Biomaterials 279 (2021) 121206

Contents lists available at ScienceDirect

Biomaterials
journal homepage: www.elsevier.com/locate/biomaterials

High-strength, porous additively manufactured implants with optimized

mechanical osseointegration
Cambre N. Kelly a, Tian Wang b, James Crowley b, Dan Wills b, Matthew H. Pelletier b,
Edward R. Westrick c, Samuel B. Adams d, Ken Gall a, William R. Walsh b, *
a
Pratt School of Engineering, Duke University, Durham, NC, USA
b
Surgical and Orthopaedic Research Laboratories (SORL), Prince of Wales Clinical School UNSW Sydney, Kensington, Australia
c
Orthopedic Surgery, Allegheny General Hospital, Pittsburgh, PA, USA
d
Department of Orthopedic Surgery, Duke University Medical Center, Durham, NC, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Optimization of porous titanium alloy scaffolds designed for orthopedic implants requires balancing mechanical
Additive manufacturing properties and osseointegrative performance. The tradeoff between scaffold porosity and the stiffness/strength
Laser powder bed fusion must be optimized towards the goal to improve long term load sharing while simultaneously promoting
Titanium
osseointegration. Osseointegration into porous titanium implants covering a wide range of porosity (0%–90%)
Osseointegration
Gyroid
and manufactured by laser powder bed fusion (LPBF) was evaluated with an established ovine cortical and
cancellous defect model. Direct apposition and remodeling of woven bone was observed at the implant surface,
as well as bone formation within the interstices of the pores. A linear relationship was observed between the
porosity and benchtop mechanical properties of the scaffolds, while a non-linear relationship was observed
between porosity and the ex vivo cortical bone-implant interfacial shear strength. Our study supports the hy­
pothesis of porosity dependent performance tradeoffs, and establishes generalized relationships between porosity
and performance for design of topological optimized implants for osseointegration. These results are widely
applicable for orthopedic implant design for arthroplasty components, arthrodesis devices such as spinal inter­
body fusion implants, and patient matched implants for treatment of large bone defects.

1. Introduction same porosity [3–6]. Additional topological design parameters such as

the unit cell size, orientation, and even the ratio of the cell size to the
Accelerating and maximizing the osseointegration of an orthopedic overall porous volume impacts mechanical properties of porous scaf­
implant through modulation of implant topology has garnered folds [2,4,7]. However, these topological design factors are all second­
increasing interest in recent years. With the ability to fabricate porous ary to the dominating effect of scaffold porosity which drives the
metallic scaffolds with increasing complexity, laser powder bed fusion mechanical performance of the scaffold (strength and stiffness) and in­
(LPBF) has enabled a paradigm shift in the topological design of fluences osseointegration [2–4].
biomedical implants [1]. The integration of advanced computational By increasing the porosity of the titanium scaffold, the apparent
design tools with LBPF, has unlocked engineering of porous scaffolds modulus can be reduced to within the range of bone [2], however, there
with prescribed properties and patient matched geometries. In fact, is also a correlated reduction in scaffold strength which must be
recent investigations into characterization of mechanical properties of balanced to support implantation and load bearing. Matching of implant
AM titanium scaffolds with varied topologies and porosities show a wide stiffness to that of bone is considered favorable in maximizing load
range of resulting strengths and stiffnesses within the range of bone [2, sharing between the two, and thus stimulating bone formation [8–10].
3]. Recent work has also demonstrated that triply periodic minimal However simply achieving stiffness matching through material selection
surfaces (TPMS) sheet-based architectures, including the gyroid-sheet, is not sufficient to ensure mechanical interlock and successful osseoin­
exhibit favorable mechanical properties, by maintaining high strength tegration, as polyetheretherketone (PEEK) spinal interbody cages have
and fatigue resistance relative to strut-based unit cell architectures of the been shown to result in fibrous encapsulation despite their relatively low

* Corresponding author.
E-mail address: [email protected] (W.R. Walsh).

https://doi.org/10.1016/j.biomaterials.2021.121206
Received 26 May 2021; Received in revised form 4 October 2021; Accepted 20 October 2021
Available online 22 October 2021
0142-9612/© 2021 Elsevier Ltd. All rights reserved.
C.N. Kelly et al. Biomaterials 279 (2021) 121206

modulus within the range of bone [11]. Thus, it is believed that through Benchtop mechanical evaluation was conducted in parallel with an
a combination of stiffness matching, surface interaction, and ovine osseointegration model to determine the tradeoff between the
three-dimensional porous networks that interfacial mechanical interlock implant’s load bearing capabilities and the biomechanics at the bone-
between the implant and newly formed bone is achieved. Thus, opti­ implant interface after 4 and 12 weeks. Histological and histomorpho­
mization of the tradeoff between initial load bearing capacity of the metric evaluations were used to assess the volume and location of bone
implant, and long-term potential for osseointegration and load sharing ingrowth into the porous implants in cortical and cancellous sites. The
must be considered. present results have implications for orthopedic implant design across
A high strength bone-implant interface is critical to successful out­ numerous clinical applications.
comes in many clinical applications including interbody fusion,
arthroplasty fixation, and bridging of large bone defects. Clinical failure 2. Results
can be caused by lack of initial interlock, poor bone integration or
subsequent resorption due to stress shielding, often resulting in revision 2.1. Scaffold design, fabrication, and benchtop mechanical performance
surgery. This is especially challenging in treatment of segmental defects, of titanium scaffolds
such as those of the lower extremity, where a need for a high strength
and fatigue resistant implant must be balanced with the need for bone Titanium implants of varying porosity for mechanical evaluation and
ingrowth over a long distance [12–15]. Currently, treatment of large preclinical implantation were designed with a gyroid-sheet architecture
bone defects is primarily achieved using fibular autograft or cadaver and produced via LPBF of medical grade titanium alloy (Ti6Al4V) (Fig. 1
bone allografts, the latter of which have a 50% failure rate due to A, B). Porosity was systematically controlled by decreasing the wall
nonunion and collapse after at least 16 months [16,17]. Another chal­ thickness of the gyroid sheet. Porosity of the printed implants evaluated
lenge is lack of implant integration with the host bone in arthrodesis by μCT revealed a decrease in as-printed porosity from the idealized
applications, leading to instability and micromotion caused by fibrous CAD model of up to 4% (Table S1). This is attributed to characteristic
encapsulation of the implant [18,19]. Thus, achieving early osseointe­ overinflation of the geometry resulting from the powder bed fusion
gration is critical in the success of fusion procedures. We recently re­ process, which was seen to be the greatest in the highest porosities.
ported the importance of topology on establishment of a stable Surface roughness (Ra) of the implants has previously been evaluated at
bone-implant interface to achieve functional repair of critically sized 7.0 μm [7], and is resultant of the powder bed fusion process, which
defects of the rat femora using gyroid-sheet implants, where the amount produces a surface with partially adherent powder particles as observed
of bone in the most proximal and distal interfaces dictated the torsional in Fig. 1 C-E. When observed at higher magnification, the spherical
strength of the repair [12]. morphology of the titanium powder is seen to create clusters, which
Better understanding of the mechanical and biomechanical tradeoffs provide a topographical texture that is favorable for bone ongrowth as
dictated by the implant porosity must be achieved to optimize implant discussed below [18].
design. In this study, porous titanium implants with gyroid architecture The effective strength and modulus of the scaffolds decreased in a
were designed with porosity varied over a physiological relevant range linear manner under compressive, tensile, and torsional loading with
and produced by LPBF of medical grade titanium alloy (Ti6Al4V). increasing porosity (Fig. 2). Strength of the scaffold increased linearly

