9
A histomorphological study of tendon reconstruction to a hydroxyapatite-coated implant: regeneration of a neo-enthesis in vivo C.J. Pendegrass a, * , M.J. Oddy a , S.R. Cannon b , T. Briggs b , A.E. Goodship c , G.W. Blunn a a The Centre for Biomedical Engineering, Institute of Orthopaedics & Musculo-Skeletal Science, Brockley Hill, Stanmore, Middlesex HA7 4LP, UK b The Royal National Orthopaedic Hospital, Stanmore, Middlesex HA7 4LP, UK c The Royal Veterinary College, North Mimms AL9 7TA, UK Received 21 January 2004; accepted 24 March 2004 Abstract The attachment of tendons and ligaments to massive endoprostheses remains a clinical challenge due to the difficulty in achieving a soft tissue implant interface with a mechanical strength sufficient to transmit the forces necessary for locomotion. We have used an in vivo animal model to study patellar tendon attachment to an implant surface. The interface generated when the patellar tendon was attached to a hydroxyapatite (HA) coated implant was examined using light microscopy and a quantitative histomorphological analysis was performed. In the Autograft Group, the interface was augmented with autogenous cancellous bone and marrow graft, and at six weeks an indirect-like insertion was observed. At twelve weeks, the interface was observed to be a layered neo-enthesis, whose morphology was similar to a normal direct tendon insertion. In the HA Group, the tendon–implant interface was not augmented, and the implant was enveloped by a dense collagenous fibrous tissue. This study shows that a tendon–implant neo-enthesis can develop in situ by employing a suitable implant surface in association with biological augmentation. Ó 2004 Orthopaedic Research Society. Published by Elsevier Ltd. All rights reserved. Introduction Inadequate attachment of the patellar tendon fol- lowing proximal tibial replacement results in impaired extensor mechanism function at the knee [4,5,24]. There are many contemporary methods employed to recon- struct the soft tissue–implant interface [4,14,17,24, 29,33], but poor fixation can result in tendon shortening and detachment, patella alta, quadriceps atrophy and altered gait [5,24]. Clinical retrieval data is limited, hence the morphology of the resulting interface is poorly understood. Despite these problems, limb-salvage sur- gery with massive endoprosthetic replacement of bone defects remains the treatment of choice for certain tu- mours of the musculoskeletal system [19]. A normal epiphyseal tendon–bone interface is a direct enthesis uniting tissues with differing biomechanical properties. The junction can be shown to redevelop fol- lowing surgery to implant tendon into bone, although the early stages of union resemble an indirect enthesis [27,31]. Experimental observations have suggested that the free end of the tendon undergoes some initial degenerative changes [39,42]. Permanent integration oc- curs as a result of progressive mineralization from the walls of the bone tunnel into a collar of soft tissue callus surrounding the tendon [15,26,38]. Regeneration of a layered enthesis occurs after an intermediate phase, during which the collagen fibres of woven bone are an- chored to the fibre bundles of the scar tissue surrounding the tendon [35]. Functional testing of a neo-enthesis to bone has shown that by twelve weeks, sufficient interface strength can be recovered, such that failure during pull-out studies oc- curred at another site [35]. These findings are comparable * Corresponding author. Tel.: +44-020-8954-0268; fax: +44- 0208420-7392. E-mail address: [email protected] (C.J. Pendegrass). 0736-0266/$ - see front matter Ó 2004 Orthopaedic Research Society. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.orthres.2004.03.022 www.elsevier.com/locate/orthres Journal of Orthopaedic Research 22 (2004) 1316–1324

A histomorphological study of tendon reconstruction to a hydroxyapatite-coated implant: regeneration of a neo-enthesis in vivo

  • Upload
    c

  • View
    214

  • Download
    2

Embed Size (px)

Citation preview

www.elsevier.com/locate/orthres

Journal of Orthopaedic Research 22 (2004) 1316–1324

A histomorphological study of tendon reconstruction toa hydroxyapatite-coated implant: regeneration of

a neo-enthesis in vivo

C.J. Pendegrass a,*, M.J. Oddy a, S.R. Cannon b, T. Briggs b,A.E. Goodship c, G.W. Blunn a

a The Centre for Biomedical Engineering, Institute of Orthopaedics & Musculo-Skeletal Science,

Brockley Hill, Stanmore, Middlesex HA7 4LP, UKb The Royal National Orthopaedic Hospital, Stanmore, Middlesex HA7 4LP, UK

c The Royal Veterinary College, North Mimms AL9 7TA, UK

Received 21 January 2004; accepted 24 March 2004

Abstract

The attachment of tendons and ligaments to massive endoprostheses remains a clinical challenge due to the difficulty in achieving

a soft tissue implant interface with a mechanical strength sufficient to transmit the forces necessary for locomotion. We have used an

in vivo animal model to study patellar tendon attachment to an implant surface. The interface generated when the patellar tendon

was attached to a hydroxyapatite (HA) coated implant was examined using light microscopy and a quantitative histomorphological

analysis was performed. In the Autograft Group, the interface was augmented with autogenous cancellous bone and marrow graft,

and at six weeks an indirect-like insertion was observed. At twelve weeks, the interface was observed to be a layered neo-enthesis,

whose morphology was similar to a normal direct tendon insertion. In the HA Group, the tendon–implant interface was not

augmented, and the implant was enveloped by a dense collagenous fibrous tissue.

