Surface wear in a custom manufactured temporomandibular joint prosthesis

Abstract The wear of a novel temporomandibular joint (TMJ) prosthesis was evaluated in an animal model. The prosthesis consisted of an additively manufactured titanium alloy (Ti6Al4V) mandibular condyle and glenoid fossa created through selective laser melting, with a machined vitamin E‐enriched ultra‐high molecular weight polyethylene (UHMWPE) surface attached to the fossa. Thirteen TMJ prosthesis were implanted in sheep, six of which had condylar heads coated with HadSat® diamond‐like carbon (H‐DLC). Euthanasia took place after 288 days, equaling 22 years of human mastication. Linear and volumetric wear analysis of the fossa was performed by optical scanning. The condylar head surfaces were assessed by scanning electron and confocal laser microscopy. The average linear UHMWPE wear, when combined with the coated condyle, was 0.67 ± 0.28 mm (range: 0.34–1.15 mm), not significantly differing (p = .3765, t‐test) from the non‐coated combination average (0.88 ± 0.41 mm; range: 0.28–1.48 mm). The respective mean volumetric wear volumes were 25.29 ± 11.43 mm3 and 45.85 ± 22.01 mm3, not significantly differing (p = .1448, t‐test). Analysis of the coated condylar surface produced a mean Ra of 0.12 ± 0.04 μm and Sa of 0.69 ± 0.07 μm. The non‐coated condylar surface measured a mean Ra of 0.28 ± 0.17 μm and Sa of 2.40 ± 2.08 μm. Both Sa (p = .0083, Mann–Whitney U test) and Ra (p = .0182, Mann–Whitney U test), differed significantly. The prosthesis exhibits acceptable wear resistance and addition of the H‐DLC‐coating significantly improved long‐term condylar surface smoothness.

replacement surgery in a patient's lifetime might be necessary. 4 This is of significant importance, as the expected lifespan of a TJR decrease is inversely correlated to the number of revision surgeries. 5 The rate at which wear appears can be influenced by both material-related factors such as material choice, surface roughness, and the geometry of the articulating surfaces, as well as patient related factors such as the amount of force that is generated and the amount of activity and movement. 6 Wear debris also can lead to foreign-body giant cell reactions, bone resorption, and aseptic implant loosening, contributing to long-term implant failure. 2,[7][8][9] Despite several TMJ systems being available on the market, there is a clear lack of both proper in vivo and in vitro wear analysis. 10,11 This lack of testing is a significant shortcoming, as mechanical properties and wear resistance play a pivotal role in determining the long term outcomes of TJR and, therefore, the need for revision or replacement surgery. 2,7 As far as the authors are aware of, Van Loon et al. 12,13 are the only group to publish their in vitro TMJ TJR wear results, prior to commercial release of their prosthesis. They designed a wear testing machine, which simulated the articulation of the mandibular head against the ultra-high molecular weight polyethylene (UHMWPE) disc, while the implant was submerged in bovine fetal calf serum, diluted with distilled water. The UHMWPE disc was weighed both before and after a 7 million cycle-run, which corresponds to 10 years in vivo functioning, resulting in a wear rate of 0.65 mm 3 /year or linear wear of less than 0.01 mm/ year. 13 While they afterwards also conducted an in vivo sheep experiment, reporting on histological findings of the peri-articular tissues, no evaluation of the amount of wear was reported on. 12 Several more studies evaluated either the histological reaction of the peri-articular tissues or wear pattern in explanted TMJ TJR after, yet as far as we are aware of, this is the first study to report on TMJR wear results through an in vivo experiment. 9,14 While wear can be evaluated via either in vitro or in vivo testing, in vivo testing is preferred for TMJ replacements for at least three reasons. Firstly, there is evidence from hip joint prostheses that in vivo wear rates are much higher than those evaluated by in vitro testing, risking underestimation of the total wear rate. 15 Secondly, the TMJ makes rotational and anteroposterior as well as mediolateral translative movements. Mimicking in vivo scenarios in an in vitro testing environment that captures the specific degrees of freedom in movement that occur during mastication would be extremely difficult.
Thirdly, the amount of force to which the TMJ is subjected remains uncertain, 14,16 which limits the ability to create a reliable in vitro experimental environment. When evaluating potential in vivo animal models, the primate TMJ is most similar to a human's, yet their daily mastication rate is rather low. In addition, several major ethical issues and cost of care prohibit the use of primates for this type of research.
It is for said reason that several different animal models, such as the pig, goat, and sheep model, have been investigated and proven reliable and relatable in vivo experimental models for TMJ investigations.
While having both their advantages (the anatomically and biomechanical resemblance to the human TMJ) and limitations (the more outspoken laterotrusive movements) both the goat and sheep model are considered the "gold standard" in large animals. 17,18 Further, sheep spend on average 4 h per day eating at a rate of 128 mastication cycles per minute and an average of 8-9 h per day ruminating at a rate of 100 cycles per minute. 19 Due to this high daily mastication rate, exceeding that of goats, the total duration of an in vivo evaluation of implant wear can be conducted over a shorter period than in humans or other species.
After the developing of a novel patient-specific additively manufactured (AM, also referred to as 3D-printed) titanium (Ti) alloy TMJ replacement system by CADskills BV, which aims to restore laterotrusive movement through reinsertion and integration of the lateral pterygoid muscle (LPM), a sheep model animal experiment was designed to further investigation. 2,[20][21][22][23] Whereas the proper implant integration and LPM insertion was previously evaluated, 24 this paper aims to evaluate the in vivo wear rate in the condylar and fossa components. Furthermore, the difference in wear between the polished condylar head, coated with a HadSat ® diamond-like carbon (H-DLC) layer, was investigated and compared to that of the non-coated condylar head. In addition, the amount of wear of the fossa composed of a machined Vitamin E-enriched and ɣ-irradiated UHMWPE component articulating with either type of condylar surface was evaluated.
The total evaluation period in the present study was 288 days, which is equivalent to 22 years of human masticatory function. 25 2 | MATERIALS AND METHODS Fourteen ewes (Swifter crossbreed) aged 2-5 years, with an average weight of 73.4 kg (range: 52-86 kg) and without any missing teeth were enrolled in the study. They were allowed to roam freely in the meadow until the operation.

