# Novel design of additive manufactured hollow porous implants

Dental Materials, 2020-11-01, Volume 36, Issue 11, Pages 1437-1451, Copyright © 2020 The Academy of Dental Materials

## Highlights

• The properties and performance of porous implants were examined by in vitro mechanical test, in vivo micro-CT and ex vivo pull out test.

• The results prove that the proposed porous implants can provide good fixation and improve the osseointegration.

• Several features were accurately predicted by the three-dimensional mechano-regulation model developed in this work.

• The proposed FE model may reduce the demand of animal tests and is sufficiently general for different dental implants.

## Objective

Our aim is to examine the mechanical properties of two types of additive manufactured hollow porous dental implants and 6 and 12-week bone ingrowth after insertion in animals. A 3D numerical model is also developed to show detailed tissue differentiation and to provide design guidelines for implants.

## Methods

The two porous and a commercial dental implant were studied by series of in vitro mechanical tests (three-point bending, torsional, screwing torque, and sawbone pull-out tests). They also evaluated by in vivo animal tests (micro-CT analysis) and ex vivo pull-out tests. Moreover, the mechano-regulation algorithm was implemented by the 3D finite element model to predict the history of tissue differentiation around the implants.

## Results

The results showed that the two porous implants can significantly improve osseointegration after 12-week bone healing. This resulted in good fixation and stability of implants, giving very high maximum pull-out strength 413.1 N and 493.2 N, compared to 245.7 N for the commercial implant. Also, several features were accurately predicted by the mechano-regulation model, such as transversely connected bone formation, and bone resorption occurred in the middle of implants.

## Significance

Systematic studies on dental implants with multiple approaches, including new design, mechanical tests, animal tests, and numerical modeling, were performed. Two hollow porous implants significantly improved bone ingrowth compared with commercial implants, while maintaining mechanical strength. Also, the numerical model was verified by animal tests. It improved the efficiency of design and reduce the demand for animal sacrifice.

## Introduction

In recent years, dental implants have become an important surgical component. Commercial implants usually are made by titanium alloys because of their excellent chemical stability and biocompatibility. However, the mismatch of stiffness between bone and titanium implant (the so-called stress shielding effect) may result in bone loss adjacent to implants and lead to the increases of late failure of reconstruction [ , ]. The situation becomes even worse for osteoporotic patients since their bones are typically too fragile to afford another surgery [ ]. The stability of implants is the key to the success of surgeries, which is significantly influenced by the design of their geometry, such as thread shape, inclination, and healing chamber shape. Calì et al. [ ] examined the influence of rectangular and trapezoidal thread shapes with different inclination on the distribution of Von-Mises stress by using finite element method, showing that thread shape and inclination can reduce stress peaks in trabecular bones, and the dental implant stability at the initial osseointegration stage can be improved by thread inclination. Zanetti et al. [ ] adopted modal analysis to correlate the resonance frequencies of different dental implant designs and the resulting osseointegration level, showing that the analysis is useful for examining tapered implants and early stage of bone maturation. Beutel et al. [ ] discovered that an implant with a trapezoidal healing chamber has the best osseointegration in the cases they considered.

Another promising solution to improve the stability of implants is porous implants. They can reduce the effect of stiffness-mismatch and improve the interfaces-roughness between implant and bone, resulting in stronger biomechanical characteristics when compared with traditional screws [ ]. There are various fabrication methods for porous metallic biomaterials [ , ], such as powder sintering [ , ], freeze casting [ , ], space-holder method [ ], surface modification [ ] and additive manufacturing [ ]. Many successful porous structures in dental implants reported by the different literatures. Weber et al. [ ] studied the hollow-cylinder (HC) and hollow-screw (HS) ITI dental implants with transverse-opened macropores, showing good osseointegration. Telleman et al. [ ] further performed a 10-year long term evaluation of these ITI dental implants, giving promising results for porous dental implants. The size, morphology, distribution, porosity, and interconnectivity of the pores directly influence the biocompatibility [ ] and the overall mechanical properties of the porous implants [ , ]. It has been reported that the porous size in the range of 100–500 μm results in excellent osseointegration [ ] and the fixation of implants [ ]. Also, continuous and more regular arranged pores showed higher Young’s modulus and strength than randomly arrange pores [ ]. Moreover, with suitable control in a certain range of porosity and pore size, the porous structure could form a strong adhesion between bones and implants, which can enhance the osseointegration of bone tissue [ ].