This study shows that a tendon–implant neo-enthesis can develop in situ by employing a suitable implant surface in association

with biological augmentation.

� 2004 Orthopaedic Research Society. Published by Elsevier Ltd. All rights reserved.

Introduction

Inadequate attachment of the patellar tendon fol-

lowing proximal tibial replacement results in impaired

extensor mechanism function at the knee [4,5,24]. There

are many contemporary methods employed to recon-

struct the soft tissue–implant interface [4,14,17,24,

29,33], but poor fixation can result in tendon shortening

and detachment, patella alta, quadriceps atrophy andaltered gait [5,24]. Clinical retrieval data is limited,

hence the morphology of the resulting interface is poorly

understood. Despite these problems, limb-salvage sur-

gery with massive endoprosthetic replacement of bone

defects remains the treatment of choice for certain tu-

mours of the musculoskeletal system [19].

* Corresponding author. Tel.: +44-020-8954-0268; fax: +44-

0208420-7392.

E-mail address: [email protected] (C.J. Pendegrass).

0736-0266/$ - see front matter � 2004 Orthopaedic Research Society. Publis

doi:10.1016/j.orthres.2004.03.022

A normal epiphyseal tendon–bone interface is a directenthesis uniting tissues with differing biomechanical

properties. The junction can be shown to redevelop fol-

lowing surgery to implant tendon into bone, although

the early stages of union resemble an indirect enthesis

[27,31]. Experimental observations have suggested that

the free end of the tendon undergoes some initial

degenerative changes [39,42]. Permanent integration oc-

curs as a result of progressive mineralization from thewalls of the bone tunnel into a collar of soft tissue callus

surrounding the tendon [15,26,38]. Regeneration of a

layered enthesis occurs after an intermediate phase,

during which the collagen fibres of woven bone are an-

chored to the fibre bundles of the scar tissue surrounding

the tendon [35].

Functional testing of a neo-enthesis to bone has shown

that by twelve weeks, sufficient interface strength can berecovered, such that failure during pull-out studies oc-

curred at another site [35]. These findings are comparable

hed by Elsevier Ltd. All rights reserved.

C.J. Pendegrass et al. / Journal of Orthopaedic Research 22 (2004) 1316–1324 1317

with functional testing of an intact tendon–bone junction,in which failure occurred preferentially at tendon mid-

substance or through sub-chondral bone, but rarely

through the layered enthesis [16]. Despite evidence for the

re-establishment of a direct enthesis at the tendon–bone

interface, there are few reports of regeneration and

attachment of a tendon to a metallic prosthesis.

The hypothesis in this study is that a tendon–implant

interface can develop, whose morphology resembles anormal enthesis. The application of a bioactive coating

on the implant surface and the use of autogenous can-

cellous bone and marrow graft were selected in an at-

tempt to generate an in vivo bone block. We feel that

this in vivo construct would provide the necessary

bioactive components for a successful tendon–implant

attachment. We have employed an ovine model to sim-

ulate the attachment of a tendon to a massive endo-prosthesis, and have studied the morphology of the

resulting neo-interface using light microscopy over a

twelve-week period.

Fig. 1. Plan view of tendon clamp; showing base plate (BP) with HA

coating and six spikes (S); and clamp lid (CL) with three spikes (S) and

drill hole perforations (DH); unassembled (A) and assembled (B).

Method

Implant

A customised titanium alloy (Ti–6Al–4V) base plate (30 mmlength) with a 70 lm plasma-sprayed coating of CAPTAL�

hydroxyapatite (Plasma Biotal Limited, Tideswell, UK) was producedto simulate the extra-medullary surface of a proximal tibial replace-ment. Nine interlocking press-fit spikes (1 mm diameter; 4 mm length)on the base plate and a clamp lid (20 mm length) provided the initialmechanical stability of the tendon–implant construct (Fig. 1(A)–(B)).The clamp provided an enclosed environment to isolate the biologicalprocesses occurring at the tendon–implant interface from the under-lying bone surface. The lid was H-shaped and perforated with eleven1 mm drill holes to allow access from overlying tissues.

