Introduction
Failure of osteosynthesis can be caused by an inadequate mechanical or biological
environment, or both. Biological inactivity at the fracture site, despite adequate
fixation, may result in inadequate bone healing and is defined as a non-union.[1]
[2] A subclassification of non-union is atrophic non-union, which is characterized by
complete absence of restorative processes.[2]
[3] Treatment of an atrophic non-union is often challenging and requires surgical debridement
of all non-viable tissue. To stimulate bone regeneration and to bridge a segmental
bone defect that is not expected to heal without a secondary (surgical) intervention,
rigid fixation is of vital importance.[3] If the fracture non-union involves an articular surface, segmental bone bridging
should preferably also restore anatomic alignment of the joint surface. To enhance
biological activity, autogenous cortico-cancellous bone grafts together with recombinant
human bone morphogenetic protein-2 (rhBMP-2) or rhBMP-7 can be used.[4]
[5] In addition, large segmental bone defects benefit from the use of free autogenous
cortico-cancellous grafts (e.g. rib sections) or vascularised grafts from and adjacent
paired bone, whenever possible.[6] In some cases of non-union, and due to a poor prognosis, high expenses and unsuccessful
treatment attempts, a high degree of morbidity may prevail, with the end-point decision
being amputation. However, recent innovations in additive manufacturing (AM) offer
a surgical solution when off-the-shelf implants with fixed sizes fail to reach an
acceptable solution to bridge a non-union with a segmental bone defect, especially
in cases with involvement of a joint surface that requires accurate anatomic restoration.[7]
[8]
Additive manufacturing is used for rapid prototyping and small series manufacturing.[7] Digital three-dimensional (3D) models are in silico created by using medical imaging. Standard Digital Imaging and Communications in
Medicine (DICOM) images are reconstructed to 3D models in readily available segmentation
software. This AM technique enables the manufacturing of implants directly from a
digital 3D stereolithography model by direct metal topographical printing using the
technique of selective laser melting. The main advantages of this technique are the
ability to manufacture complex 3D geometries, to develop scaffolds with well-controlled
pore size, porosity and interconnecting pore size as well as an adequate resistance
force, which is very difficult to accomplish through conventional manufacturing technologies.[9] This technique has a twofold effect on efficiency. It will reduce waiting time between
diagnosis and surgery as well as a reduction in surgical time with improved gap bone
healing.[7]
[8] The additional advantage of porous metals lies in their open space surface for the
ingrowth of bone tissue, hence accelerating the osseointegration process.[9] Over the last decade, AM is increasingly used in both human and veterinary medicine
and multiple case studies have been reported.[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26] In veterinary surgery this technique has mainly been used for reconstruction of
skull defects but recently, new applications for the treatment of hip dysplasia have
been published.[27]
[28]
[29]
[30]
The aim of this study was to describe the implant design and the clinical use of patient-specific
3D-printed osteotomy guides, and patient-specific 3D-printed porous titanium implants
to bridge segmental antebrachial non-unions in two dogs involving the elbow joint.
Case Description
Case 1
A 1-year-old, 40 kg, female, Bernese Mountain Dog was referred 11 weeks after bi-oblique
dynamic proximal ulnar osteotomy (BOD-PUO) of the right antebrachium for elbow dysplasia
with medial coronoid disease and incongruency. The osteotomy in the ulna was performed
unusually close to the elbow joint. Three weeks after the BOD-PUO the dog developed
a surgical site infection, which was treated with broad-spectrum antibiotic medications.
Despite conservative management with antibiotic medications and non-steroidal anti-inflammatory
drugs, the dog developed a persistent lameness of the right thoracic limb grade 4
out of 5, moderate-to-severe muscle atrophy of the antebrachial musculature and joint
effusion of the elbow joint.[31] Concurrent orthopaedic abnormalities included bilateral subluxations of the hip
joints due to hip dysplasia. Radiographic examination revealed an atrophic non-union
of the right proximal ulna with malalignment and intra-articular extension ([Fig. 1A]). There was severe osteoarthrosis with subchondral bone lesions, joint effusion
and incongruency of the right elbow. Computed tomography (CT) of both forelimbs revealed
a chronic osteolytic defect of 10 mm of the right proximal ulna and severe right elbow
osteoarthritis with fragmentation of the medial coronoid process, incongruity and
mineralization and erosion of cartilage and subchondral erosive lesions.
