The Porcine Accessory Carpal Bone as a Model for Biologic Joint Replacement for Trapeziometacarpal Osteoarthritis

Abstract

Given its complex shape and relatively small size, the trapezium surface at the trapeziometacarpal (TMC) joint is a particularly attractive target for anatomic biologic joint resurfacing, especially given its propensity to develop osteoarthritis, and the limited and sub-optimal treatment options available. For this to advance to clinical translation, however, an appropriate large animal model is required. In this study, we explored the porcine accessory carpal bone (ACB) as a model for the human trapezium. We characterized ACB anatomy, geometry, joint and tissue-scale mechanics, and composition across multiple donors. We showed that the ACB is similar both in size, and in the saddle shape of the main articulating surface to the human trapezium, and that loads experienced across each joint are similar. Using this information, we then devised a fabrication method and workflow to produce patient-specific tissue-engineered replicas based on CT scans, and showed that when such replicas are implanted orthotopically in an ex vivo model, normal loading is restored. Data from this study establish the porcine ACB as a model system in which to evaluate function of engineered living joint resurfacing strategies.

Graphical Abstract

Introduction

Articular cartilage lines the boney surfaces of joints, and is responsible for transmitting the very high loads associated with activities of daily living.[1] Articular cartilage consists of a dense network of type II collagen fibers organized zonally through the tissue depth as well as negatively charged proteoglycans that repel one another and attract water creating a fluid pressure when the joint is loaded.[2] This combination results in a material well adapted to bearing loads throughout a lifetime of use. However, damage due to wear from aging, disease, or trauma is very common, and given its avascular nature and low cellularity, articular cartilage has a poor endogenous healing response.[3] The societal impact of cartilage damage and degeneration is enormous. It is estimated that over 50 million adults in the United States are living with physician-diagnosed osteoarthritis (OA), and by 2040 this number is projected to reach nearly 80 million.[4] As many as 1 in 10 US adults experience some degree of activity limitation due to arthritis[4] and total costs associated with OA in the United States are more than $120 billion annually.[5]

While most studies of OA focus on the larger joints of the body, including the hip, knee, and shoulder, OA of the thumb is particularly prevalent and adversely impacts quality of life.[6] The trapeziometacarpal (TMC) joint is where the first (thumb) metacarpal articulates on the saddle-shaped cartilage surface of the trapezium. Due to its unique shape, the high mobility of the joint, relatively large forces and small contact areas present, this joint is particularly susceptible to OA.[7–10] A recent study noted that >80% of patients in their 8th and 9th decades had radiographic evidence of TMC arthritis.[11] The hand, and the opposable thumb in particular, are essential to most activities of daily living, and as such the degeneration of this joint significantly impairs quality of life. Hand function is significantly reduced in people with arthritis of the TMC. Compared to the normal TMC joint, the arthritic TMC joint causes decreased thumb strength and range of motion. Grip strength is decreased 14% and pinch strength is decreased 16–22%.[12] Motion is decreased in all planes: 24% flexion-extension, 41% abduction-adduction, and 21% pronation-supination; this results in an overall decrease of 49% in circumduction area.[13]

Patients with TMC OA are initially managed with activity modification, Nonsteroidal Anti-Inflammatory Drugs (NSAIDs), splinting, and corticosteroid injections. Oftentimes these conservative treatments are insufficient, and surgery is required.[9] Over the past 60 years, many procedures have been described, but there continues to be controversy over the optimal technique. One recent review of eight surgical techniques highlighted the limitations of each and concluded that no single method provided superior outcomes.[14] Current techniques remove all or part of the trapezium and often fill this void with tendon, fascia, or an artificial substrate or implant with some including supplemental ligament reconstruction or soft tissue stabilization.[9,14,15] This preserves motion and motor control of the thumb, but at the expense of stability and pinch strength, and, when these procedures fail, they may result in joint subsidence, recurrent pain, and disfigurement of the hand.[16] Alternatively, patients may opt for fusion, which provides stability for pinch strength, but restricts range of motion.[14] Given the high prevalence of this condition and the lack of effective treatment options, there is an urgent need for a better solution.

One such solution may exist in the form of cartilage tissue engineering. Efforts to replace articular cartilage (and bone) with living, functional tissue have matured substantially over the past two decades.[17] In engineered cartilage, chondrocytes (or a chondrogenic cell source such as mesenchymal stem cells (MSCs)) are combined with a scaffolding material to sequester cells and capture accumulated extracellular matrix (ECM), forming a cartilage-like replacement tissue. Hydrogels are attractive biomaterials for cartilage regeneration[18], given that chondrocytes and stem cells can be well dispersed and the cells can take on the round shape and phenotype of a chondrocyte.[19] Furthermore, these gels efficiently entrap the cartilage-like ECM produced by the cells, and can rapidly assemble a neo-cartilage-like matrix with functional properties. [20] In previous work, our group and others have shown that small, cylindrically shaped constructs seeded with chondrocytes or mesenchymal stem cells can be grown in vitro to match the biochemical, histological, and mechanical properties of native tissue.[21–24] In addition, multiple groups consider the integration of tissue engineered cartilage with the underlying bone and much work has been done to engineer the entire osteochondral unit.[25–28]