Fig. 1. Evaluation of topology and topography of ti­

tanium implants produced via laser powder bed
fusion (A) Micrographs and cross-sectional images
from (B) microCT reconstructions of implants with
increasing porosity produced via LPBF of titanium
alloy. (C–E) Micrographs of implant surface, showing
the topography produced by laser powder bed fusion.
At higher magnitudes, the particles are observed to
create a surface topography with defined by clusters
of particles forming three-dimensional features.

2
C.N. Kelly et al. Biomaterials 279 (2021) 121206

Fig. 2. Benchtop mechanical evaluation of scaffolds with increasing porosity. A linear relationship is observed with increasing porosity for scaffold (A)
modulus and (B) strength under each loading mode. For all trendlines shown R2 > 0.99.

ranging from 15 to 250 MPa with porosity. Across the range of porosity 2.2. Evaluation of bone structure at implant surface and inside the pore
studied, compressive modulus varied across an order of magnitude. network
Shear modulus as observed to be lower than those under axial loading.
This is attributed to a decrease in stiffness at the circumference of the In vivo implantation of implants with increasing porosity was con­
porous gage section in torsion, particularly due to the limited number of ducted in a randomized manner into cancellous and cortical defect sites
unit cell repeats in the sample [20]. All mechanical results are tabulated of adult merino sheep as shown in Fig. 3. We studied the osseointegra­
in Table S1. tion in cortical and cancellous sites and cortical bone-implant interface
mechanical properties over a relevant implant porosity range of 50–90%
by increasing the wall thickness of the gyroid-sheet and compared to a
solid printed titanium implant (Fig. 1A). Histological evaluation at 4-
weeks in cancellous bone sites (Fig. 4) demonstrates early new bone

Fig. 3. Surgical implantation of titanium implants. Dowel implants were placed into one of three sites. (A) The proximal tibia in a press-fit manner, (B) the
cortical bone of the diaphysis of the tibia in a line-to-line manner, (C) the cancellous bone of medial distal femoral condyles in a press-fit manner. (D) Representative
ex-vivo anterior-posterior radiograph showing placement of the implants in the three sites.

3
C.N. Kelly et al. Biomaterials 279 (2021) 121206

Fig. 4. Representative PMMA histology from cancellous sites at 4-weeks and 12-weeks. Limited bone ingrowth into the void space of the porous implants is
observed at both time points. Most of the porous volume is filled with connective tissue in cancellous sites. Note the various geometries of the gyroid implants present
in each representative image are due to location of the slice through the cross section.

deposition on the surface and integration to the titanium implant sur­ surface. However, by 12 weeks, the implants with more void volume
face. By 12-weeks, limited ingrowth into the porous architecture was available had higher relative ingrowth, indicating that the availability of
observed in all porous implant groups embedded in the cancellous bone, void space facilitates neo-bone formation inside the interconnected pore
consistent with previous reports using this model in cancellous bone network.
[21]. Conversely, in the cortical sites, early osseointegration was
observed to at 4 weeks, and was seen to significantly increase by
12-weeks for all groups. Direct surface apposition to the surface 2.3. Implantation of porous scaffolds in sheep model to assess neo-bone
topography of the implants was observed Fig. 5, and no evidence of formation and biomechanics
fibrous encapsulation or inflammatory response was seen histologically
in either cortical or cancellous sites. Evaluation of the biomechanics in the cortical sites after 4 and 12
Histomorphometry analysis of the tissues ingrown into the scaffolds weeks of implantation was conducted using a push-out test as illustrated
in cortical sites at 4 and 12 weeks was conducted to quantify the bone in Fig. 7A and B. In all implant groups, the shear strength significantly
ongrowth for the solid implants and ingrowth into the porous implants increased (p < 0.05) from the earlier to later time point. By 4 weeks,
(Fig. 6). Tabulated histomorphometric results for solid and porous im­ bone ongrowth to implant surface has occurred, and by 12 weeks the
plants are given in Table S2 and Table S3 respectively. Bone Contact deposited bone has increased in quality. This is reflected in the changes
Length over Total Length (BCL/TL) for the solid implants in cortical sites observed in the histology, histomorphometry, and pushout results from
increased from 51% to 63% at 4 and 12 weeks respectively. Bone Vol­ 4 to 12 week. Biomechanical results are tabulated in Table S4.
ume over Total Volume (BV/TV) for all porous implant groups increased Despite higher BIAV values, high porosity implants did not exhibit
from 4 to 12 weeks, as shown in Fig. 6A. To normalize for the higher void higher bone-implant shear strength (Fig. 6C). The maximum force, en­
volume in higher porosity implants, Bone Volume in the Available Void ergy to failure, and shear strength were seen to have a parabolic rela­
(BIAV) was also calculated. At 4 weeks, the BIAV was approximately one tionship with porosity, with a peak between 60 and 70%, the interface
third for all groups regardless of porosity (Fig. 6B). However, by 12 stiffness was seen to have an inverse linear relationship (Fig. 7C–F). At
weeks, differences in bone ingrowth behavior were observed and BIAV both timepoints, the 60% porosity group had the highest average shear
was observed to increase with porosity, ranging from 62% for the 50% strength (13.3 and 33.8 N/mm2 at 4 and 12 weeks respectively). Stiff­
porosity implant, up to 87% BIAV for the 90% implant groups (Fig. 6B). ness and shear strength of the porous implants normalized to that of the
These results indicate that bone ingrowth behavior is both temporal and solid implant at both time points are given in Fig. 7G and H respectively.
porosity dependent. Specifically, at 4 weeks, woven bone deposition A decrease in relative stiffness of the bone-implant interface from 4 to 12
occurred in all groups, and was generally localized to the titanium weeks is seen. The relatively higher stiffness of the 50% and 60% porous
implants at 4 weeks can be attributed to the greater surface area in direct