Fig. 1 (A) A 1-year-old Bernese Mountain dog (case 1) was presented with an atrophic non-union
of the right proximal ulna, malalignment with intra-articular extension and severe
elbow osteoarthrosis. (B) The left ulna (also showing elbow osteoarthrosis) was mirrored to obtain an approximation
of the original anatomy of the right ulna for the design and additive manufacturing
of a three-dimensional patient-specific implant (blue) for ulnar reconstruction. (C) Immediate postoperative radiograph showing the bridging titanium implant restoring
elbow congruency and fixation with two plates. (D) Twenty months postoperative radiograph showing stable position of the implant.
Implant Design
A patient-specific 3D implant was designed to fill the segmental defect in the right
ulna and restore the ulnar elbow joint alignment. To develop the patient-specific
implant, first the DICOM files of the CT scan (250 mAs, 120 kV, 0.6 mm slice thickness)
were exported from the imaging archive system to Mimics (v21, Materialise NV., Leuven,
Belgium) for anatomical segmentation. Standard bone threshold values (226 HU – upper
threshold) were taken to segment the bone. The left ulna was mirrored to obtain an
approximation of the original anatomy of the right ulna for the accomplishment of
a 3D patient-specific implant for ulnar reconstruction ([Fig. 1B]). Hereafter, the non-union was manually segmented together with a digitally grown
resection margin that was used to simulate the minimal needed resection area to obtain
straight osteotomies enhancing the press fit placement of the 3D patient-specific
implant. Subsequently the anatomical models were transferred as stereolithography
files to 3-matic software (v13, Materialise NV., Leuven, Belgium) in which the design
took place. The proximal and distal resection guides were designed, to remove the
irregular non-viable bony edges of the non-union. The resulting gap between the two
osteotomies determined the size of the required implant (20 mm length × 18 mm in width).
The ulnar reconstruction implant was designed as press fit porous titanium (70% porous,
500–600 µm pore size, Dodecahedron unit cell) to allow bony ingrowth at the implant
bone interface. Additionally, the central region of the implant contained two holes
for application of an autogenous cancellous bone graft together with demineralized
bone matrix ([Fig. 2A]).
Fig. 2 (A) Case 1: the ulna was reconstructed with a three-dimensional (3D)-printed porous
titanium implant with central canals for the application of an autogenous cancellous
bone graft with demineralized bone matrix. (B) Intraoperative view showing the 3D-printed surgical saw guide (white) that was applied on the caudal ulnar surface. (C) The implant was fitted in between both ulnar fragments and fixated with two plates
(not shown). (D) Micro-computed tomography at 24 months postoperatively (after euthanasia). Bone
ingrowth and bridging are evident on a sagittal median view through the central canal
of the implant.
Production and Presurgical Planning
The osteotomy saw guides and the left and right ulna were 3D-printed in Nylon (PA12)
on an EOS P110 printer (EOS, Krailling, Germany). The implant was 3D-printed in medical
grade titanium alloy Ti-6Al-4V ELI grade 23 using direct metal printing on a ProXDMP320
printer (3D Systems, Leuven, Belgium) by the process of selective laser melting of
powders containing titanium using topographical printing. Post-processing included
chronologically: hot-isostatic-pressing, screw wiretapping, polishing and manual cleaning.
Rehearsal surgery was performed using the 3D-printed bone specimen together with the
saw guides and titanium scaffold, allowing precise selection of the size and length
of the plates and contouring to the bone and presurgical planning for the screw trajectories.
Before surgery both the guides and the implant were manually cleaned, and standardized
autoclave sterilized at the in-house sterilization facility.
Surgery
The patient was placed in dorsal recumbency with the affected limb suspended and retracted
caudally for draping. A caudal approach to the proximal shaft of the ulna was performed.[32] Subperiosteal elevation and medial retraction of the flexor carpi ulnaris muscle
and lateral retraction of the extensor carpi ulnaris muscle exposed the ulnar shaft.
