacross the defect exhibiting a level 40 timesgreater than the contralateral femur. At 12 weeks,the Tc uptake decreased to approximately 15 timesthat of the contralateral femur, and its distri-bution was slightly greater in the middle of thedefect than at both host bone–defect junctions. By18 weeks, the Tc uptake at both proximal anddistal thirds of the defect was comparable to that ofthe contralateral intact femur; while the mid-thirdof the defect remained ‘‘hot.’’ In contrast, Group Dconsistently demonstrated minimal radioactivitythroughout the entire defect at 18 weeks. How-ever, all Group D femurs demonstrated increasedTc99m MDP uptake at the site of the distalinterlocking screw (Fig. 7).

Biomechanical Analysis

In the Groups A, B, and C femurs, the meantorsional stiffness at 6, 12, and 18 weeks was 44.4,45.7, and 72.5% of the contralateral intact femur,respectively. The mean torsional strength was51.0, 73.6, and 83.4% of the contralateral intactfemur at 6, 12, and 18 weeks, respectively (Fig. 8).All specimens in these groups failed at theproximal host bone–cage interface. Although

1442 LINDSEY ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006 DOI 10.1002/jor

there was a trend toward increasing rotationalstiffness and strength over time in Groups A, B,and C, these differences were not statisticallysignificant. In the Group D femurs, the meantorsional stiffness and strength at 18 weeks were17.5 and 39.2% of contralateral control femurs,respectively. All specimens in this group consis-tently failed through the middle of the defect.A statistically significant difference was demon-strated in the rotational stiffness and strengthbetween cage (Group C) and noncage (Group D)specimens at 18 weeks (p¼ 0.046 and p¼ 0.032,respectively).

Histological Analysis

In Groups A, B, and C, bone formation spanned thedefect correlating with the time of healing. InGroup A, the proximal and distal longitudinalsections (Fig. 9A and C) demonstrated evidence ofearly new bone formation both within and outsidethe cage, which extended onto the margins ofthe host cortical bone. Likewise, the transversesections (Fig. 9B) demonstrated new bone forma-tion across the middle of the defect without directapposition to the metal framework of the cage. InGroup B, the proximal and distal longitudinalsections (Fig. 10A and C) as well as the transverse

Figure 2. Anteroposterior (A) and lateral (B) plainX-rays of harvested, soft tissue-freecage reconstructed canine femur at 18 weeks. New bone formation can be visualizedabout the cage but not accurately detected within the cage with plain radiography.

RECONSTRUCTION OF A CRITICAL-SIZE CANINE SEGMENTAL FEMORAL DIAPHYSEAL DEFECT 1443

DOI 10.1002/jor JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006

sections (Fig. 10B) demonstrated the progressionof new bone formation. At the cage–host bonejunctions, this new bone continued to form bothinside and outside the cage in a manner that fullyincorporated the cage with the host cortical bone.Within the defect, there was further evidence ofthe uniform integration of the bone formed outsideand inside the cage through its fenestrationswithout directly contacting the cage. In Group C,the proximal and distal longitudinal sections(Fig. 11A and C) as well as the transverse sections(Fig. 11B) progressively demonstrated extensive

areas of new bone formation spanning the defect.Less porous, maturing new bone was evidentoutside and inside the cage seamlessly integratingthrough its fenestrations without intervening softtissue. Compared to Groups A and B, Group Cspecimens demonstrated a progressive decrease inthe lucency at the interface between the newlyformed bone and the metallic framework of thecage. In Group D, all proximal and distal long-itudinal as well as transverse sections (Fig. 12A–C) demonstrated a paucity of bone within thedefect. Instead, there were sparse isolated regions

Figure 3. Anteroposterior (A) and lateral (B) plain X-rays of a harvested soft tissuefree canine femur reconstructed without a cage. A prominent lucency is clearly visible inthe middle of the defect without bony bridging at 18 weeks. At the distal interlockingscrew site, where the screw has failed there is focal cortical hypertrophy.

1444 LINDSEY ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006 DOI 10.1002/jor

Figure

4.

A3D

reconstru

cted

CT

at18wee

ksdem

onstratesnew

bon

eform

ation

,which

completely

envelop

sthe

cage

(A).