4
C.N. Kelly et al. Biomaterials 279 (2021) 121206

Fig. 6. Bone ingrowth increases with increasing porosity. (A) Bone volume
(BV) assessed via histomorphometry for implants with increasing porosity. (B)
Normalization of BV by the void volume is represented by bone in the available
void (BIAV). (C) An indirect relationship is observed between BIAV versus shear
strength assessed at 12-weeks.

2.4. Limb salvage in patients with critical sized defects of the lower
extremities
Fig. 5. Representative PMMA histology from cortical sites at 4-weeks and
12-weeks. Progressive bone ingrowth into the void space of the porous implants Early clinical evidence for use of anatomically matched porous ti­
is observed from early the mid time points. At 4-weeks At 12-weeks the porous tanium implants produced by LPBF indicate a significant improvement
volume of most of the porous implants were filled with newly formed bone. upon previous reconstructive options using bulk allograft [14,15,21,22].
Note the various geometries of the gyroid implants present in each represen­ However, reconstructive failure due to implant fracture and inability to
tative image are due to location of the slice through the cross section. achieve implant integration have been reported and is increased in pa­
tients with comorbidities such as diabetes, smoking, or history of
apposition to the cortex, allowing early fixation involved at the inter­ infection [14,15].
face. By 12 weeks, despite the increased bone volume observed in the Two representative case reports demonstrating successful outcomes
higher porosity implants, increased shear strength as not observed. Bone after treatment with patient-specific implants with gyroid lattice for
located in the center of the porous implants does not contribute to shear reconstruction of large segmental defects of the tibia and femur are
resistance, thus the shear strength is dependent on the amount and shown here. In both cases, postoperative radiographic results have
maturity of the bone at the cortical interface. shown bone formation in and around the implant surface. In the case of
the 21-year-old female who sustained an open fracture of the distal tibia
with substantial bone loss, bone growth is observed radiographically as
early as 4 months along the posterior side of the implant (Fig. S1). In the
case of the 60-year-old male treated for fracture of the distal femur, bone

5
C.N. Kelly et al. Biomaterials 279 (2021) 121206

Fig. 7. Biomechanical evaluation via push-out test performed at 4 and 12 weeks after implantation. (A, B) Schematic and images of sectioning and push-out
of implant from cortex. (C–F) Push-out results given at 4 weeks (grey) and 12 weeks (black). Max force, energy to failure, stiffness, and shear strength. Biomechanical
results for implants with increasing porosity normalized by that of the solid implants at the same time point; (G) Normalized stiffness, (H) Normalized shear stress.

formation along the surface of the implant is observed radiographically topologies are hypothesized to enable enhanced bone ingrowth due to
at 12 months (Fig. S2). Given the radiodensity of the titanium implants, the high surface area, permeability, capacity to effectively carry bi­
quantification of bone ingrowth into the porous network of the implants ologics, and potential for the sheet curvature to drive osteogenic cues.
was not possible using X-ray or CT imaging because of the inability to We hypothesized that with increasing porosity, a reduction in modulus
distinguish bone formation from metal artifact. would be observed, and result in improved osseointegration perfor­
mance due to increased void volume for bone ingrowth. The results of
3. Discussion the current study agree with previous reports using the same ovine
model [28–30] and allows for the first time an understanding or the
The present mechanical, preclinical, and clinical results establish the influence of porosity volume fraction on mechanical properties and in
use of porous titanium implants with gyroid-sheet architecture produced vivo performance in a controlled manner.
by AM for treatment of load-bearing bone defects and overall implant Porous scaffolds can be thought of as composites, where the topo­
fixation. Driven by recent evidence that the role of substrate curvature is logical distribution of the material in a void matrix gives varied prop­
important in cuing tissue regeneration [23], gyroid and other erties at the macroscale [3,21,28,31–33]. For bone scaffolds, this means
TPMS-based architectures have received greatly increased interest for modulation of strength and stiffness, by controlling the porosity and
use as tissue engineering scaffolds as they have been shown to have local architecture. Evaluation of porous architectures such as the gyroid-sheet
curvature and stiffness similar to human bone [3,6,24–27]. Gyroid-sheet under torsional and compressive loading is pertinent to its use in