After debridement of both fragments, the 3D-printed patient-specific surgical saw
guides were applied to the caudal ulnar bone surface using manual press fit and with
the help of a double pointed bone clamp. Perfect fit of the saw guide was confirmed
before the distal and proximal osteotomies were performed with a 0.6 mm thick oscillating
saw (DePuy Synthes, Johnson-Johnson, Oberdorf, Switzerland) ([Fig. 2B]). Following removal of the ulnar non-union, the newly formed bone ends were inspected
for viability and bleeding. The patient-specific 3D implant was embedded with the
autogenous cancellous bone graft harvested from the ipsilateral tuberculum majus and
mixed with artificially engineered demineralized bone matrix (Attrax Putty, Nuvasive,
San Diego, California, United States). The titanium implant was fitted in between
both ulnar fragments ([Fig. 2C]) and after restoration of the ulnar alignment and elbow congruence a hybrid dynamic
compression plate 3.5/2.7 mm (hybrid DCP [HDCP]) (DePuy Synthes, Johnson-Johnson,
Oberdorf, Switzerland) was placed caudally on the tension side of the ulna under fluoroscopic
guidance. An additional 2.7 mm DCP (DePuy Synthes, Johnson-Johnson, Oberdorf, Switzerland)
was applied medially and fixed with two bicortical cortical screws to prevent medial
displacement of the titanium implant. Before routine closure, 4 × 32.5 mg gentamycin
sponges (Garacol, SERB SA, Brussels, Belgium) were applied locally around the implant.
Follow-Up
Postoperative radiographs showed a stable construction ([Fig. 1C]). At the 4 months postoperative check, there was clinical improvement in use of
the right thoracic limb and CT revealed no changes with postoperative radiographs.
At 5 months postoperatively, the dog developed a progressive worsening lameness at
the right thoracic limb. Five out of seven cortical screws used with the caudal HDCP
plate were broken. During surgical removal of the hybrid plate, inspection of the
3D implant showed a macroscopic stable embedding between the proximal and distal adjacent
bone segments. Eleven months after surgery, the medial DCP plate was removed because
of failure of the two screws and intra-operative inspection showed continued embedding
of the 3D implant. At 20 months follow-up (after the initial surgery), the dog showed
significant clinical improvement with minimal lameness compared to previous orthopaedic
examinations. The 3D implant was radiologically stable in position with a radiolucent
line around the entire implant ([Fig. 1D]). Whether the radiolucent line was caused by failure of the implant to fully osseointegrate
or by the Uberschwinger artefact (radiolucent halo around metal when there is a large
density difference between adjacent objects) could not be determined.[33]
[34]
[35] There was a mild-to-moderate progression of the elbow joint osteoarthrosis. Two
years postoperatively the dog developed right pelvic limb lameness. Radiographs of
the stifle revealed an osteolytic process in the distal femur with aggressive characteristics.
The owner elected euthanasia of the dog and consented to postmortem collection of
the antebrachial segment. At the time of collection of the bone sample, it appeared
that the implant was macroscopically completely integrated with the ulna segment.
3D micro-CT scan (VECTor6/CT system, MILabs B.V., Utrecht, the Netherlands) was obtained
of the bone specimen with the following parameters: multi-circle 360 degrees acquisitions,
tuber voltage of 55KV, tube current of 0.19 mA, exposure time of 75 ms per projection,
angle increment of 0.5 degrees and 50 μm reconstructed isotropic voxel size using
3D Feldkamp filtered back-projection reconstruction. Micro-CT revealed bone in-growth
in the medullar cavity of the porous segment ([Fig. 2D]); however, ingrowth of bone in the smaller porous structure was too small for detection.
Histopathology was performed on the collected antebrachial segment to study bone ingrowth.
The formalin fixed samples were embedded in polymethylmethacrylate. The polymethylmethacrylate
embedded plug was cut with a Leica 4 SP1600 Saw Microtome system (Leica) to yield
30 to 50 µm sections and these were stained with basic fuchsine-methylene blue. Overview
pictures were made with a Thunder imaging system (Leica) and revealed that there was
new bone formation present in the medullary cavity of the porous segment but absence
of osseointegration in the smaller porous segment at the bone–implant interface ([Fig. 3]).
Fig. 3 (A) Overview – New bone formation present in the medullary cavity of the porous segment
(*). (B) Absence of osseointegration in the smaller porous segment at the bone–implant interface
(<).