Coron

al(B

),sa

gittal(C

),and

axial(D

)reconstru

ctionsthrough

themiddle

ofthedefectdep

ictuniform

distribution

ofbon

ewithin

thecage.

Anaxialsectionthroughthemiddle

ofthecage(B

)of

mineralizedbon

ewhichis

hom

ogen

eouslydistributed.

RECONSTRUCTION OF A CRITICAL-SIZE CANINE SEGMENTAL FEMORAL DIAPHYSEAL DEFECT 1445

DOI 10.1002/jor JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006

of nonhomogeneous trabecular bone primarily atthe defect–host bone junctions, suggesting thateither the previously placed graft had beendislodged and/or had been resorbed. None of thespecimens demonstrated bone bridging the defectwith new bone visualized only at the host femurmargins, while the middle region of the defectconsisted predominantly of fibrous tissue.

DISCUSSION

A segmental defect in long bone is considered ofcritical size if it will not heal without surgicalaugmentation. The size which renders a defect‘‘critical’’ is not well understood. A critical defecthas been defined as segmental bone loss of a lengththat exceeds 2.5 times the diameter of the affectedbone. In the present study, the size of the defectwas approximately two times the mean diameterof the diaphysis in the canine femurs. Althoughthis was slightly less than the proposed size to be

‘‘critical,’’ it was clearly of sufficient size to inhibithealing even when augmented with bone graft andstabilized with an intramedullary nail. Therefore,a critical defect in long bone cannot simply bedefined by its size, but may also be dependent onthe species phylogenetic scale, the anatomic loca-tion, the associated soft tissue envelope, and theloads on the affected limb. Furthermore, the host’sage, metabolic and systemic conditions, andrelated morbidities would also affect the defect’shealing potential.

A variety of treatment modalities have beenadvocated for large long bone segmental defect and,to date, no single treatment method is consideredthe standard. This is, in part, due to the diverseetiologic conditions that result in segmental bonedeficiency, and the unique clinical considerationsthat are specific to each. Bony resection secondaryto osteomyelitis or neoplasm requires healthyretained bone margins and, it is not possible tocompromise on the length of bone to be resected.

Figure 5. Biplanar surface (A, B) and axial (C) 3D CT reconstructions of a femurtreated without a cage demonstrates sparse bone formation without bridging of thedefect. Axial (D) reconstruction at the distal interlocking screw site demonstrates focalbony hypertrophy and screw failure.

1446 LINDSEY ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006 DOI 10.1002/jor

Deformity correction requires that the reconstruc-tion technique be stable enough to permit the earlyactive mobilization of adjacent joints. The treat-ment of segmental bone loss due to high energytraumamandates that reconstructive technique beversatile, and applicable in the acute setting with-out inflicting additional trauma to the host tissues.All currently recommended treatment options forlarge longbonedefects donot reliably address theseissues in a comprehensive manner.

The application of a cylindrical titanium meshcage for the treatment of critical size long-bonesegmental defects was first reported by theauthors.32 In clinical cases involving large tibialdiaphyseal defects, the authors were able todemonstrate defect healing clinically and radio-graphically while achieving excellent functionaloutcomes.33 The technique consisted of a singlestage procedure that permitted the early recoveryof adjacent joint mobility, and immediate axialloading of the extremity. Compared with alterna-tive modalities, the cage technique was simpleand cost-effective. Ostermann et al.34 applied thissame technique in a Grade IIIB proximal tibiametaphyseal-diaphyseal fracturewith similar clini-

cal and radiographic results. These authors alsonoted a significant decrease in the time of treat-ment compared to other techniques. Reynderset al.35 also used the cylindrical titanium meshcage technique to successfully treat an 11 cm-longsegmental defect in the femur. They reportedbridging bone formation about the periphery ofthe cage, but could not document the presence ofosseous continuitywithin the cage. They concludedthis technique to be as a salvage procedure whenexisting alternatives were not feasible, and ampu-tation was imminent. These clinical reports docu-ment the potential merits of cylindrical cagereconstruction but alone do not accurately char-acterize the extent andnature of the defect healing.