6
C.N. Kelly et al. Biomaterials 279 (2021) 121206

orthopedic implants. Although compressive loads are the dominant gyroid which is advantageous for maximizing bone contact for early
physiological loading mode, bending and torsional loads are also stabilization. At the later time point, bone formation patterns appeared
observed [12,34]. The gyroid-sheet architecture evaluated has previ­ to be guided by the curvature of the implant topology. Given the sinu­
ously been shown to have high strength and stiffness relative to other soidal nature of the gyroid architecture, which has both convex and
porous architectures, for example compared to an octet truss of the same concave surfaces, bone ingrowth into the center of the implant was
porosity [3,6,20,35,36]. The superior compressive strength and energy facilitated. Recent evidence surrounding topology driven bone growth,
absorption is attributed to the continuous nature of the sheets, which and particularly the role of substrate curvature further supports the
serves to more homogeneously distribute load [37]. Further, the sheets design of present implants with gyroid-sheet architecture [31,40].
are self-supporting during the printing process, which results in fewer The remodeling of bone at later time points is dependent on the
geometric imperfections that can drive enhanced local stress concen­ interface mechanics which are themselves dependent on the topology of
trations [3], or that may act as notched stress concentrations leading to the implant, dictated by the structure and porosity. The tradeoff be­
early fatigue failure [4,6]. tween osseointegration, as evaluated by shear strength, and the initial
The moduli and strength of the gyroid-sheet scaffolds under both load bearing capacity, as measured by compressive strength, is not direct
loading modes was seen to be similar and within the range of cancellous (Fig. 8A). Thus, based on this work, a mechanically and biomechanically
and cortical bone [38]. The dependence on porosity was greater than on balanced implant can be designed based on the necessary orthopedic
loading mode, supporting previous studies showing the relative application. However, a limitation of the present study is the uncertainty
isotropic nature of the gyroid architecture [39]. Compressive strength of the in vivo performance at longer endpoints where more significant
ranged over an order of magnitude (30–300 MPa) depending on bone remodeling may occur. Given longer implantation periods, higher
porosity, indicating there is a wide design window for topological porosity scaffolds may indeed reach shear strengths equivalent to lower
optimization of an implant dependent on loading criteria for various porosity implant and provide for improved load sharing. Further, eval­
orthopedic applications. Both compressive strength and moduli were uation of bone remodeling in the adjacent cortex inferior and superior to
reduced from previously reported properties of gyroid-sheet scaffolds, the defect sites, as well as within the implants at later timepoints would
which is attributed to the influence of free boundary effects which be valuable for understanding the role of implant stiffness on long-term
reduce the properties of porous scaffolds when the number of unit cell biomechanics.
repeats is low, as it was in this study [20]. The roles of both surface topography and porous topology are
Early bone formation at in both the cancellous and cortical sites was important in achieving osseointegration of implants. Present gyroid
mediated by the titanium implant’s osteoconductive surface. Direct implants were observed to have higher shear strength at 12 weeks than
apposition on bone to the rough surface of the implant indicates the previously reported values for smooth titanium, as well as plasma spray
roughness resulting from LPBF is appropriate for facilitating attachment coated, or grit blasted + HA coated surface (Fig. 8B) [41,42].
via ongrowth. Previous results from other osteointegration preclinical Non-additive manufacturing methods such as plasma spray or HA
studies have also shown the influence of topography of rough surfaces coating enable rough surfaces or deposition of mineralized coating to­
resulting from PBF of titanium implants [18,19]. The 4-week shear wards improve bone-implant interface strength. However, these pro­
strength of gyroid-sheet implants is greater than previously reported cesses require additional manufacturing steps to deposit the surface
values for porous implants with diamond grid architecture using the coating, which are typically limited by line-of-sight deposition, and
same model [28]. This is attributed to the higher surface area of the vulnerable to delamination from the substrate. Conversely, one of the

Fig. 8. Bone-implant interface biomechanics are

driven by topography and topology, both of
which are highly tunable using 3D Printing
Technology. (A) Shear strength at 4 and 12 weeks
versus compressive strength of empty scaffold. (B)
Shear strength of solid titanium implants with varied
surface roughness, compared with surface porous,
and porous 3D printed implants evaluated at 12-
weeks in an ovine bone defect model. 3D Printing
enables “complexity for free” including the ability to
manufacture implants with high surface roughness
and complex porous architectures to optimize me­
chanical interlock to bone [21,52].