Study design

The study was based on a Project Licence protocol accepted underthe UK Home Office Animals (Scientific Procedures) Act 1986. Twentyfour skeletally mature ovine ewes (Friesland breed) (55–90 kg) wereused in the study. In 12 animals (HA Group), the patellar tendon wasclamped directly onto the HA surface of the implant, whilst in theother 12 (Autograft Group), the tendon-HA interface was augmentedwith autogenous cancellous bone and marrow graft. The ovine stiflejoint is equivalent in form and function to the human knee, despiteanatomical differences, which do not necessarily allow a completecomparison. The ovine patellar tendon was selected for this modelsince sheep have a single true patellar tendon, whose most significantattachment is to the tibial tuberosity [1], with medial and lateral reti-nacular expansions as in the human knee. Other ruminants possessthree discrete patellar ligaments [13]. The size of the sheep, and hencethe ovine patellar tendon, are of a similar magnitude to that of thehuman, and during normal gait, a comparable physiological loading ofthe tendon–implant interface would occur.

Surgery

The animals received a sub-cutaneous pre-medication dose ofxylazine hydrochloride (0.2 mg/kg; Bayer plc, Bury St Edmonds,Suffolk, UK). Anaesthesia was induced using intravenous midazolam(2.5 mg stat dose; Roche Products Limited, Welwyn Garden City, UK)and ketamine hydrochloride (2 mg/kg; Fort Dodge Animal HealthLtd., South Hampton, UK), and was maintained with halothane 3%

(Merial Animal Health Ltd., Harlow, UK) and oxygen (4 l/min). Theright patellar tendon was approached through an 8 cm mid-line inci-sion over the stifle joint. The tendon was defined and elevated from thetibial tuberosity by sharp dissection. An osteotomy of the insertion sitewas performed with an oscillating saw to create a flat bone bed toaccommodate the base plate, which was attached with 2.7 mm self-tapping cortical bone screws (SYNTHES�, STRATEC Medical Ltd.,Welwyn Garden City, UK). In the Autograft Group, a slurry ofautogenous cancellous bone chips and marrow (1.5 g wet weight)harvested from the ipsilateral iliac crest was packed onto the wholelength of the HA surface of the implant. The clamp lid was assembledwith the stifle joint in extension (Fig. 2). The animals received doses ofsub-cutaneous ceftiofur antibiotic (1 mg/kg; Pharmacia & Upjohn,Northants, UK) intraoperatively and for five post-operative days,and intramuscular buprenorphine hydrochloride (10 lg/kg; Reckitt& Colman Products Ltd., Melton Mowbray, UK) for one post-oper-ative day. The animals were allowed to mobilize freely within a pen(3 · 4 m).

Histological analysis

Six animals from both Groups underwent euthanasia at six weeksusing intravenous 20% pentobarbital solution (J.M. Loveridge plc,Southampton, UK), and the remaining animals were killed by thesame method at twelve weeks. The normal (left) and operated patella-patellar tendon–proximal tibia units were harvested and fixed in 10%formalin for 10 days. The samples underwent ascending graded alcohol

Fig. 2. Plan view of clamped patellar tendon (T) with underlying bone

graft (BG), beneath clamp lid (CL).

1318 C.J. Pendegrass et al. / Journal of Orthopaedic Research 22 (2004) 1316–1324

dehydration, de-fatting in chloroform for five days, and embedding inLR White Resin (London Resin Company Limited, Reading, UK).Sections were cut, ground, and polished to 100 lm, stained withtoluidine blue and paragon, and underwent qualitative morphologicalassessment using an Olympus BH2 microscope (Olympus OpticalCompany Limited, Tokyo, Japan), linked to Zeiss KS300 3.0 imageanalysis software (Imaging Associates, Thame, UK). Three sections atone-third width intervals across the clamp were prepared for eachanimal. For each section, five fields of view along the length of the baseplate were observed at a magnification of ·100. A double-blind,quantitative analysis of collagen fibre orientation at the implant sur-face was performed by measuring the angles of elevation of the col-lagen fibres per field of view, with respect to the whole length of theHA-coated base plate which was taken as a point of zero reference,using Adobe� Photoshop� version 6.0.1 (Fig. 3(B)–(C)). The collagenfibre orientation in the normal patellar tendon–bone enthesis wasmeasured similarly, using the tangent to the surface of the tibialtuberosity as a point of zero reference (Fig. 3(A)). Comparisons weremade between the HA and Autograft Groups, and normal patellartendon–bone insertion using a Mann–Whitney U test (SPSS version10.1 for Windows). Semi-quantitative analysis of the bone–tendoninterface was carried out for all sections where bone graft was observedwithin the constraints of the clamp lid (20 mm from the proximalextent of base plate), whether associated with the HA coating or not.The tissue was described as either perforating-fibrous (Fig. 6(F)) orfibrocartilaginous (Fig. 7(B)) on the basis of the histological appear-ance at six and twelve weeks, and was expressed as a percentage. Theresults were tested for significance using a Mann–Whitney U test.