Our basic segmental defect study design wasselected to limit the bone-healing variables tosimply the presence or absence of the cage.Although this model would not accurately repre-sent themultifactorial demands of an open fractureassociated with bone loss, it could reflect theclinical situation presented by those patients whorequire surgical bone resection for tumor, infection,revision joint arthroplasty, or limb lengthen-ing. Like all animal studies, our model has its

Figure 6. The graph depicts the spatial distribution of Tc99m MDP isotope uptakethroughout the defect relative to the corresponding region of the contralateral intactfemur.

RECONSTRUCTION OF A CRITICAL-SIZE CANINE SEGMENTAL FEMORAL DIAPHYSEAL DEFECT 1447

DOI 10.1002/jor JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006

limitations, andamong them include themanner inwhich the defect was created, variations betweenhuman and canine femur anatomy and bloodsupply, the modified nature of the intramedullaryimplant, and the inherent differences between two-and four-leg gait. However, the surgical techniquesemployed (with andwithout the cage), and the boneallograft composite placed in the defect are bothcurrently used in daily clinical practice.

Hypothetically, the cylindrical titanium meshcage technique exhibits a number of advantageousbiologic andbiomechanical properties that promotelong-bone defect reconstitution. The principalbiologic advantages of the cage technique are thebiocompatibility of material (i.e., titanium), itsfenestrated design, and its ability to contain bonegraft. Thehollowand fenestratednature of the cageallows for the bulk of the construct to consist ofcancellous bone graft, ensures retention of the graftat the defect, and provides a favorable scaffold forbony restoration. Biomechanically, the cage tech-

nique allows for the immediate restoration of limbalignment and stability, which permits earlyweight bearing of the limb. This results in theactive loading of the bone graft, which appears toenhance the healing process.

This study was designed to not only determinethe merits of the titanium mesh cylindrical cage/bone graft reconstruction technique to repaircritical size long bone defects in vivo, but tospecifically employ implants and instruments thatare currently FDA approved and clinically avail-able. Likewise, a canine model was selected due toits more clinically relevant biology, the availabilityof canineDBMand canine allograft, a limb size andconfirmation that would accommodate existingorthopedic instruments and implants, and theability of the canine to resume full weightbearingpostoperatively.

In all Group D canine defects reconstructedwithout a cage, all specimens failed to heal at18 weeks by all of the assessment modalities

Figure 7. Tc99m MDP isotope uptake at the defect in cage reconstructed femurs at6weeks (A), 12weeks (B), and18weeks (C).Note the limited isotopeuptakeat thedefect at18 weeks in the femur reconstructed without the cage (D). In this specimen markeduptake was evident at the site of the distal interlocking screw focal bony hypertrophy.

1448 LINDSEY ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006 DOI 10.1002/jor

employed. At the defect,manual inspection demon-strated gross rotational motion, X-rays revealed aprominent lucency, CT confirmed the absence ofbone reconstitution, and SPECT was negative fortechnetium isotope uptake. Furthermore, biome-chanical testing corroborated the lack of torsionalstability, and histological assessment demon-strated bone graft resorption with isolated regionsof nonhomogenous trabecular bone fragments.

In contrast, in the Groups A, B, and C experi-mental cage specimens all defects demonstratedprogressive bone healing of the defect at 6, 12,and18weeks, respectively. All specimens exhibitedrigidity with nondestructive manual bending ortorsion; confluent new bone formation was visibleand palpable circumferentially about the cagespanning between the proximal and distal femurfragments. Plain radiography revealed the pro-

gression of new bone formation about the cage andat the cage–host bone junctions. The metallicnature of the cage obstructed the accurate visua-lization of bone within the implant; however, CT ofthe experimental femurs confirmed the presence ofbone inside and around the cage, to include thecage–host bone junctions. CT was particularlyuseful in determining the presence, location, andquality of the bone about the cage. The quality ofthe CT image was not distorted by the metalliccage/nail construct.

SPECT analysis was important in establishingthe pattern of bone graft reconstitution throughoutthe defect. In Group A (6 weeks) specimens, thebiologic activity at the defect was significantly (40times) greater than that of the contralateral intactfemur, and extended diffusely across the defect toinclude both the proximal and distal cage–hostbone junctions. This patternof activity couldalsobeappreciated in Group B (12 weeks) specimens,although its intensity had decreased approxi-mately 60% (15 times that of the contralateralintact control). In Group C (18 weeks) specimens,the biologic activity at both cage–host bone junc-tions further subsided, while the center of thedefect remained relatively ‘‘hot.’’ These findingssuggest that bone graft reconstitution occurredthroughout the entire defect with the maturationoccurring at the cage–host bone junctions prior tothe center of the defect. These data support thehypothesis that the cage promotes confluent bonehealing across the entire defect, and a longerfollow-up would be necessary to determine thetime to complete bone maturation in the defect.