7
C.N. Kelly et al. Biomaterials 279 (2021) 121206

hallmark advantages of AM technologies is “complexity for free”, increasing thickness, the porosity of the resulting scaffold decreased, as
meaning design of surface topography and porous topology are inherent shown in S.I. Table 1. Additionally, solid coupons and implants with no
to the manufacturing. Given this advantage, use of LPBF to manufacture porosity were designed as a control. Coupons and implants were fabri­
porous implants for hip, knee, spine, and foot and ankle indications has cated via LPBF according to ASTM F3001. LPBF was conducted on a 3D
been increasingly adopted [1]. Systems ProX DMP 320, using Ti6Al4V ELI powder under inert argon
In a recent case series reporting outcomes from complex re­ atmosphere. Printing parameters followed previously reported methods
constructions of the foot and ankle using AM titanium implants in high- optimized for gyroid lattices [4]. Following printing, excess powder was
risk patients, 87% were successful, however the failures were due to cleaned from all samples, which then underwent hot isostatic pressing
non-union (6%) and infection (6%) [14]. Although there are some (in accordance with ASTM F3001, 900◦ C at 1000 bar for 2 h) prior to
preliminary preclinical results [12,43], further work in evaluation of the removal from the build plate by wire electrical discharge machining. A
efficacy of such high strength porous implant to support load bearing surface blasting treatment which removes any partially adhered powder,
while acting as a carrier for delivery of osteoinductive or antibiotic followed by passivation in nitric acid, and cleaning was conducted on all
materials would be highly valuable in translation to the clinic. In the coupons and implants. Implants were then sterilized via gamma irradi­
reconstructive case shown in Fig. S2, for example, the patient was pre­ ation prior to surgery. In LBPF, and other additive manufacturing
viously treated for an infection due to the open fracture prior to place­ technologies, deviation of the printed part from the idealized CAD model
ment of the porous titanium implant. While the implants in this is common, particularly when there are features close to the resolution
preclinical study were not packed with biologics, however in clinical use of the printing method or unsupported geometries. Micro-computed
porous titanium implants are typically packed with autogenic, allogenic, tomography (μCT) was used to compare the porosity of the printed
or synthetic bone graft to promote bony fusion [28,44–48]. Similarly, implant to that of the designed CAD model.
treatment of infections with localized antibiotics could be facilitated
through such composited implants, which could be an avenue for future 5.3. Mechanical evaluation
work in this area. Additionally, a growing area of research is investi­
gation of degradable metallics, including Mg, Zn, and Fe, produced by Representative coupons of the solid and various gyroid-sheet po­
LPBF for orthopedic applications [49–51]. These materials may offer an rosities were tested under compressive, tensile, and torsional loading.
advantage to titanium scaffolds, as they can degrade over time and be All mechanical testing was conducted on a calibrated servo-electric
replaced by newly forming bone. testing frame (Test Resources 830, 50 kN load cell) under displace­
ment control. Compressive and tensile loads were applied at an axial
4. Conclusion displacement rate of 1 mm/min, and torsional tests were conducted at a
rate of 1◦ /second. All tests were conducted until sample failure. For each
In this work, demonstration of a linear relationship between scaffold loading mode, peak stress, yield stress, and modulus were calculated
porosity and mechanical performance was observed, whereas a para­ from the stress-strain curves, using and the total cross-sectional area of
bolic relationship with ex vivo pushout strength was seen. All porosities the sample to calculate stress. The highest porosity (90%) specimens for
showed an increase in pushout shear strength from the 4 week–12 week tensile and torsion testing were too fragile for mechanical testing, and
timepoint. The highest pushout strength of the porous titanium gyroid thus no results are reported.
implants at both the time points evaluated in a bicortical defect model
was 60% porosity implants, which exceeded previously reported values 5.4. Ovine bicortical defect model
for titanium implants with plasma spray coated or other modified sur­
faces previous reported in this preclinical model. These results show that Following institutional ethical approval (UNSW ACEC 20/36A) an
while an increase in mechanical interlock strength through osseointe­ established ovine bone ingrowth/ongrowth model and endpoints that
gration can be achieved with porous scaffolds, there is diminishing re­ have been reported for over two decades with various technologies for
turn in both strength and bone ingrowth in scaffolds with porosity osseointegration [21,29,47,52–60] was used for the in-vivo portion of
exceeding 80%. the study. Sample size (n = 5) was based on a power calculation to detect
a 20% difference between groups. Ten skeletally mature adult male
5. Materials and methods sheep (Border Leicester Merino Cross, Ovis Aries, 18 months) underwent
a bilateral procedure with implants placed in a press fit manner (5.5 mm
5.1. Study design hole and 6 mm implant) into the cancellous bone of the distal femur (2
implants per femur) and proximal tibias (1 implant per tibia) well as
The objectives of the study were to establish a relationship between bicortical diaphysis of the tibia (3 implants per tibia) in a line-to-line
mechanical and biological performance of porous titanium scaffolds manner (6 mm hole and 6 mm implant) (Fig. 3). The model provided
produced by AM for use in treatment of bone defects using patient- a total of 6 cancellous and 6 cortical sites per animal.
specific implants. Laser powder bed fusion (LPBF) of medical grade ti­ Pre-operative animal preparation included a clinical veterinary re­
tanium alloy was used to manufacture implants with increasing porosity view as well as routine haematology and biochemistry blood work to
which were evaluated in a bicortical defect model in sheep. Bone- confirm animal status. Fentanyl patches (2–3 μg/kg/hr) (company name
implant interface biomechanics were assessed, along with histological here) were used to provide an opioid analgesic 24 h prior to surgery
evaluation at 4 and 12 weeks. Analogous porous specimens were pro­ [61]. The animals were sedated with xylazine Intramuscular injection,
duced for benchtop mechanical evaluation under tensile, torsional, and (IM)) followed by ketamine (IM) 15 min later. The animals were intu­
compressive loading to determine each’s modulus and strength. bated and maintained throughout the procedure on oxygen and iso­
flurane using an anesthetic machine. All animals received 1 g of
5.2. Design and manufacturing Cefazolin intravenously and Benacillin (Procaine penicillin) 1mL/10 kg
(IM) after induction as antibiotic prophylaxis as well as IV Hartmann’s
As described above, the gyroid sheet architecture is defined by si­ fluid during the procedure via an 18-gauge cannula placed in the ce­
nusoidal functions which can be used to generate a unit cell that can be phalic vein. The fentanyl patches were replaced to provide analgesia
patterned to fill a defined geometric volume. The present coupons and cover for approximately 72 h ([61].
implants were designed based on a repeating gyroid-sheet cubic unit The skin on the medial aspect of tibias and femurs were clipped and
cell, with side length of 6 mm. The porosity of the scaffolds was aseptically prepared for surgery with chlorhexidine gluconate 4% w/v in
modulated by increasing the wall thickness of the sheets, such that with ethanol (96%) and allowed to dry. The animals were transferred to the