Stubinger et al. [ ] and Traini et al. [ ] developed implants with complicated porous structure by using a direct laser metal sintering technique. In recent years, additive manufacturing has gained attention for fabricating porous implants with greater complexity [ ]. This kind of porous titanium scaffold allowed osteoblasts to maintain their metabolic activity on the surface of the scaffold [ ]. The pore structures enable the implants to have better elastic adaptability to the bone. This minimizes the stress shielding effects resulting in better osseointegration between bones and the implant. In some works, the fabricated porous scaffolds were also coated with HA/pDA (polydopamine-assisted hydroxyapatite) to enhance the attachment and proliferation of cells on the scaffold surface [ ].

Finite element method (FEM) has been adopted to study orthopedic biomechanics, to predict the deformation and stress in human bones under functional loadings, and to provide design guidelines for implants. In the considered structures, FEM can reveal the distribution of stresses and displacements subjected to the loads. It allows the complexity of modeling of implants, and thus, the designs can be iterated and optimized efficiently even before implants are manufactured [ ]. Furthermore, several algorithms based on biomechanics or fluid dynamics can be implemented in FEM, and have been applied extensively in orthopedics and dentistry. The results provide insights on implant design, bone remodeling, fracture healing, mechanical properties of biomedical coatings on implants, and the interactions at the bone-implant interfaces [ ].

The extracellular microenvironment plays an important role in controlling the behavior and differentiation of tissues and bones [ ]. Lin et al. [ ] adopted the adaptive strain energy density algorithm to study the preliminary bone remodeling around the implant under the different loading conditions and based on implant position. Their calculations suggested that reducing the lateral occlusal force can enhance bone growth around the implant. Desai et al. [ ] used a two-dimensional finite element analysis to reveal the stress and strain contour of several implants and obtained the optimal design of the implant. Chou and Müftü [ ] studied the bone formation and healing pattern after the dental implant placement using a two-dimensional finite element model. In the study of Yamanishi et al. [ ], the neck design of the implants subjected to an oblique loading was evaluated by a three-dimensional finite element analysis. Recently, Mukherjee and Gupta [ ] adopted the mechano-regulatory algorithm to predict the tissue differentiation around a porous acetabular component and to investigate the bone ingrowth due to the surface design of the implants. Thus, computational analyses not only give an insight into the progressive tissue differentiation but also enhances the efficiency of the design process of implants.

The studies mentioned above provided profound knowledge for the design of porous implants through either careful experiments or numerical calculations. However, a systematic study on dental implants with multiple approaches, including new design, mechanical tests, animal tests, and numerical modeling simultaneously is still desirable. In the current work, two types of novel internal fixation dental implants (Type I: circular porous structure and Type II: biomimetic mesh network structure) with complicated porous scaffold and hollow structure are manufactured by additive manufacturing technology, as shown in Fig. 1 . They were designed to improve the overall mechanical properties of implants, bone-implant contact areas, and bonding strength. The implants were evaluated by series of in vitro mechanical tests, in vivo animal tests, and computational tissue differentiation analyses.

(a)–(c) are the geometries of Type I, Type II, and Control implants. (d)–(f) and (g)–(i) are the SEM images of the surface of Type I, Type II, and Control implants at 100× and 250×.