Biomechanically, the femurs reconstructedwithout a cage were unstable at 18 weeks. Thetorsional stiffness and strength of these specimenswere considerably less than that of even the 6-weekcage reconstructed femurs (17.5 and 39.2% vs. 44.4and 51.0%, respectively). Although extending thetime of healing may increase the mechanicalproperties of the femurs without the cage, it isunlikely that would be clinically relevant. Con-versely, biomechanical analysis of the cage femursdemonstrated a marked progression in the defect’sstability up to 18 weeks (72.5% stiffness and 83.4%strength compared to the contralateral intact side).This could be attributed to the bone formationwithin the defect, its remodeling, and its inter-digitationwith the cage and host bone. Therewas alinear relationship between the improved mechan-ical properties of the cage augmented defects andthe time of healing. Although the torsional stiffnessand strength of the cage femurs did not achievesymmetry with the contralateral intact femur at

Figure 8. The graph depicting the stiffness (A) andstrength (B) of the defect specimens treated with andwithout a cage. There was a statistically significantdifference in rotational stiffness and strength betweenthe cage (GroupC) and theno-cage (GroupD) at 18weeks(p¼ 0.046 and p¼ 0.032, respectively).

RECONSTRUCTION OF A CRITICAL-SIZE CANINE SEGMENTAL FEMORAL DIAPHYSEAL DEFECT 1449

DOI 10.1002/jor JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006

18 weeks, the data suggest that a minor extensionin the time of implantation could fully restore limbstability. The effect of the implants (cage and nail)on the defect stability prior to healing wasbiomechanically tested in a separate ex vivoexperiment, and the results were negligible. Thissuggests that the stability of the in vivo healeddefect was primarily the result of bone formationand thepresence of the cage facilitated this process.

Histological analysis of the cage specimenssuggests that there was a progressive increase inbone mineral density in all defect regions studiedover time. These specimens demonstrated a con-sistent pattern of bone formation at both cage–hostbone junctions with subsequent progressiontowards the middle of the defect. It appears thatthe early bone formation at the cage–host bonejunctions provides the initial biomechanical

Figure 9. At 6 weeks postcage implantation, the proximal (A) and distal (C)longitudinal sections demonstrate new bone formation both within and outside the cagethat extend into the host cortical bone margins. A transverse section through the middleof the defect (B) also depicts new bone formation inside and outside the cage, withoutdirect apposition to its metal frame work.

Figure 10. At 12 weeks postcage implantation, the proximal (A) and distal (C)longitudinal sections depicted new bone formation both inside and outside the cage,which is contiguous with the host control bone. Transverse sections (B) also depicteduniform bone formation outside and inside the cage through its fenestration withoutdirect opposition to the cage.

1450 LINDSEY ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006 DOI 10.1002/jor

integrity of the limb; consequently, the completerestitution of bone continuity across the defectmaynot be as crucial for restoring limb stability. Incontrast, defects reconstructedwithout a cagemustcompletely restore bone continuity to becomestable. Additionally, new bone was formed outsideand inside the cage, and was most pronounced inthe circumferential regions about the cage, whichclosely approximated the limb musculature. This

supports the placement of bone graft in bothlocations. The new bone, outside and inside thecage, communicated through the implant’s fenes-trations, and this further augmented the defect’sstability. Finally, the lucency between the newbone formedand themetallic framework of the cageprogressively decreased over time further, suggest-ing that the cage provided a suitable scaffold topromote bone healing across the defect. These

Figure 11. At 18 weeks postcage implantation, the proximal (A) and distal (C)longitudinal sections and the transverse sections (B) depicted progressive new boneformation throughout the defect. This bone, which was evident both inside and outsidethe cage integrated through its fenestrations.The seambetween themetal and thisnewlyformed bone appeared to decrease as healing progressed.