| In vivo test subjects
First, a pilot surgery was performed on two sheep, consisting of a sham surgery with surgical TMJ approach, including opening of the joint capsule, but without condylectomy or prosthesis implantation in one sheep. The second sheep received a TMJ TJR to establish standard procedures before the following twelve sheep were operated.
During the first post-operative week, the sheep were kept in solitary confinement, after which they were put together in a larger indoor pen.

| Implant manufacturing
Six weeks before surgery, each sheep was subjected to a computed tomography (CT) scan of the head. This data was provided to the engineers of CADskills BV in DICOM format. Using the derived standard template library (STL) files, a virtual condylectomy was performed on the left side, from which a total joint prosthesis was designed using Geomagic Freeform Plus (3D Systems). The overall design of the prosthesis was slightly enhanced for animal use after examining a 3D-printed plastic model of the first sheep's skull (Makerbot, MakerBot Industries and Formlabs II, Formlabs). However, the actual prosthetic design, including the number of screws and screw diameters, was devised to be similar to those used in humans. The specific course of the inferior alveolar nerve in the sheep was taken into account for screw length and position in the mandibular stump.
The ramal component was produced in a medical-grade Ti alloy (Ti 6 Al 4 V ELI grade 23) by AM, more specifically selective laser melting (SLM 125 HL, SLM Solutions Group AG). A scaffold structure (500 μm interconnected pores with a diamond unit cell structure) was provided both at the boney interface with the mandible as well as at the condylar neck to provide optimal conditions for boney union and enthesis reconstruction of LPM respectively. A narrow tunnel with a diameter of 2.4-2.5 mm, to accommodate a size 0 suture, was designed in the neck of the condyle (Figure 1). After printing, all condylar heads were first milled to achieve a 0.02 mm accuracy to the "design-STL", after which they are polished using a chalk-based polishing paste. Six of the 13-condylar heads were further treated with a H-DLC coating using the non-disclosed HadSat ® protocol, whereas the other seven condylar heads were left untreated after polishing. The identity of the supplier, as well as the means for applying the H-DLC-coating onto the condylar head surface are proprietary information. The surface roughness of one, non-implanted, coated condyle was determined using a confocal laser microscope (Ra = 0.09 μm, Rt = 0.53 μm) to serve as a comparison for the explanted condyles.  This is of importance, as it evaluates the total amount of debris that is formed and does not just evaluate the deepest point of material loss on the bearing surface.
To determine the amount of linear wear, first a 3D scanner applying blue-light technology (ATOS CORE 135, GOM GmbH,) was used.
The scanner was first calibrated according to the company prescribed calibration procedure, using a type CP40/170 calibration plate. This glass plate has circular markers with several markers having a larger diameter compared to the rest of the markers. These larger markers define the coordinate origin of the panel coordinate system. The 3D coordinates of the central points of each circular marker are measured, as well as distances between certain defined markers. This calibration process was performed and certified by a GOM-employed specialist, resulting in a 13-μm accuracy. However, because this 3D scanner does not allow for evaluation beyond a depth of 1 mm, the linear wear of these samples was recalculated and confirmed using a The "explanted-STL" was then subtracted from the "design-STL" to quantify the volume lost due to wear. Next, the articulating areas of the UHWMPE were isolated and evaluated rather than the entire UHMWPE fossa part. This was done to prevent overestimation of the wear volume, due to the scalpel reduction that was performed during implantation.
Wear volume was calculated using VGSTUDIO MAX Version 3.3.2 (Volume Graphics GmbH). The linear and volumetric UHMWPE one fossa could not be analyzed because the software was not able to retrieve a "best fit" between the design and the scan of the explanted fossa. The error margin in the overlap between the two STL models was too large to provide reliable results due to the intraoperative trimming of UHMWPE in non-load-bearing regions, as well as the posterior UHMWPE ridge being erroneously trimmed down during the post-euthanasia implant retrieval as well as the titanium part for fixation onto the zygoma. (Figure 3) While this does not affect the articulating surface, the difference between the "design-STL" and "explanted-STL" was too significant for the best-fit algorithm, thus resulting in non-cooperative surfaces. One additional fossa could not be analyzed for volumetric wear because the software was unable to provide a "best fit" between both explanted fossa and their 'design-STL', within the margin of error. As a result, a reliable volumetric wear volume could not be determined.