## Hollow porous implants design and manufacturing

Two types of newly-designed implants were proposed here and manufactured by Biomedical Technology and Device Research Laboratories, Industrial Technology Research Institute (ITRI), Taiwan. The geometry of the implants has an average diameter and total height of 4.2 mm and 21.5 mm, respectively, as shown in Fig. 1 (a)–(c). The shape of the implant body is analog to a commercial dental implant Superline FX4514SWC (Dentium, USA), which is tapered for a good success rate in immediate loading cases [ ] and with a flat end design for reducing bone destruction and perforation risk. Type I implant adopts a circular porous scaffold structure with pore size in the range of 50−200 μm. Type II implant adopts a biomimetic mesh network structure which is analog to the trabeculated bone tissues in cancellous bone, with pore size in the range of 300−500 μm. Based on the study of Li et al. [ ], titanium specimen with a porosity of 60–70% processed similar mechanical behavior with human cancellous bone. However, implants with such high porosity resulted in poor quality in additive manufacturing, leading to early failure in the following mechanical tests. Thus, we thickened the frame width around the pores for both implants, giving relatively lower porosity of Type I and Type II implants of 36% and 55%, respectively. The thread design of Type I and Type II implants is similar to a commercial implant Acutrak standard (Acumed, Hillsboro, Oregon, USA), with relatively wider spacing between each thread compared with it in Superline implants and with cutting flutes through the threads. This allows greater exposure area between bones and the pores on the surface of the implant body. In order to examine the performance of the two hollow porous implants, a non-porous implant that is identical to a commercial dental implant Superline FX4514SWC was manufactured by additive manufacturing technique and served as the control group (known as “Control” in the follows).

Three implants were manufacturing from Ti-6Al-4 V powder by using the ITRI additive manufacturing laser powder bed fusion technology (ITRI-AM100, Tainan, Taiwan). All the implants were manufactured by using a beam power of 170 W with the laser scanning speed of 1250 mm/s. The temperature was increased by the thermal effect of the laser, leading to the bonding of powder. The layer thickness of the printed process was 20−30 μm. The surface morphology was observed by scanning electron microscopy (SEM, HITACHI S-3400, Tokyo, Japan), as shown in Fig. 1 (d)–(i). The pores were distributed uniformly over the surface. The 3D printer showed a reliable accuracy and the implants were well-produced. Four standard tests were then performed to analyze the mechanical properties of these implants.

## Three-point bending tests

The Instron ElectroPuls® E3000-LT Series Test Instrument (INSTRON, Canton, MA, USA) was used for the three-point bending test. The implants were supported by two holders with a 10 mm span length. A stainless-steel loading device with a tip, which can be regarded as a cylinder with 1 mm diameter, was located over the center of the implant. A small preload of 5 N was initially applied to stabilize the implants. Then, the force was applied with a loading rate of 0.5 mm/s under the displacement control mode. The force-displacement response of the implants was measured and illustrated by the computer monitor of the testing machine with an error of ±5%.

## Torsional tests

The torsional tests were done by SE MODEL 2205NS (SE Test system Co., LTD., Taiwan) torque tester. Both ends of each implant were mounted with a resilient sleeve and fixed at the thread site at 20% of the implant length away from the top surface. The torsional load was applied to the implant at a constant speed of 3 rpm with a load limit of 1 kgf until the failure was reached. Three sets of torsional properties were obtained from external and internal rotations of the actuator after a preconditioning cycle. The mean values were then calculated.

## Screwing torque tests

The sawbones (Pacific Research Laboratories Inc, Vashon Island, WA, USA) with 43 mm × 43 mm × 40 mm and a density of 20 pcf was used as the test bone. A hole was pre-drilled on the sawbones for implant insertion by using a tap with a diameter of 2.8 mm. A torsional load was applied to the implant at a constant speed of 3 rpm with a load limit of 1 kgf. The torque tests were carried out for 4.5 laps with clockwise and counter-clockwise. The mean values were obtained from five samples in each case.

## Pull-out tests

The pull-out strength was tested by HT-2402EC (Hung Ta Instrument LTD., Taichung, Taiwan). The implants were inserted in the sawbones under the same conditions as described in the screwing torque test. The specimen was then placed on a specially designed fixture with a self-aligning function to ensure a vertically oriented pull-out test. A pre-load of 40 N with the accommodation time of 30 s was applied, followed by a progressive load at a speed of 5 mm/min. The failure pull-out strength is obtained based on the first peak force in the force-displacement diagram. The stiffness of each type of implants was evaluated by the slope of the linear region in the force-displacement diagram. The energy absorption before failure was determined by the area under the force-displacement curve before the failure.