Fig. 3. Collagen fibre angle at insertion. Normal enthesis (A) ·100;Autograft Group (B) ·100; HA Group (C) ·200.

Results

All animals recovered well from the procedure and

early mobilization was observed in every case. One ani-mal developed a superficial wound infection that was

treated successfully with a five day course of ceftiofur. At

six and twelve weeks post-operation, vascular scar tissue

was observed overlying the implants. There was obvious

continuity between the unclamped and clamped tendon,

and no failures due to pull-out occurred. Regions of

ectopic calcification were observed in the tendon outside

the clamp in three out of six, and four out of six of theHA and Autograft specimens, respectively at six weeks,

and in two out of six, and four out of six of the HA and

Autograft specimens, respectively at twelve weeks.

Normal osseo-tendonous junction

The normal ovine patellar tendon insertion is a direct

enthesis composed of four distinct zones. The tissue

morphology alters as tendon, consisting of a dense

connective tissue of crimped, parallel, collagen fibres

Fig. 4. Normal osseo-tendonous junction, showing tendon (T), un-

mineralized fibrocartilage (UFC), tidemark (Tm), mineralized fibro-

cartilage (MFC) and bone (B) layers, stained with toluidine blue and

paragon; (A) bar¼ 1 mm; (B) chondrocyte-like cells (Ch) in lacunae

observed in UFC layer, bar¼ 100 lm.

Fig. 5. Low power image of a section showing the tendon (T), mid-

substance (MS), interface (I) and ingrowth (D) regions.

C.J. Pendegrass et al. / Journal of Orthopaedic Research 22 (2004) 1316–1324 1319

interspersed with spindle-shaped elongated fibroblast

nuclei, inserts into bone through regions of un-miner-alized and mineralized fibro-cartilage, with rounded cells

set in lacunae surrounded by a darkly stained amor-

phous matrix (Fig. 4(A)–(B)). The bone layer shows a

thin shell of cortical lamellar bone, beneath which lies

the cancellous bone of the proximal tibia. The deep

surface of the attachment shows a higher proportion of

cartilaginous tissue in keeping with the compressive

loads encountered there.

Tendon–implant interface

The low power image shown in Fig. 5 shows a lon-gitudinal section through the implant and clamped

tendon. At low magnification at both time points,

normal tendon (T) was seen to progress into a region

of clamped mid-substance tendon (MS). The tendon–

implant interface (I) in the HA Group consisted of

a tendon–HA interface only. By comparison, in the

Autograft Group both tendon–bone and tendon–HAinterfaces were observed, due to inconsistent bone graft

retention at the HA surface. A region of ingrowth (D)

from overlying vascular connective tissue was observed

in both Groups through the drill holes in the clamp lid

and distal to the end of the clamp lid. Subsequent figure

regions will be referred to those denoted in Fig. 5.

Six weeks post-operation

In both study Groups, normal tendon (T) running

from the patella towards the entrance of the clamp

consisted of crimped dense collagenous tissue with

parallel-orientated fibres interspersed with numerouselongated fibroblast nuclei. This region progressed into

the mid-substance (MS) clamped tendon which was a

tissue of mixed morphology consisting predominantly of

tissue displaying a normal parallel alignment of collagen

fibres along the line of tensile loading (Fig. 6(A) (Region

MS)), however some loss of fibre orientation was ob-

served (Fig. 6(B) (Region MS)). Fibroblast nuclei were

plump and oval in shape, and a mixed inflammatoryinfiltrate was observed, with populations of lympho-

cytes, macrophages and the occasional giant cell.

In the HA Group a disorganized fibrous tissue layer

was observed at the tendon–HA interface (Fig. 6(E)

(Region I)). In the Autograft Group, bone was observed

undergoing active remodelling (Fig. 6(C) (Regions I and

MS)). Where bone graft was intimately associated with

the HA layer, an interface was observed, with tendonattached to the bone graft–HA construct via perforating

collagen fibres (Fig. 6(F)–(G) (Region I)). Where no

bone was seen at the HA surface the interface between

the tendon and HA was similar to that observed in the

HA Group. In some sections, bone was observed away

from the HA surface. The interface between the tendon

and these bone islands (Fig. 6(D) (Region MS)) was

Fig. 6. Histology at six weeks; (A) HA Group section showing clamped mid-substance with some parallel-orientated fibres, bar¼ 100 lm; (B) HAGroup section of disorganized collagenous tissue, bar¼ 100 lm; (C) Autograft Group section of bone undergoing active remodelling with visible

osteoblasts (Ob) and osteoclasts (Oc), bar¼ 100 lm; (D) Autograft Group section showing perforating-fibrous interface. Collagen fibres (CF)

anchoring to woven bone (WB), bar¼ 100 lm; (E) HA Group section of disorganized fibrous tissue adjacent to HA-coated base plate (BP) in HA

Group, bar¼ 200 lm; (F) bar¼ 100 lm, and (G) bar¼ 50 lm, Autograft Group section of soft tissue–bone–HA perforating-fibrous interface with

collagen fibres (CF) anchoring soft tissue to bone; (H) HA Group section of tissue ingrowth (D) over edge of clamp lid (CL), bar¼ 200 lm.