8
C.N. Kelly et al. Biomaterials 279 (2021) 121206

operating theatre and placed on the surgical table and positioned supine. assess the amount of bone ingrowth as well as implant material in each
A final spray of 70% providone iodine was applied all over the clipped section.
surgical site and draped for surgery. The cancellous or cortical bone sites
were exposed, and a hole created using a 4.5 mm drill bit for a pilot hole 5.5. Statistical methods
followed by 5.5 mm drill bit in cancellous bone to allow a press fit or a 6
mm drill bit in cortical bone for a line to line fit. Saline hydration was Mechanical data is presented as the average and standard deviation.
used to minimize any thermal damage during drilling. Six different Histomorphometry data was analyzed by determining a single mean
implants (table x) were randomly allocated cortical and cancellous sites. value for each implant site from the multiple sections and pooled for
The subcutaneous and skin were closed in layers using resorbable su­ each group in cortical or cancellous sites for statistical analysis using a
tures. Post-operative pain relief was provided by the fentanyl patch as two-way ANOVA and post-hoc multiple comparisons with p < 0.05
well as Carprofen (Rimadyl) for the first 48 h (subcutaneous injection). chosen for significant using SPSS (IBM Ver 25). The pushout biome­
Animals were carefully monitored throughout the study. Five animals chanical results are presented as average and standard error. A two-way
were euthanized and 4- and 12-weeks after surgery for the endpoints as ANOVA and post-hoc multiple comparisons test with p < 0.05 was used
outlined below and reported in previous work [21,29,47,52–60]. to assess the pushout results.
On the day of euthanasia, general health status, ambulation and
blood work was repeated along with the sedation and anesthetic pro­ Declaration of competing interest
cedure described above prior to lethal injection with an overdose so­
dium pentobarbitone (Lethobarb) via the jugular vein. The right and left The authors declare the following financial interests/personal re­
skin and subcutaneous sites were inspected for any sign of wound lationships which may be considered as potential competing interests:
breakdown or infection. The femur and tibia were harvested intact and Cambre Kelly reports a relationship with restor3d, Inc. that includes:
radiographed in the antero-posterior and lateral planes using a Faxitron equity or stocks. Ken Gall reports a relationship with restor3d, Inc. that
(Faxitron, Wheeling, IL) and digital plates (AGFA CR MD4.0 Cassette). includes: equity or stocks. Samuel B. Adams reports a relationship with
Radiographs in the anteroposterior and lateral views were carefully restor3d, Inc. that includes: equity or stocks.
examined to assess for any adverse bony reactions, and evidence of
radiographic changes at the implant bone interface.
Acknowledgments
The implants placed in the cancellous bone were isolated with a saw,
fixed in cold phosphate-buffered formalin and processing using routine
John Ward, Daniel Pasqualino, John Rawlinson, Greg Mitchell,
polymethylmethacrylate (PMMA) embedding. The bicortical implants in
Rebecca Smith and James O’Connor for technical services.
the tibia were isolated using a saw in the axial plane and sectioned in the
sagittal plane to isolate the medial and lateral specimens for push-out
Appendix A. Supplementary data
testing followed by PMMA hard-tissue histology. Prior to mechanical
testing, the specimens were polished using a Buehler polisher perpen­
Supplementary data to this article can be found online at https://doi.
dicular to the long axis of the implant to remove any periosteal bone
org/10.1016/j.biomaterials.2021.121206.
overgrowth.
A calibrated servohydraulic testing machine (MTS Mini Bionix, MTS
Funding
Systems Inc., Minneapolis, MN, USA) was used with a 25 kN load cell to
perform a standard pushout test of the implants in cortical bone at 0.5
None.
mm/min until the peak load was reached. All samples were fixed in
phosphate buffered formalin and processed for routine PMMA histology.
Author contributions
Peak load, stiffness, and energy to failure were determined by plotting of
the load-deformation curve and calculated using a MATLAB script
Conceptualization: CK, KG, WRW. Investigation: CK, TW, MHP, DW,
(MATLAB R2016a, MathWorks, Natick, MA, USA). The shear stress (Eq.
JC, WRW. Writing – original draft: CK, WRW. Writing – review & edit­
(1)) where σ is the shear stress, c1 and c2 are the cortical thickness on
ing: CK, TW, MHP, DW, JC, TW, KG, WRW.
each side of the implant in the histology section, and d is the implant
diameter. After obtaining cortical thickness values from the PMMA
Data and materials availability
histology.
Load All data are available in the main text or the supplementary
σ= (Equation 1)
(c1 +c
2
2
)πdi materials.

Histology processing included phosphate buffered formalin fixation

References
followed by dehydration in increasing concentrations of ethanol fol­
lowed by methylmethacrylate and polymerization. Embedded cortical [1] L.M. Ricles, J.C. Coburn, M. Di Prima, S.S. Oh, Regulating 3D-printed medical
and cancellous samples were sectioned along the long axis of the im­ products, Sci. Transl. Med. 10 (2018).
plants using a Leica SP 1600 Microtome (Leica Biosystems, Nussloch, [2] C.N. Kelly, A.T. Miller, S.J. Hollister, R.E. Guldberg, K. Gall, Design and
structure–function characterization of 3D printed synthetic porous biomaterials for
Germany). A minimum of two thin (~15–20 μm) sections were cut from tissue engineering, Adv. Healthcare Mater. 7 (2018) 1701095.
sample, etched in acidic ethanol (98 mL ethanol 96% and 2 mL HCl [3] O. Al-Ketan, R. Rowshan, R.K. Abu Al-Rub, Topology-mechanical property
37%) and stained with methylene blue followed by basic fuchsin. relationship of 3D printed strut, skeletal, and sheet based periodic metallic cellular
materials, Addit. Manuf. 19 (2018) 167–183.
The stained slides were reviewed under low magnification to provide [4] C.N. Kelly, et al., Fatigue behavior of as-built selective laser melted titanium
an overview of the section and histomorphometry for bone ingrowth and scaffolds with sheet-based gyroid microarchitecture for bone tissue engineering,
ongrowth. The implant-bone interfaces and local reactions were care­ Acta Biomater. 94 (2019) 610–626.
[5] O. Al-Ketan, R.K. Abu Al-Rub, Multifunctional mechanical metamaterials based on
fully examined at higher magnification for the presence of inflammatory
triply periodic minimal surface lattices, Adv. Eng. Mater. 21 (2019).
cells or local particulate in the cancellous and cortical sites as well as the [6] F.S.L. Bobbert, et al., Additively manufactured metallic porous biomaterials based
bone reaction on as well as withing the porous domains of the gyroid. on minimal surfaces: a unique combination of topological, mechanical, and mass
transport properties, Acta Biomater. 53 (2017) 572–584.
Histology images were used for bone ingrowth and ongrowth histo­
[7] C.N. Kelly, et al., The effect of surface topography and porosity on the tensile
morphometry using validated internal programs (MatLab scripts) to fatigue of 3D printed Ti-6Al-4V fabricated by selective laser melting, Mater. Sci.
Eng. C 98 (2019) 726–736.