## The three-dimensional mechano-regulatory tissue differentiation model

The numerical model was developed in the current work to predict the bone ingrowth and tissue differentiation around the implants. The results can establish a direct comparison with those obtained in the animal test, such as micro-CT analysis. More detailed information on the healing pattern in the region of interest can also be predicted numerically. It is worth noting that the implants considered in the current work consisted of porosity, and thus three-dimensional finite element calculation was needed. The strategy taken here was a two-stage simulation. At the first stage, a mechanical calculation is undergone for the entire bone and implant system, Fig. 2 (a). Then, the resulting displacements extracted from the previous simulation were applied to the selected part of the model, and the mechano-regulatory algorithm was applied to study the healing pattern. The selected part was one-eighth of the implant and its surrounding bones located at the bottom half of the system, as shown in Fig. 2 (b), where several features of bone formation patterns were shown in the animal tests.

The three-dimensional finite element model of the implant and the surrounding bones for (a) the first-stage mechanical simulation and (b) the second stage tissue differentiation simulation. (For interpretation of the references to colour in the text, the reader is referred to the web version of this article.)

The three-dimensional finite element model was built by finite element software ANSYS for the first-stage mechanical calculation. Half of the geometry is shown in Fig. 2 (a) for illustration purposes. The implant was set as Ti6Al4V and the corresponding material properties can be found in the literature [ ]. A thin cylindrical callus region around the implant was set as to where the implantation surgery proceeded. The callus region was filled with granulation tissues, which can later differentiate into fibrous tissues, cartilages, or bones. The top layer of the cylinder was cortical bones, and the remaining part was cancellous bones. The implant, cortical bone, and cancellous bone were then meshed by the SOLID185 element type provided by ANSYS. The material properties of these two types of bone are listed in Table 1 . The region of callus was meshed by pore-pressure mechanical solid elements, type CPT215 provided by ANSYS. Elements of this type can determine both the strain state and the interstitial flow velocity of the tissue due to the applied loads, and thus the environment for cell activity can be simulated at the second stage of the simulation for tissue differentiation. The boundary conditions follow those used in the animal tests done by Duyck et al. [ ]. The top surface of the implant is subjected to a transverse loading of 29.4 N and the surface at the bottom of the cylinder is fixed. The contact settings allow sliding and gap formation between bone and implant. Here, we assume that the coefficient of friction in the current work is 0.3 [ ].

Table 1
Material properties of the tissues used in the current model [ ].
Young’s modulus (MPa) Poisson’s ratio Permeability (

${}^{}$

m

4

/

N

s

)

Granulation tissue 1 0.17 1.0

10 −14

Fibrous tissue 2 0.17 1.0

10 −14

Cartilage 10 0.17 5.0

10 −15

Immature bone 1000 0.30 1.0

10 −13

Mature bone 6000 0.30 3.7

10 −13

Cortical bone 20,000 0.30 1.0

10 −17

Cancellous bone 6000 0.30 3.7

10 −13

Ti-6Al-4V 113,000 0.30 N/A

The model for the second stage calculation applied the displacement based on the displacement generated in the first stage. It is worth noting that according to tests, tissue differentiation is governed only by the vertical component of the load, and thus, only vertical displacement (∼10 μm) was applied at the top surface of the implant. The load induced the distortional strain and the fluid flow in the granulation elements, and thus affected the bone healing and tissue differentiation. The tissue differentiation was based on the mechano-regulatory algorithm and computational procedures proposed by Lacroix and Prendergast [ ] and governed by biophysical stimulus ( S ). It can be defined as

S = γ/ a + ν / b

where a = 0.0375 and b =3 μm/s are empirical constants; γ is the octahedral shear strain; ν is the interstitial flow velocity. The biophysical stimulus of each element determined the cell type of that element at the next time step. The tissue types that granulation can transform into were fibrous tissue, cartilage, immature bone, and mature bone. The corresponding range of biophysical stimulus and the material properties for these tissue types are listed in Table 2 .