1320 C.J. Pendegrass et al. / Journal of Orthopaedic Research 22 (2004) 1316–1324

C.J. Pendegrass et al. / Journal of Orthopaedic Research 22 (2004) 1316–1324 1321

similar to that between tendon and bone retained onthe HA surface (Fig. 6(F) (Region I)). In some sections

in both Groups, small regions of the central tendon

mid-substance displayed areas of fatty and vacuolar

degeneration with a sparse network of collagen fibres,

interspersed with denser collagenous matrix. Ingrowth

of the overlying tissues occurred through the drill holes

in the lid of the clamp and into the open region above

the distal base plate (Fig. 6(H) (Region D)). Bone andmarrow slurry that had been packed onto the HA sur-

face of the distal section of the base plate (10 mm length

not constrained by clamp lid) was not observed, and the

tendon–implant interface in this region was similar to

that observed in the HA Group, as a disorganized fi-

brous tissue layer.

Twelve weeks post-operation

In both Groups, the tendon mid-substance (MS) and

ingrowth (D) regions consisted predominately of densecollagenous tissue, with crimped fibres parallel to the

direction of tensile loading (Fig. 7(A)). The fibroblast

nuclei had adopted a more elongated morphology

compared with those observed at six weeks. In the HA

Group the tendon remained in close contact with the

HA coating, and was observed as a dense collagenous

Fig. 7. Autograft Group histology at twelve weeks; (A) mid-substance mater

(B) Autograft Group fibrocartilaginous interface with regions of tissue contai

tissue interface in the HA Group showing soft tissue in close contact with the

HA Group interface showing dense collagenous tissue running parallel to H

tissue orientated parallel to the surface of the implant(Fig. 7(C)–(D)(Regions I and MS)). In the Autograft

Group, the tendon–bone interface had developed a

layered morphology with a clearly defined fibrocarti-

laginous zone (Fig. 7(B) (Region I)). This was observed

both in regions where bone remained intimately asso-

ciated with the HA layer, and also where bone islands

were found away from the HA surface. The unclamped

region of the base plate (10 mm length) in the AutograftGroup showed a tendon–implant interface similar to the

HA Group. Quantitative analysis showed there to be no

significant difference in the collagen fibre orientation to

the implant surface (mean± standard error) between the

HA (13.68�±0.87�) and Autograft (19.73�±1.02�)Groups (p > 0:05). The fibre orientations in both

Groups were significantly lower than the fibre insertion

angle at a normal patellar tendon enthesis (34.96�±2.55�) (p < 0:05). However, the difference in site-specific

collagen fibre orientation angle between the Autograft

Group and the normal patellar tendon enthesis (14.13�±1.77�) was found to be significantly lower than the dif-

ference observed between the HA Group and normal

patellar tendon enthesis (27.18�±3.54�) (p < 0:01). Inthe Autograft Group the nature of the interface between

the bone graft and tendon (within the constraints of theclamp lid), expressed as a mean percentage (± standard

ial with morphology resembling that of normal tendon, bar¼ 100 lm;ning chondrocyte-like cells (Ch) in lacunae, bar¼ 100 lm; (C) HA-softHA-coated base plate (BP), bar¼ 100 lm; (D) higher magnification ofA surface, bar¼ 50 lm.

1322 C.J. Pendegrass et al. / Journal of Orthopaedic Research 22 (2004) 1316–1324

deviation) were 16% (±1.4) fibrocartilaginous, and 84%(±5.7) perforating-fibrous at six weeks. By twelve weeks

the interface proportions were 76% (±3.4) and 24%

(±1.6), respectively. The bone graft–tendon interfaces in

the Autograft Group were significantly more fibrocar-

tilaginous in nature at twelve weeks compared to those

at six weeks (p < 0:01). In the HA Group no bone was

found within the clamp and hence no semi-quantitative

analysis of tendon–bone interface was performed.

Discussion

Tendon and ligament insertions into bone are char-

acterized based on their differences in morphology into

direct and indirect [2,8,36]. The normal ovine patellar

tendon insertion displayed the characteristics of a direct

enthesis, with four distinct zones; tendon, fibrocartilage,

mineralized fibrocartilage and bone [10,41].