Figure 12. At 18 weeks without cage implantation, the proximal (A), and distal (C)longitudinal as well as the transverse (B) sections demonstrated sparse regions ofnonhomogeneous trabecular bone primarily at the defect postbone junctions. There wasno evidence bone bridging the defect, and the middle of the defect consisted ofpredominantly fibrous tissue.

RECONSTRUCTION OF A CRITICAL-SIZE CANINE SEGMENTAL FEMORAL DIAPHYSEAL DEFECT 1451

DOI 10.1002/jor JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006

biological observations culminated in a defecthealing process that seamlessly integrated thecage with the femur and resulted in excellent limbstability.

CONCLUSIONS

A cylindrical titanium mesh cage in combinationwith cancellous allograft, demineralized bonematrix and an intramedullary nail successfullyrestores bone continuity and augments limbstability in a canine femur diaphyseal segmentaldefect model. Conversely, the same constructwithout a cage does not heal the defect. In all cagespecimens, new bone formation was evident at thedefect and progressed throughout the time ofimplantation. New bone seamlessly integrated toand through the fenestrations of the cage at itshost bone junctions as well as in the middle of thedefect. The cage appeared to promote allograftcomposite healing throughout the defect andincrease biomechanical stability of the caninefemur. Cylindrical titanium mesh implants maybe a viable alternative for the treatment ofsegmental long bone defects.

ACKNOWLEDGMENTS

The study was funded by a grant from DePuy Motech(Johnson & Johnson, Warsaw, IN).

REFERENCES

1. Johnson KD, August A, Sciadini MF, et al. 1996. Evalua-tion of ground cortical autograft as a bone graft material ina new canine bilateral segmental long bone defect model.J Orthop Trauma 10:28–36.

2. DeCoster TA, Gehlert RJ, Mikola EA, et al. 2004. Manage-ment of posttraumatic segmental bone defects. J Am AcadOrthop Surg 12:28–38.

3. Lindsey RW, Probe R, Miclau T, et al. 1993. The effects ofantibiotic-impregnated autogenic cancellous bone graft onbone healing. Clin Orthop 291:303–312.

4. Stevenson S, Li XQ, Martin B. 1991. The fate of cancellousand cortical bone after transplantation of fresh and frozentissue-antigen-matched and mismatched osteochondralallografts in dogs. J Bone Joint Surg 73A:1143–1156.

5. Chmell MJ, McAndrew MP, Thomas R, et al. 1995.Structural allografts for reconstruction of lower extremityopen fractures with 10 centimeters or more of acutesegmental defects. J Orthop Trauma 9:222–226.

6. Banic A, Hertel R. 1993. Double vascularized fibulas forreconstruction of large tibial defects. J Reconstr Microsurg9:421–428.

7. Srouji S, Blumenfeld I, Rachmiel A, et al. 2004. Bone defectrepair in rat tibia by TGF-beta1 and IGF-1 released fromhydrogel scaffold. Cell Tissue Bank 5:223–230.

8. Rasmussen MR, Bishop AT, Wood MB. 1995. Arthrodesisof the knee with a vascularized fibular rotatory graft.J Bone Joint Surg 77A:751–759.

9. Malizos KN, Zalavras CG, Soucacos PN, et al. 2004. Freevascularized fibular grafts for reconstruction of skeletaldefects. J Am Acad Orthop Surg 12:360–369.

10. Cierny G, Zorn KE. 1994. Segmental tibial defects.Comparing conventional and Ilizarov methodologies. ClinOrthop 301:118–123.

11. Green SA. 1994. Skeletal defects. A comparison of bonegrafting and bone transport for segmental skeletal defects.Clin Orthop 301:111–117.

12. Ilizarov GA. 1989. The tension-stress effect on the genesisand growth of tissues. Part I. The influence of stability offixation and soft-tissue preservation. Clin Orthop 238:249–281.

13. Ilizarov GA. 1989. The tension-stress effect on the genesisand growth of tissues: Part II. The influence of the rate andfrequency of distraction. Clin Orthop 239:263–285.

14. Gao TJ, Lindholm TS, Kommonen B, et al. 1997. The use ofa coral composite implant containing bone morphogeneticprotein to repair a segmental tibial defect in sheep. IntOrthop 21:194–200.