Table 2
The ranges of cell differentiation factor for different phenotypes of the cell.
S Phenotype
3.000 < S Fibrous tissue
1.000 < S

3.000

Cartilage
0.266 < S

1.000

Immature bone
0.010 < S

0.266

Mature bone
S

0.010

Initial resorption

The material settings for the next step were related to the mesenchymal stem cell migration [ ], which can be regarded as the level of the transition from granulation tissue to the other tissue types. The stem cell migration was modeled by the classical diffusion equation

$\frac{}{}$

d

n

d

t

=

D

2

n

where n is the concentration of mesenchymal stem cells, t is time and D is diffusion constant. The coefficient D was determined based on a pre-set healing time at which all the granulation elements reach the maximal stem cell concentration. Here, the healing time was set for 12 weeks to fit the healing time in the animal tests. The callus region located around the implant was where the tissue differentiation occurred. The boundary marked in yellow lines in Fig. 2 (b) is the origin of the cells, which is defined as a source of the diffusion of the mesenchymal stem cells, i.e. n = 100%. Thus, the distribution of stem cell concentration at each time step can be obtained. The effective material property X , such as Young’s modulus, Poisson’s ratio, and permeability, for the next step, was the linear combination between the material properties of the granulation tissue X g and the differentiated tissue X d , such that

n

m

a

x

-

n

n

m

a

x

X

g

+

n

n

m

a

x

X

d

where

${}^{}$

n

m

a

x

is the maximum concentration. In addition, the smoothing procedure was applied by averaging X with the material properties in the previous nine steps to avoid numerical instability and dramatic changes. In this way, the material properties and tissue types transformed from granulation tissue in the callus region around the implant in each time step can be obtained [ ].

## Animal experiments

The surgical procedures and experimental protocols were approved and under supervision by the Institutional Animal Care and Use Committee (IACUC) of Medical College’s Animal Research Committee of the National Taiwan University and all procedures have been carried out in accordance with the Declaration of Helsinki. Eighteen adults (1-year-old) female New Zealand white rabbits, weighing 2.0–2.5 kg, which were used as experimental subjects, feeding with Purina Laboratory Chow (LabDiet, St. Louis, MO, USA) ad libitum and housing in temperature-, humidity-, and light-controlled environment. The rabbits were premedicated with 0.5 mg atropine and then anesthetized using ketamine (50 mg/kg) and Rompun [2-(2,6-xylidino-)-5,6-dihydro-4H-1,3-thiazine hydrochloride] (12 mg/kg). Surgical interventions were performed under strict sterile conditions. All implants were sterilized by radiation before the implantation. The surgical area was shaved and disinfected with iodine, and a straight incision was made to expose the distal femur bone. A hole was pre-drilled on the bones for each implant insertion by using a tap with a diameter of 2.8 mm. Eighteen rabbits were randomly divided into three groups for the test of Control, Type I, and Type II implants, each of which included six rabbits. After the implant insertion, each group was randomly divided into two subgroups (three rabbits each), representing 6 and 12 weeks of follow-up. Note that each rabbit can afford two implants, and thus, we performed 6 replicates in animal tests for each type of implant. At the end of experiments, rabbits were then sacrificed by injection of pentobarbital (200 mg/kg, IV). Samples were dissected and soaked in fixative (6% formaldehyde) and ready for the analyses described as follows.

## Micro-CT analyses

The images of the implant and the surrounding bones were examined by micro-CT (SkyScan 1176, Bruker, Belgium) and reconstructed by CTvox 2.4 software. A cylindrical region around the implant was defined as the region of interest (ROI). Where the bone volume fraction to the total volume of ROI (BV/TV, %) for each case was determined.

## Ex vivo pull-out tests

After micro-CT analyses, an axial pull-out force was applied to the specimen using a material testing machine (MTS 858, MTS Systems, Minnesota, American). The rabbits’ distal femur bones with inserted implants were embedded in plaster and fixed in the pipes. Each implant was extracted from the femur at a rate of 5 mm/min until the failure of the implant. The maximum pull-out strength was defined as the highest load reached before the decrease in strength due to the failure.

## Statistical analyses

Statistical analyses were performed by using SPSS version 16.0 software (SPSS Inc., Chicago, IL, USA). Results from in vitro mechanical tests were expressed as mean ± standard deviation. The unpaired two-tailed student’s t -test was performed for the results of BV/TV obtained from micro-CT analysis, as the groups of Type I and II implants were assumed to be independent and with different standard deviation. The results from the ex vivo pull-out test were statistically analyzed by analysis of variance (ANOVA), where a post-hoc test with Bonferroni correction was performed. Differences were considered as significant as p < 0.05.