The morphology reported in tendon–bone tunnel

models has shown collagen fibres spanning the devel-oping interface between tendon and the woven bone

tunnel lining [25,27,31,35,38,39]. The environment

within the clamp is unlike that observed with tendon

healing in a bone tunnel due to the presence of a metal

implant. However, we feel that augmentation with

autogenous bone and marrow graft applied to the HA-

coated implant surface created an environment capable

of supporting the development of a tendon–implantattachment. The presence of perforating collagen fibres

between tendon and bone in the Autograft Group at six

weeks is consistent with development of an indirect

enthesis. In the HA Group, and areas of HA not cov-

ered with bone graft in the Autograft Group, fibrous

tissue encapsulation was observed. Absolute values

for fibre orientation at the interface did not reach sig-

nificance between the HA and Autograft Groups. Inthe Autograft Group, the interface contained regions

with both tendon–HA and tendon–bone attachment,

consisting of both indirect and direct type interfaces

(described here as perforating-fibrous and fibrocarti-

laginous, respectively). The collagen fibre angles in both

of these regions within the same sections contributed to

the overall results, and were in part responsible for the

apparent lack of statistical significance despite obviousqualitative differences in morphology. However, the site-

specific analysis highlighted that regions with and

without bone at the interface display significant differ-

ences of collagen fibre orientation. We speculate that the

overall variation in interface morphology between the

HA and Autograft Groups is predominately due to

biological augmentation with marrow and bone, since

both Groups were exposed to a similar local mechanicalenvironment. However, the presence of the additional

bone graft volume may have influenced the degree of

tendon compression, and thus affect the morphology of

the tissue formed. Collagen fibre insertion angle in thenormal enthesis varies with position along the tuberosity

and we have presented a mean value. Variation in fibre

orientation with respect to bone or implant surfaces has

been shown to reflect the nature and integrity of the

resulting attachment [11]. The flat implant surface, and

angle of osteotomy may have contributed to the shallow

insertion angles observed in the experimental groups,

and thus the local mechanical environment may haveinfluenced our findings. In addition, collagen fibre ori-

entation varies with indirect and direct type insertions

[43], and their relative proportions will affect our ob-

served values, hence reflect the degree of maturation of

the neo-enthesis.

The marrow provided a source of pluripotent cells

with osteogenic potential [3,6,28], whilst the trabeculae

of the cancellous bone provided an osteoconductivescaffold for subsequent remodelling. HA has been shown

to induce differentiation of pluripotent cells along the

osteoblast linage [37], provide a bioactive surface that

can support mineralization [9,30,32] and induce collagen

synthesis [40]. In our study, biointegration was observed

with bone in intimate apposition with the HA-coated

base plate.

At twelve weeks, fibrocartilaginous regions betweenthe tendon and bone trabeculae in the Autograft Group

differentiated the interface from that observed in the

Autograft Group at six weeks and the HA Group at

the both time points. Semi-quantitative analysis of the

amount of fibrocartilaginous tissue present at the bone

graft–tendon interface showed an increase of 60% be-

tween six and twelve weeks. This was consistent with

observations in longer-term tendon–bone tunnel healingstudies [35] and models of anterior cruciate ligament

reconstruction [34], where a transition from an indirect

to direct morphology has been characterized. Other

studies have shown the absence of interface maturation

[20,25], however the type of model and study duration

may have contributed to these findings. We acknowl-

edge that the fibrocartilaginous tissue could represent

progression of endochondral ossification rather than atrue four-zoned direct interface and feel that longer term

follow-up would help clarify the fate of this tissue.

Limitation of bone graft retention meant that the

tendon–bone–HA interface was observed to a maximum

of 66% of the base plate length (constrained clamped (20

mm) as a proportion of base plate (30 mm)). The semi-

quantitative analysis of the tendon–bone interface was

performed regardless of whether the bone was attachedto the HA surface, and reflects the nature of the inter-

face rather than the problem of bone–HA association

we experienced with our methodology. We feel that a

combination of factors including graft resorption, de-

fects created during resin polymerisation and the use of

morcelised bone chips contributed to poor graft reten-

tion. We propose that the addition of a macro-textured

C.J. Pendegrass et al. / Journal of Orthopaedic Research 22 (2004) 1316–1324 1323

substrate for the HA coating and the use of a bone blockrather than morcelised graft may encourage more min-

eralized tissue ingrowth, as demonstrated by previous

Research Groups [9,12,30]. Other attempts to reattach

soft tissues to metal implants augmented with bone

graft, have observed graft resorption and replacement

with a ‘thick collagen bundle’ [22,23]. However, marrow

supplementation of a bone plate has been shown to

initiate an osteoinductive reaction at the tendon–implant interface in other studies in this field [7]. The

implant we used and our study design are similar to

previous work published in this journal, which showed

comparable histological reconstitution of a layered

tendon–bone–implant interface with functional weight-

bearing recovery of 90.3% at sixteen weeks [21].