9
C.N. Kelly et al. Biomaterials 279 (2021) 121206

[8] J.H. Chen, C. Liu, L. You, C.A. Simmons, Boning up on Wolff’s Law: mechanical [35] L. Zhang, et al., Energy absorption characteristics of metallic triply periodic
regulation of the cells that make and maintain bone, J. Biomech. 43 (2010) minimal surface sheet structures under compressive loading, Addit. Manuf. 23
108–118. (2018) 505–515.
[9] H.M. Frost, Wolff’s Law and bone’s structural adaptations to mechanical usage: an [36] I. Maskery, N.T. Aboulkhair, A.O. Aremu, C.J. Tuck, I.A. Ashcroft, Compressive
overview for clinicians, Angle Orthod. 64 (1994) 175–188. failure modes and energy absorption in additively manufactured double gyroid
[10] J.C. Rice, S.C. Cowin, J.A. Bowman, On the dependence of the elasticity and lattices, Addit. Manuf. 16 (2017) 24–29.
strength of cancellous bone on apparent density, J. Biomech. 21 (1988) 155–168. [37] E. Gawronska, R. Dyja, N. Sczygiol, Numerical simulations of stress distribution in
[11] W.R. Walsh, N. Bertollo, C. Christou, D. Schaffner, R.J. Mobbs, Plasma-sprayed complex structures with various average volume fraction, Proceed. World Congr.
titanium coating to polyetheretherketone improves the bone-implant interface, Eng. Comput. Sci. II (2018).
Spine J. 15 (2015) 1041–1049. [38] W.R. Walsh, N. Guzelsu, Compressive properties of cortical bone: mineral-organic
[12] C.N. Kelly, et al., Functional repair of critically sized femoral defects treated with interfacial bonding, Biomaterials 15 (1994) 137–145.
bioinspired titanium gyroid-sheet scaffolds, J. Mech. Behav. Biomed. Mater. 116 [39] H. Barber, C.N. Kelly, K. Nelson, K. Gall, Compressive anisotropy of sheet and strut
(2021). based porous Ti–6Al–4V scaffolds, J. Mech. Behav. Biomed. Mater. 115 (2021)
[13] J.C. Reichert, et al., The challenge of establishing preclinical models for segmental 104243.
bone defect research, Biomaterials 30 (2009) 2149–2163. [40] A.A. Zadpoor, Mechanics of additively manufactured biomaterials, J. Mech. Behav.
[14] T.J. Dekker, J.R. Steele, A.E. Federer, K.S. Hamid, S.B. Adams Jr., Use of patient- Biomed. Mater. 70 (2017) 1–6.
specific 3D-printed titanium implants for complex foot and ankle limb salvage, [41] M. Svehla, et al., Morphometric and mechanical evaluation of titanium implant
deformity correction, and arthrodesis procedures, Foot Ankle Int. 39 (8) (2018) integration: comparison of five surface structures, J. Biomed. Mater. Res. 51 (2000)
916–921, 1071100718770133. 15–22.
[15] K.S. Hamid, S.G. Parekh, S.B. Adams, Salvage of severe foot and ankle trauma with [42] W.R. Walsh, et al., Bone ongrowth and mechanical fixation of implants in cortical
a 3D printed scaffold, Foot Ankle Int. 37 (2016) 433–439. and cancellous bone, J. Orthop. Surg. Res. 15 (2020) 177.
[16] C.L. Jeng, J.T. Campbell, E.Y. Tang, R.A. Cerrato, M.S. Myerson, Tibiotalocalcaneal [43] J. van der Stok, et al., Full regeneration of segmental bone defects using porous
arthrodesis with bulk femoral head allograft for salvage of large defects in the titanium implants loaded with BMP-2 containing fibrin gels, Eur. Cell. Mater. 29
ankle, Foot Ankle Int. 34 (2013) 1256–1266. (2015) 141–153, discussion 153-144.
[17] W. Wang, K.W.K. Yeung, Bone grafts and biomaterials substitutes for bone defect [44] H. Yuan, et al., Osteoinductive ceramics as a synthetic alternative to autologous
repair: a review, Bioact Mater 2 (2017) 224–247. bone grafting, Proc. Natl. Acad. Sci. U. S. A 107 (2010) 13614–13619.
[18] A. Cheng, et al., Laser sintered porous Ti-6Al-4V implants stimulate vertical bone [45] P. Habibovic, K. de Groot, Osteoinductive biomaterials–properties and relevance in
growth, Ann. Biomed. Eng. 45 (2017) 2025–2035. bone repair, J Tissue Eng Regen Med 1 (2007) 25–32.
[19] A. Cheng, A. Humayun, D.J. Cohen, B.D. Boyan, Z. Schwartz, Additively [46] R. Dimitriou, E. Jones, D. McGonagle, P.V. Giannoudis, Bone regeneration: current
manufactured 3D porous Ti-6Al-4V constructs mimic trabecular bone structure and concepts and future directions, BMC Med. 9 (2011) 66.
regulate osteoblast proliferation, differentiation and local factor production in a [47] D. Chen, et al., Osseointegration of porous titanium implants with and without
porosity and surface roughness dependent manner, Biofabrication 6 (2014), electrochemically deposited DCPD coating in an ovine model, J. Orthop. Surg. Res.
045007. 6 (2011) 56.
[20] A. Pham, C. Kelly, K. Gall, Free boundary effects and representative volume [48] L. Mullen, R.C. Stamp, W.K. Brooks, E. Jones, C.J. Sutcliffe, Selective Laser Melting:
elements in 3D printed Ti–6Al–4V gyroid structures, J. Mater. Res. (2020) 1–9. a regular unit cell approach for the manufacture of porous, titanium, bone in-
[21] W.R. Walsh, et al., Does implantation site influence bone ingrowth into 3D-printed growth constructs, suitable for orthopedic applications, J. Biomed. Mater. Res. B