Case 2
A 4-year-old, 25 kg, male, Schapendoes, was presented with a history of a left-sided
transverse olecranon fracture due to a traffic accident 3 years ago which was treated
by open reduction and internal fixation and complicated by osteomyelitis. Implants
were removed after 12 weeks and with physiotherapy the dog regained some weight-bearing
function of the left thoracic limb with severe intermittent non-weight-bearing lameness.
Two weeks before the presentation at the university, the dog suffered another road
traffic accident and was therefore referred. Radiological examination revealed a chronic,
complete, transverse, intra-articular non-union of the left olecranon and moderate
osteoarthrosis of the elbow joint and subchondral sclerosis of the ulna ([Fig. 4A]). A CT scan was performed as well as an arthrocentesis to assess whether there was
an infectious arthritis. A chronic osteolytic defect of approximately 12 mm of the
proximal left ulna was present with concurrent bilateral signs of medial coronoid
disease. Prior to surgery, and at 6 weeks and 10 weeks after surgery, ground reaction
forces were measured with a quartz crystal piezoelectric force plate (Kistler type
9261, Charnwood Dynamics Limited, Rothley, UK) together with the Kistler 9865E charge
amplifiers, as described previously.[36]
[37] Measurements were obtained with a frequency of 100 Hz. Ground reaction forces were
measured in the mediolateral (Fx), craniocaudal (Fy), and vertical (Fz) direction.
The presurgical measurement revealed a Fzmax of 12 N/kg bodyweight of the right thoracic limb and Fzmax of 0 N/kg bodyweight of the left thoracic limb (i.e. no weight bearing).
Fig. 4 (A) A 4-year-old Schapendoes (case 2) was presented with a chronic, complete, transverse,
intra-articular non-union of the left olecranon and a moderate amount of osteoarthrosis
of the elbow joint. (B) Design of the three-dimensional (3D) patient-specific implant with incorporation
of a 3D-printed titanium plate with seven holes. Immediate (C) and 6 weeks (D) postoperative radiographs showing stable presentation of the bridging porous implant
and plate that was fixed with five 2.7 mm and two 2.4 mm titanium screws. The semilunar
ulnar trochlea has been restored resulting in elbow congruency.
Implant Design
Saw guides and a patient-specific 3D implant were designed from the CT images in a
similar manner as in case 1. The ulnar reconstruction implant was manufactured as
press fit porous titanium (70% porous, 500–600 µm pore size, Dodecahedron unit cell)
to allow bony ingrowth at the implant bone interface ([Fig. 5A]). The porous titanium implant was modified in comparison to case 1 by incorporating
a 3D-printed titanium 6-hole 2.7 mm plate on the caudal side and designing two drill
guides for the screws to capture the screw direction and screw length for optimal
bone stock in the proximal and distal fragment ([Fig. 4B]). Additionally, the central region of the bone-bridging-implant contained 1 hole
for application of an autogenous cancellous bone graft. The surgical saw guides and
drill guides were 3D-printed in Nylon (PA12) on an EOS P110 printer (EOS, Krailling,
Germany) and the implant was 3D-printed in medical grade titanium alloy Ti-6Al-4V
ELI grade 23 using direct metal printing on a ProXDMP320 printer (3D Systems, Leuven,
Belgium). Post-processing of the implant included polishing and HIP treatment. Before
surgery, both, the guides and the implant were manually cleaned and sterilized.
Fig. 5 (A) Case 2: the ulna was reconstructed with a three-dimensional (3D)-printed titanium
implant consisting of an ulnar bridging porous part and a caudal 2.7 mm plate. The
porous part contained 1 hole for application of an autogenous cancellous bone graft
and two screw holes for 2.4 mm titanium screws. (B) Intraoperative view showing the 3D-printed surgical saw guide (white) that was held in place with a Kirschner wire at the time of the osteotomy. (C) The patient-specific 3D implant was fitted in between both ulnar fragments and fixed
with the plate and seven screws.