The mechanical fixation provided by the interlocking-

spike mechanism enabled immediate unprotected weightbearing, with no evidence for attachment failure. We

acknowledge that functional assessment with gait ana-

Fig. 8. (A) Prototype clamp device used in pilot study with solid lid; (B)

necrotic tendon at autopsy processed in paraffin wax and stained with

haematoxylin and eosin, bar¼ 100 lm.

lysis or direct biomechanical testing would provide use-ful objective data to supplement the histomorphological

results presented here. By its nature, within the clamp

there will be compression of the tendon, and the internal

environment will be relatively avascular. We did not

observe tissue necrosis as a consistent finding, which

concurs with other work in this field where speculation

that perforation of a tendon under compression results

in less tissue damage [18]. A prototype design (Fig. 8(A))tested in a pilot study showed that a more enclosed

environment did lead to significant tendon necrosis (Fig.

8(B)). The modified H-shaped clamp lid with perforated

drill holes was designed as a compromise to optimise

ingrowth from the overlying tissues and allow vascular

infiltration, whilst maintaining sufficient mechanical

fixation to clamp the tendon. However, we speculate

that this design modification represents only one factorthat could influence the observed differences. We feel that

the clamp is a valid device for modelling tendon attach-

ment to an implant experimentally, but further study is

necessary before considering clinical application. We

conclude that tendon reconstruction to a metal implant,

augmented with autogenous cancellous bone and mar-

row has been shown to result in the development of a

neo-enthesis with morphological features similar to thoseobserved in time-course studies of tendon reattachment

to bone.

Acknowledgements

This study was supported by Stanmore Implants

Worldwide Ltd.

References

[1] Allen MJ, Houlton JEF, Addams SB, Rushton N. The surgical

anatomy of the stifle joint in sheep. Vet Surg 1998;27:596–605.

[2] Benjamin M, Evans EJ, Copp L. The histology of tendon

attachments to bone in man. J Anat 1986;149:89–100.

[3] Beresford JN. Osteogenic stem cells and the stromal system of

bone and marrow. Clin Orthop 1989;240:270–80.

[4] Bickels J, Wittig JC, Kollender Y, et al. Reconstruction of the

extensor mechanism after proximal tibia endoprosthetic replace-

ment. J Arthroplasty 2001;16:856–62.

[5] Cannon SR. Massive prostheses for malignant bone tumours of

the limbs. JBJS Br 1997;79:497–506.

[6] Caplan AI. Mesenchymal stem cells. J Orthop Res 1991;9:641–50.

[7] Chao EY, Inoue N, Ikeda K, et al. Formation of pseudo-

subchondral bone plate for tendon attachment to metallic implant

via osteoinduction. Trans Orthop Res Soc 1997;22(Suppl 1):1.

[8] Clark J, Stechschulte Jr DJ. The interface between bone and

tendon at an insertion site: a study of the quadriceps tendon

insertion. J Anat 1998;192(Pt 4):605–16.

[9] Cook SD, Thomas KA, Kay JF, Jarcho M. Hydroxyapatite-

coated porous titanium for use as an orthopedic biologic

attachment system. Clin Orthop 1988;230:303–12.

[10] Cooper RR, Misol S. Tendon and ligament insertion. A light and

electron microscopic study. JBJS Am 1970;52:1–20.

1324 C.J. Pendegrass et al. / Journal of Orthopaedic Research 22 (2004) 1316–1324

[11] Donley TG, Gillette WB. Titanium endosseous implant-soft

tissue interface: A literature Review. J Periodontol 1991;62:153–

60.

[12] Ducheyne P, Hench LL, Kagan A, et al. Effect of hydroxyapatite

impregnation on skeletal bonding of porous coated implants.

J Biomed Mater Res 1980;14:225–37.

[13] Dyce KM, Sack WO, Wensing CJG. Textbook of veterinary

anatomy. In: The Hind Limb of the Ruminants, 2nd ed., 1996.

p. 759–60 [chapter 33].

[14] Eckardt JJ, Matthews JG, Eilber FR. Endoprosthetic reconstruc-

tion after bone tumor resections of the proximal tibia. Orthop Clin

North Am 1991;22:149–60.

[15] Forward AD, Cowan RJ. Tendon suture to bone. JBJS Am

1963;45-A(4):807–23.

[16] Gao J, Rasanen T, Persliden J, Messner K. The morphology of

ligament insertions after failure at low strain velocity: an evalua-

tion of ligament entheses in the rabbit knee. J Anat 1996;189(Pt 1):

127–33.

[17] Gosheger G, Hillmann A, Lindner N, et al. Soft tissue recon-

struction of megaprostheses using a trevira tube. Clin Orthop

2001;393:264–71.

[18] Gottsauner-Wolf F, Egger EL, Schultz FM, Sim FH, Chao EY.

Tendons attached to prostheses by tendon–bone block fixation:

an experimental study in dogs. J Orthop Res 1994;12(6):814–

21.