porous implants? Spine J. 11 (2019) 1885–1898. Appl. Biomater. 89 (2009) 325–334.
[22] S.C. Anne-Marie Pobloth, Hajar Razi, Ansgar Petersen, James C. Weaver, [49] Y. Li, et al., Additively manufactured biodegradable porous iron, Acta Biomater. 77
Katharina Schmidt-Bleek, Markus Windolf, Andras Á. Tatai, Claudia P. Roth, Klaus- (2018) 380–393.
Dieter Schaser, Georg N. Duda, Philipp Schwabe, Mechanobiologically optimized [50] Y. Qin, et al., Additive manufacturing of biodegradable metals: current research
3D titanium-mesh scaffolds enhance bone regeneration in critical segmental status and future perspectives, Acta Biomater. 98 (2019) 3–22.
defects in sheep, Sci. Transl. Med. 10 (2018) 423. [51] Y. Yang, et al., In-situ deposition of apatite layer to protect Mg-based composite
[23] S.J.P. Callens, R.J.C. Uyttendaele, L.E. Fratila-Apachitei, A.A. Zadpoor, Substrate fabricated via laser additive manufacturing, J. Magn. Alloys (2021), https://doi.
curvature as a cue to guide spatiotemporal cell and tissue organization, org/10.1016/j.jma.2021.04.009.
Biomaterials 232 (2019) 119739. [52] M. Svehla, et al., Morphometric and mechanical evaluation of titanium implant
[24] J. Shi, et al., A TPMS-based method for modeling porous scaffolds for bionic bone integration: comparison of five surface structures, J. Biomed. Mater. Res. 51 (2000)
tissue engineering, Sci. Rep. 8 (2018) 7395. 15–22.
[25] D. Mahmoud, M.A. Elbestawi, Selective laser melting of porosity graded lattice [53] M. Svehla, P. Morberg, W. Bruce, B. Zicat, W.R. Walsh, The effect of substrate
structures for bone implants, Int. J. Adv. Manuf. Technol. 100 (2018) 2915–2927. roughness and hydroxyapatite coating thickness on implant shear strength,
[26] S. Vijayavenkataraman, L. Zhang, S. Zhang, J.Y.H. Fuh, W.F. Lu, Triply periodic J. Arthroplasty 17 (2002) 304–311.
minimal surfaces sheet scaffolds for tissue engineering applications: an [54] M. Svehla, P. Morberg, W. Bruce, W.R. Walsh, No effect of a type I collagen gel
optimization approach towards biomimetic scaffold design, ACS Appl. Bio Mater. 1 coating in uncemented implant fixation, J. Biomed. Mater. Res. B Appl. Biomater.
(2) (2018) 259–269. 74 (2005) 423–428.
[27] O. Al-Ketan, D.W. Lee, R. Rowshan, R.K. Abu Al-Rub, Functionally graded and [55] N. Bertollo, et al., Effect of surgical fit on integration of cancellous bone and
multi-morphology sheet TPMS lattices: design, manufacturing, and mechanical implant cortical bone shear strength for a porous titanium, J. Arthroplasty 26
properties, J. Mech. Behav. Biomed. Mater. 102 (2019) 103520. (2011) 1000–1007.
[28] W.R. Walsh, et al., Does implantation site influence bone ingrowth into 3D-printed [56] N. Bertollo, R. Da Assuncao, N.J. Hancock, A. Lau, W.R. Walsh, Influence of
porous implants? Spine J. 19 (11) (2019) 1885–1898. electron beam melting manufactured implants on ingrowth and shear strength in
[29] N. Bertollo, E. Sandrini, P.D. Pria, W.R. Walsh, Osseointegration of multiphase an ovine model, J. Arthroplasty 27 (2012) 1429–1436.
Anodic spark deposition treated porous titanium implants in an ovine model, [57] W.R. Walsh, N. Bertollo, C. Christou, D. Schaffner, R.J. Mobbs, Plasma-sprayed
J. Arthroplasty 30 (2015) 484–488. titanium coating to polyetheretherketone improves the bone-implant interface,
[30] M. Svehla, P. Morberg, W. Bruce, B. Zicat, W.R. Walsh, The effect of substrate Spine J. 15 (2015) 1041–1049.
roughness and hydroxyapatite coating thickness on implant shear strength, [58] W.R. Walsh, et al., The in vivo response to a novel Ti coating compared with
J. Arthroplasty 17 (2002) 304–311. polyether ether ketone: evaluation of the periphery and inner surfaces of an
[31] A.A. Zadpoor, Bone tissue regeneration: the role of scaffold geometry, Biomater. implant, Spine J. 18 (2018) 1231–1240.
Sci. 3 (2015) 231–245. [59] W.R. Walsh, et al., Bone ongrowth and mechanical fixation of implants in cortical
[32] W.R. Walsh, et al., Does implantation site influence bone ingrowth into 3D-printed and cancellous bone, J. Orthop. Surg. Res. 15 (2020) 1–10.
porous implants? Spine J. 19 (2019) 1885–1898. [60] G.C. Causey, et al., In-Vivo response to a novel pillared surface morphology for
[33] A.A. Zadpoor, Mechanical performance of additively manufactured meta- osseointegration in an ovine model, J. Mech. Behav. Biomed. Mater. (2021)
biomaterials, Acta Biomater. 85 (2018) 41–59. 104462.
[34] J. van der Stok, et al., Enhanced bone regeneration of cortical segmental bone [61] C. Christou, R.A. Oliver, J. Rawlinson, W.R. Walsh, Transdermal fentanyl and its
defects using porous titanium scaffolds incorporated with colloidal gelatin gels for use in ovine surgery, Res. Vet. Sci. 100 (2015) 252–256.
time- and dose-controlled delivery of dual growth factors, Tissue Eng. 19 (2013)
2605–2614.

10