Surgery
Surgical approach to the caudal side of the ulna was performed in a similar fashion
as in case 1. The anconeal process was used as lever to dislocate the articulating
surfaces to gain better visualization of the proximal fragment and to better assess
the intra-articular ulnar defect. After debridement of both fragments, the 3D-printed
patient-specific osteotomy saw guides were applied to the caudal ulnar bone surface
and when perfect fit was confirmed, the saw guides were temporarily fixed with 1.6 mm
Kirschner wires ([Fig. 5B]). Distal and proximal osteotomies were performed with a 0.6 mm thick oscillating
saw and new bone formations on the caudal and medial part of the ulna were removed.
The implant was filled through the central hole with an autogenous cancellous bone
graft harvested from the ipsilateral tuberculum majus and was then fitted in between
both ulnar fragments. First the implant was fixated to the proximal fracture fragment
using the proximal drill guide by placing a central 2.4 mm titanium cortex screw (Unilock,
DePuy Synthes, Johnson-Johnson, Oberdorf, Switzerland). After reduction in the anconeal
process and alignment of the proximal and distal ulnar fragments, fixation of the
most distal part of the plate occurred using the distal drill guide and placing three
2.7 mm titanium screws (Kyon, Zürich, Switzerland) under fluoroscopic guidance of
elbow congruency. Fixation of the plate was completed by applying one 2.7 mm titanium
screw (Kyon, Zürich, Switzerland) and one 2.7 mm 316 L stainless steel cortical screw
using the proximal drill guide ([Fig. 5C]). Before closure an autologous bone graft was applied around the implant.
Follow-Up
Postoperative radiographs showed a stable construction ([Fig. 4C]). The patient was discharged with a Modified Robert Jones bandage containing a splint.
Bandage changes were performed on a weekly base for 6 weeks. The dog started to use
the thoracic limb during the ensuing 6-week period of bandage changes and after removal
a gradual return to full weight bearing without lameness occurred and was confirmed
at 15-month follow-up. Radiographs at 6 and 10 weeks after surgery showed a stable
presentation of the metallic implants and congruent elbow with osteoarthrotic changes
similar to previous studies ([Fig. 4D]). Six- and ten-weeks postoperative force plate analysis was rechecked. At 6 weeks
Fzmax of the left thoracic limb was 3.5 N/kg and Fzmax of the right thoracic limb was 9 N/kg. Measurements were repeated at 10 weeks postoperatively
and showed an increase in Fzmax of the left thoracic limb of 5.5 N/kg and a Fzmax of the right thoracic limb of 8 N/kg.
Discussion
This report described the use and feasibility of additive titanium manufacturing to
create a 3D patient-specific implant for repair of atrophic non-unions with segmental
antebrachial bone defects in dogs with incorporating a polished joint segment in one
case. Both patients were presented with atrophic non-unions in close vicinity of the
elbow joint as a result of prior unsuccessful osteosyntheses. The alternative surgical
options for these cases were either limb amputation or the use of fixed size off-the-shelf
implants. The latter were not considered optimal due to the absence of a complete
articular surface or accurate anatomical reconstruction which would result in severe
morbidity.[38] Reconstruction of the ulna, and in one case the joint surface of the elbow, with
the 3D patient-specific implants, led to functional limb use and excellent clinical
outcome in both cases.
Both implants were made of porous titanium scaffold with a hollow centre intended
to allow bone ingrowth and therefore permanent osseointegration of the implant in
the ulna. However, in case 1 bone ingrowth was seen macroscopically around the 3D-printed
scaffold but on radiographs a radiolucent line remained around the entire implant.
Whether the radiolucent line was caused by failure of the implant to fully osseointegrate
or by the Uberschwinger artefact (radiolucent halo around metal when there is a large
density difference between adjacent objects) could not be determined on the radiographs.