[19] Grimer RJ, Taminiau AM, Cannon SR. Surgical outcomes in

osteosarcoma. JBJS Br 2002;84:395–400.

[20] Hausman M, Bain S, Rubin C. Reluctance of metaphyseal bone to

heal to tendon: histologic evidence for poor mechanical strength.

Trans Orthop Res Soc 1989;14:277.

[21] Inoue N, Ikeda K, Aro H, et al. Biologic tendon fixation to

metallic implant augmented with autogenous cancellous bone

graft and bone marrow in a canine model. J Orthop Res 2002;

20:957–66.

[22] Inoue N, Ikeda K, Young DR, et al. Tendon fixation to porous

metallic implant using autogenous bone graft augmentation.

Trans Orthop Res Soc 1996;21:352.

[23] Inoue N, Young DR, Ikeda K, et al. Fiber orientation in soft

tissue attachment to metallic prosthesis. Trans Orthop Res Soc

1995;20:615.

[24] Jeon DG, Kawai A, Boland P, Healey JH. Algorithm for the

surgical treatment of malignant lesions of the proximal tibia. Clin

Orthop 1999;358:15–26.

[25] Jones JR, Smibert JG, McCullough CJ, et al. Tendon implanta-

tion into bone: an experimental study. J Hand Surg [Br] 1987;

12:306–12.

[26] Kernwein GA. A study of tendon implantation into bone. Surg,

Gynaecol Obstet 1942;75:794–6.

[27] Liu SH, Panossian V, al Shaikh R, et al. Morphology and matrix

composition during early tendon to bone healing. Clin Orthop

1997;339:253–60.

[28] Majors AK, Boehm CA, Nitto H, et al. Characterization of

human bone marrow stromal cells with respect to osteoblastic

differentiation. J Orthop Res 1997;15:546–57.

[29] Malawer MM, McHale KA. Limb-sparing surgery for high-grade

malignant tumors of the proximal tibia. Surgical technique and a

method of extensor mechanism reconstruction. Clin Orthop 1989;

239:231–48.

[30] Meffert RM. Do implant surfaces make a difference? Curr Opin

Periodontol 1997;4:104–8.

[31] Oguma H, Murakami G, Takahashi-Iwanaga H, et al. Early

anchoring collagen fibers at the bone–tendon interface are

conducted by woven bone formation: light microscope and

scanning electron microscope observation using a canine model.

J Orthop Res 2001;19:873–80.

[32] Ohgushi H, Goldberg VM, Caplan AI. Heterotopic osteogenesis

in porous ceramics induced by marrow cells. J Orthop Res 1989;7:

568–78.

[33] Ozaki T, Kunisada T, Kawai A, et al. Insertion of the patella

tendon after prosthetic replacement of the proximal tibia. Acta

Orthop Scand 1999;70:527–9.

[34] Panni AS, Milano G, Lucania L, Fabbriciani C. Graft healing

after anterior cruciate ligament reconstruction in rabbits. Clin

Orthop 1997;343:203–12.

[35] Rodeo SA, Arnoczky SP, Torzilli PA, et al. Tendon-healing in a

bone tunnel. A biomechanical and histological study in the dog.

JBJS Am 1993;75:1795–803.

[36] Rufai A, Ralphs JR, Benjamin M. Structure and histopathology

of the insertional region of the human Achilles tendon. J Orthop

Res 1995;13:585–93.

[37] Rust PA, Blunn GW, Cannon SR, Briggs TW. Use of a novel

bioreactor to grow a bone block using human mesenchymal stem

cells on a porous HA scaffold. Trans Orthop Res Soc 2003:74.

[38] Shaieb MD, Singer DI, Grimes J, Namiki H. Evaluation of

tendon-to-bone reattachment: a rabbit model. Am J Orthop 2000;

29:537–42.

[39] Skoog T, Persson BH. An experimental study of the early healing

of tendons. Plast Reconstr Surg 1954;13:384–99.

[40] Soballe K, Hansen ES, Brockstedt-Rasmussen H, et al. Hydroxy-

apatite coating enhances fixation of porous coated implants. A

comparison in dogs between press fit and noninterference fit. Acta

Orthop Scand 1990;61:299–306.

[41] Thomas DB, Inoue N, Cosgarea A, Chao EYS. A histomorpho-

metric analysis of the human patellar tendon insertion. Trans

Orthop Res Soc 1999;24:1090.

[42] Whiston TB, Walmsley R. Some observations on the reaction of

bone and tendon after tunnelling of bone and insertion of tendon.

JBJS Br 1960;42-B(2):377–86.

[43] Woo SL-Y, Gomez MA, Sites TJ, et al. The biomechanical and

morphological changes in the medial collateral ligament of the

rabbit after immobilization and remobilization. JBJS Am 1987;69-

A(8):1200–11.