Both the non-coated and coated Ti 6 Al 4 V condylar surfaces were evaluated using a 3D scanner (ATOS CORE 135, GOM GmbH) to   (D) UHMWPE fossa of sheep #7998 that articulated with a non-coated condyle. Black arrow: main worn out UHMWPE volume due to articulating with the condylar surface. Orange arrow: initial, centered, UHMWPE wear volume due to articulating with the condylar surface. Red arrow: UHMWPE creep deformation, in non-articulating region densely concentrated on the non-coated Ti 6 Al 4 V condylar surfaces then on the H-DLC-coated surfaces ( Figure 7B,C). For both types, the surface damage was limited to the load-bearing surface of the condyle. In comparison to the pristine condyle, similar multi-directional scratches were seen on the retrieved coated condyles, indicating that these scratches are due to the polishing protocol that is applied before coating the condyle ( Figure 7A). The amount of surface marks found on the explanted non-coated condyles was markedly higher, indicating that some abrasion had occurred during usage. For a more detailed investigation of the coated surfaces, SEM analysis was performed. This analysis confirmed that in five out of six condyles, multidirectional scratches were present without significant damage to the articular surface ( Figure 8A,B). The condylar surface of ewe #2177 presented deeper marks, for which an additional surface topography analysis using MeX (Alicona Imaging GmbH) was performed, revealing that the surface damage penetrated through the DLC coating ( Figure 8C,D).
The surface roughness of the condylar bearing surface was analyzed using a confocal laser microscope. The 3D as well as 2D surface roughness amplitude parameters are presented in Tables 1 and 4. Overall, these quantitative results indicate that the roughness for the non-coated Ti 6 Al 4 V condylar surface was higher than for the DLC-

| UHMWPE wear analysis
Linear wear, expressed in mm/year, is used in orthopedic surgery to determine the lifecycle of an implant. It does not however determine the total amount of UHWMPE volume that is lost. This is of importance as, along with particle size and shape, the wear volume is a significant determinant for the occurrence of periprosthetic osteolysis. 8 Dumbleton et al. 27 concluded that the risk of osteolysis occurring is rare as long as the total amount of linear wear remains under 0.1 mm/ year. Similar findings were reported by Oparaugo et al., 28    linear wear per year for a total knee implant, our results can be considered excellent. 29 Important to notice is that, while upon inspection, there was a qualitative difference observed between the fossa articulating with either a coated or non-coated condyle, no statistically significant difference was observed between these samples. A Shapiro-Wilk test confirmed the Gaussian distribution of both the linear and volumetric wear data, supporting the use of a t-test, yet post hoc power calculations indicated that this study would have needed 15 sheeps per group to achieve adequate power to detect a significant difference between these two groups of fossa. While the sample size of this study was chosen to minimize the number of animals subjected to, the invasive procedures required for this study it is highly likely that the non-statistical difference that was found was due to the small group sizes.
Secondly, a displacement of the fossa was found in several ewes.
While a 3-month post-operative CT scan revealed a good positioning of the fossa in ewe #7998, during the post-mortem CT scan and dissection a significant caudodorsal displacement of the fossa was seen.
T A B L E 2 Amount of linear ultra-high molecular weight polyethylene wear Notes: For sample 4473, the error margin in the overlap between the two STL models was too large for the "best fit" iterative closest-point algorithm to provide reliable results. Based on the 3D scanner analysis, the linear wear was found exceed 1 mm, yet no specific result was determined. Abbreviation: TMJR: temporomandibular joint replacement. This was also reflected by the wear pattern that was found through 3D scanning of the fossa component. In addition, sheep #5158