Uberschwinger artefact may simulate loosening of orthopaedic devices.[33]
[34]
[35]
Postmortem histology of the bone specimen revealed new bone formation in the medullary
cavity of the porous segment, but absence of osseointegration in the smaller porous
segment at the bone–implant interface of the scaffold in the ulnar bone. In previous
experimental studies micromovements between bone and implant inhibited bony ingrowth
and led to the development of a fibrous membrane.[39] The strength in implant embedding is increased by bone ingrowth.[40]
[41]
[42]
[43] Li and colleagues concluded that increase in porosity and pore size of titanium
alloy implants have a positive influence on osteoconductive properties.[43] Biofunctionalizing surface treatments like alkali-acid-heat treatment (AlAcH) can
improve the apatite forming ability of AM scaffolds, and have a positive effect on
cell attachment, cell proliferation and osteogenic gene expression. The relationship
between these properties and the bone–implant biomechanics is, however, not trivial.[44]
[45]
[46]
[47]
In the second case with a chronic non-union and a segmental bone defect extending
into the elbow joint, the use of an interlocking nail, hybrid external fixator or
a trans articular external fixator, plate fixation or a combination of these techniques
would not have resulted in a clinical anatomic union. The use of a porous 3D patient-specific
implant combined with a polished joint segment and a plate on the tension side of
the ulna resulted in a stable construction.
In both cases, the dogs started to use the limb again and full weight bearing was
achieved with good clinical function. However, in case 1 the bone bridging implant
was not connected to the plates that acted as buttress plates, making them probably
more susceptible to mechanical failure. This resulted in multiple failures of plate
screws which necessitated plate and screw removal at 5 and 11 months after surgery.
At the time of plate removal, it was possible to macroscopically assess osseointegration
of the porous implant and this was found to be a solid component of the ulna and micro-CT
at 24 months confirmed bone bridging through the central hole in the titanium scaffold.
At that time bone bridging was not complete caudal to the implant which may be due
to remaining micromotion on the tension site of the ulna or simply because the dog
did not survive long enough because of euthanasia for multiple concurrent orthopaedic
abnormalities.
Ongoing insight in patient-specific implants and their application together with experience
of the implant failures in case 1 resulted in adaptations of the design of the implant
in case 2. The modification existed of incorporating a 3D-printed 2.7 mm titanium
plate on the caudal site of the porous titanium implant as well as the design of two
screw holes through the porous implant itself. The porous scaffold between the bone
parts contained one central hole for application of a cancellous bone graft and two
screw holes so the implant itself was also fixated to the bone but was also an integral
part with the caudal plate that was additionally fixated with screws to the ulna.
The incorporation of a plate on the caudal side of the scaffold has added stability
to the construct and the design likely reduced stress. In addition, in case 2 drill
guides were designed pre-surgically for screw placement which allowed precise selection
of screw direction and screw length in relation to the bone stock avoiding the joint
surface of the elbow. This may have led to distribution of scaffold and plate fixation
forces, again reducing stress shielding, and therefore help bone and implant survival.
Force plate analysis was performed in case 2 and showed preoperatively overcompensation
of the right thoracic limb. Postoperatively at 6 weeks there was a noticeable difference
in Fzmax between both the left and the right thoracic limbs, whereas at 10 weeks the absolute
difference in Fzmax between both thoracic limbs decreased. One can conclude that these results indicate
a clinical improvement of the left thoracic limb lameness. However, a force plate
analysis is a snapshot of a particular moment in time and provides little information
about the use of the limb during the entire day, e.g., while playing, lying down and
standing. Because the limited value of force plate gait analysis to evaluate the complete
function of the dog it is not routinely used in clinical practice; however, in this
case the gait analysis objectively confirmed improvement in the functional use of
the limb from preoperative to postoperative.[48]
The patient-specific nature of the implants may cause an inherent delay in the availability
of implants, and this should be considered in the surgical planning and when addressing
the owner's expectations. In these two cases, the lead time for the implants between
imaging and surgery was approximately 3 months and is due to time needed for design,
feedback to the surgeon and adapting the design, manufacturing of the implant, post-processing
and quality control, transportation and sterilization procedures. Protocols and algorithms
can be written to have a faster design process.[49]
This report has several limitations. This is a description of two cases with similar
lesions in nature and ongoing insight and experiences with case 1 have changed the
design for case 2. Imaging at follow-up is not uniform. Minimal requirements for follow-up
are radiography to monitor implant position and implant failure; however, CT is considered
of added benefit to assess bone in-growth in the titanium scaffold. For more detail
on bone ingrowth, micro-CT and histology are recommended when bone specimens become
available. It is concluded that the use of additive titanium manufacturing to create
an implant with a polished joint surface to restore the elbow joint, to fill the segmental
defect to stabilize the non-union and to promote osseointegration through the hollow
centre is a novel technique and may be used to treat segmental bone defects in dogs.