15. Ito M, Abumi K, Moridaira H, et al. 2005. Iliac crestreconstruction with a bioactive ceramic spacer. Eur SpineJ 14:99–102.

16. Jansen J, Ooms E, Verdonschot N, et al. 2005. Injectablecalcium phosphate cement for bone repair and implantfixation. Orthop Clin North Am 36:89–95.

17. Fialkov JA, Holy CE, Shoichet MS, et al. 2003. In vivo boneengineering in a rabbit femur. J Craniofac Surg 14:324–332.

18. Gugala Z, Gogolewski S. 2002. Healing of critical-sizesegmental bone defects in the sheep tibiae using bioresorb-able polylactide membranes. Injury 33(Suppl 2):B71–B76.

19. Meinig RP, Rahn B, Perren SM, et al. 1996. Boneregeneration with resorbable polymeric membranes: treat-ment of diaphyseal bone defects in the rabbit radius withpoly(L-lactide) membrane. A pilot study. J Orthop Trauma10:178–190.

20. Wang MY, Levi AD, Shah S, et al. 2002. Polylactic acidmesh reconstruction of the anterior iliac crest after boneharvesting reduces early postoperative pain after anteriorcervical fusion surgery. Neurosurgery 51:413–416.

21. Chen D, Zhao M, Mundy GR. 2004. Bone morphogeneticproteins. Growth Factors 22:233–241.

22. Cornell CN. 2004. Osteobiologics. Bull Hosp Joint Dis 62:13–17.

23. Einhorn TA, Trippel SB. 1997. Growth factor treatment offractures. Instr Course Lect 46:483–486.

24. Etienne G, Ragland PS, Mont MA. 2004. Use of cancellousbone chips and demineralized bone matrix in the treat-ment of acetabular osteolysis: preliminary 2-year follow-up. Orthopedics 27(1 Suppl):s123–s126.

25. Gao J, Knaack D, Goldberg VM, et al. 2004. Osteochondraldefect repair by demineralized cortical bone matrix. ClinOrthop 427(Suppl):S62–S66.

26. Issack PS, DiCesare PE. 2003. Recent advances toward theclinical application of bone morphogenetic proteins in boneand cartilage repair. Am J Orthop 32:429–436.

27. Meyer U, Joos U, Wiesmann HP. 2004. Biological andbiophysical principles in extracorporal bone tissue engi-neering. Part I. Int J Oral Maxillofac Surg 33:325–332.

28. Zegzula HD, Buck DC, Brekke J, et al. 1997. Boneformation with use of rhBMP-2 (recombinant human bonemorphogenetic protein-2). J Bone Joint Surg 79:1778–1790.

1452 LINDSEY ET AL.

JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006 DOI 10.1002/jor

29. Rodeo SA, Maher SA, Hidaka C. 2004. What’s new inorthopaedic research. J Bone Joint Surg 86A:2085–2095.

30. Salgado AJ, Coutinho OP, Reis RL. 20904. Bone tissueengineering: state of the art and future trends. MacromolBiosci 4:743–765.

31. Vats A, Tolley NS, Buttery LD, et al. 2004. The stem cell inorthopaedic surgery. J Bone Joint Surg 86B:159–164.

32. Cobos J, Lindsey RW, Gugala Z. 2000. The cylindricaltitanium mesh cage for the treatment of a long segmentalbone defect: description of a new technique and report oftwo cases. J Orthop Trauma 14:54–59.

33. Lindsey RW, Gugala Z. 2004. Cylindrical titanium meshcage for the reconstruction of long bone defects. OsteoTrauma Care 12:108–115.

34. Ostermann PAW, Hasse N, Rubberdt A, et al. 2002.Management of a long segmental defect at the proximalmeta-diaphyseal junction of the tibia using a cylindricaltitanium mesh cage. J Orthop Trauma 16:597–601.

35. Reynders P, Broos PLO, Stoffen D. 2003. The use ofcylindrical titanium mesh cages in the treatment of post-traumatic segmental bone loss of the femur. Osteo TraumaCare 11:99–104.

RECONSTRUCTION OF A CRITICAL-SIZE CANINE SEGMENTAL FEMORAL DIAPHYSEAL DEFECT 1453

DOI 10.1002/jor JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2006