T A B L E 3 Amount of volumetric ultra-high molecular weight polyethylene wear
showed a normal positioning of the TMJR at 3 months after surgery, yet a limited latero-inferior displacement of the fossa was found during the post-operative dissection. A similar displacement was found in sheep #2177 at both the 6-month post-operative CT scan that was made as an exception, for a study analyzing the LPM insertion to the TMJR, as well as during explantation. However, as the displacement of the fossa was rather limited, this only led to a slightly more laterally positioned wear volume in case of ewe #5158 and the edges of the wear volume were less sharply marked in case of sheep #2177 ( Figure 6). In addition to these three displaced fossa, also the fossa of ewe #4246 showed a deviant wear pattern, with a slight latero-medial extension of the wear track. This could potentially be caused due to laterotrusive movements of the contralateral joint, with the implanted side functioning as stabilizing joint.
The displacement of these three fossa was most likely due to the use of 2 mm diameter screws for the fixation of the fossa component, as is done in human TMJ TJR. Keeping the higher mastication rate and laterotrusive movement in mind, the force the fossa is subjected to is higher compared to that in humans. This might have led to excessive stress in the bone surrounding the screws, resulting in bone resorption and micromovements between the fossa and the underlying bone, causing aseptic loosening of the implant component. [30][31][32] In order to ascertain the effect of the altered wear patterns and volumes, the results of either only sheep # 7998 or all three sheeps were removed from the results and a renewed statistical evaluation was made. However, the difference in linear and volumetric wear between both groups remained non-significant, and in both cases, the human equivalent for the measured linear and volumetric wear remained well within the acceptable range. Despite the deviant wear pattern for the fourth fossa, we kept these results included, as there was no displacement that occurred.

| Condylar wear analysis
In knee and hip arthroplasty, there is an industry standard for surface  [33][34][35] and can lead to the formation of larger wear particles, which can cause third body wear. 14,36 In this study, the industry standard for total knee prostheses was applied to the TMJ implant surfaces. These surfaces were polished to obtain a Ra below 0.1 μm, which was confirmed by the surface rough-  37,43 In this study, this limitation was addressed by using the patented HadSat ® -coating; no delamination was observed on the surfaces of any of the coated condyles.

| Limitations
In total hip prostheses, the unworn volume of the acetabular component can be reconstructed when conducting a CMM measurement, out of an unworn surface, no such application exists at this moment for reconstruction of the fossa. 7 Thus it would have been preferable to scan the pre-wear UHMWPE component of the fossa before implantation, to limit any error margin. However due to sterilization issues, it was not achievable to scan the fossa after production. However, this error margin did not significantly affect the UHWMPE fossa part under investigation, as they were oversized 3D-printed and consequently milled down to the original STL file boundaries with a precision of 0.02 mm, as was also the case for the titanium condylar component. In addition, as we were not able to scan the implants prior to implantation, we were unable to predetermine reference points as to use a closed loop information system to overlap the "pre-implantation" STL and "explanted-STL" and instead relied on the "best-fit" method using GOM Inspect (GOM GmbH).
A second limitation we faced, were the fitting difficulties of the UHMWPE fossa during implantation, resulting in the trimming down of the non-load-bearing UHMWPE surfaces. While this allowed for easier implantation, this did result in problems determining the both linear and volumetric wear in one sample and volumetric wear in one additional sample. This was due to the 'best-fit' algorithm no longer A significant limitation we were confronted with as well, was the lack of prior research into both in vitro and in vivo wear analysis in TMJ TJR. Thus, we were forced to compare our results to wear evaluation in total knee replacement (TKR).

| CONCLUSION
Our custom additively manufactured TMJ replacement system is well-

CONFLICT OF INTEREST
Maurice Y. Mommaerts is co-owner and innovation manager at CADSkills BV. Stijn Huys is R&D Officer at CADSkills BV.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. Some data concerning the production processes of the implants can not be shared, as well as some of the medication used during euthanasia, as this is proprietary information. This has also been mentioned in the manuscript.
The data are not publicly available due to privacy or ethical restrictions.