In addition to marked improvement in the quality of cartilage and bone that can be generated in the lab, the last two decades have witnessed considerable developments in 3D printing and rapid prototyping. Coupling these advances has allowed for marked progress in the production of anatomic 3D tissue engineered constructs.[29–33]. Image-guided tissue engineering is ideal for cartilage engineering, as it allows for precise shape and size-matching for each patient. Work in this area has shown promising results[34–40]. For example, Chang and colleagues developed molds for facial implants that were filled by injection molding with cell-seeded alginate hydrogels.[41] This work was extended to fashion anatomic meniscus implants.[30,42] Hollister and others have developed methods for 3D printing solid porous materials with controlled geometry and pore spacing.[33] Similar methods were implemented by Lee and colleagues to develop fully anatomic replacements for the humeral head in a rabbit model.[35] Since this early and seminal work, numerous other studies have deployed similar technologies to fashion 3D, living, engineered cartilage constructs.[38–40,43]

Due to its relatively small size and the complexity of its load bearing surface, as well as the lack of viable treatment modalities for advanced joint damage, the TMC joint is an attractive candidate for pursuing a biologic joint resurfacing strategy. In order for such an approach to advance to translation, however, an appropriate large animal model is required. To that end, in this study, we explored the porcine accessory carpal bone (ACB)[44] as a model for the human trapezium. We characterized ACB anatomy, geometry, joint and tissue-scale mechanics, and composition across multiple donors. Using this information, we then devised a fabrication method and workflow to produce patient-specific tissue-engineered replicas and developed methods for the implantation of such constructs in situ. Finally, we determined whether these implanted constructs could restore normal loading patterns across the joint. Data from this study establish the porcine ACB as a model system in which to evaluate function of engineered living joint resurfacing strategies.

2. Methods:

2.1. Human and Animal Tissue Acquisition

Fourteen accessory carpal bones were isolated from the forelimbs of adult (15–24 months of age, 60–110 kg) Yucatan minipigs (Sinclair Bioresources, Columbia, MO) from an unrelated study covered under an approved protocol by the Institutional Animal Care and Use Committee at the University of Pennsylvania. Eight of these (right limbs only) were used to assess ACB geometry, and another six to assess cartilage mechanical properties. Six additional limbs were maintained as intact joints and used to evaluate ACB joint mechanics (described below). Four human trapezia were isolated from fresh frozen cadaveric donors, with approval from the University of Pennsylvania committee for cadaveric testing. These included two male specimens (ages 81 and 84) and two female specimens (ages 77 and 84).

2.2. Accessory Carpal μCT and 3D Object Generation

Samples were fixed in formalin and imaged via microcomputed tomography (μCT) before and after equilibration in Lugol’s solution (5% I₂, 10% KI in water) to enhance cartilage contrast (VivaCT 75, Scanco Medical). DICOMs from the initial scan were imported into ITK-SNAP[45] and the bone was segmented in a semi-automated manner. A surface mesh was exported and opened in MeshLab[46] (ISTI, Pisa, Italy), where the mesh was smoothed and simplified. This mesh was imported into Solidworks (Dassault Systèmes, Vélizy-Villacoublay, France) and a 3D object was created in order to compute the bone volume and surface features. Scans post Lugol treatment were manually registered with the bone scan and processed similarly, with the cartilage layer segmented in a semi-automated manner. Cartilage thickness was measured across the 3D object by overlaying a grid with a spacing of 1.25 mm and using the Solidworks thickness analysis tool.

2.3. Histological Analysis of the Accessory Carpal

After μCT imaging, human and porcine samples were decalcified in Formical 2000 for five weeks (with regular changes of fresh solution), and then processed into paraffin. Blocks were sectioned to 10 microns thickness onto glass slides, and then stained with Safranin-O/fast green to visualize cartilage, bone, and fibrous tissue, and with Picrosirius red to visualize collagen. Immunohistochemistry was used to assess distribution of collagen II as in Kim et al.[47] A Nikon Eclipse 90i upright microscope (Nikon Instruments Inc. Melville, NY) equipped with a color camera was used to image each section at 4x and 10x magnification.

2.4. Accessory Carpal Contact Mechanics:

Clinical CT images of the forelimb of skeletally mature Yucatan minipigs were obtained with a portable 8-slice CT scanner (CereTom, Samsung Neurologica) and 3D models of the forefoot were generated with ITKSnap. A three-segment model was assembled using OpenSim[48], which included a proximal forelimb (consisting of the radius, ulna, and radial, intermediate, and ulnar carpal bones), a distal forelimb (consisting of the 1^st-4^th carpal bones and the metacarpals), and an accessory carpal bone. Motion of the proximal and distal forelimb segments were limited to flexion/extension of the wrist, while the accessory carpal bone was afforded six degrees of freedom relative to the rest of the model.

The model had no active force generating muscles, but instead included a single virtual actuator that was capable of prescribing wrist flexion/extension. The Ulnaris Lateralis tendon was represented with three parallel ligament fibers with normalized quadratic force-deformation curves. The ligaments were tuned such that their maximum resting length was achieved when the wrist was in 0 degrees of flexion extension, and an offset was added such that when the ligaments were fully relaxed there was no force at the contact. Wrist flexion led to ligament shortening, resulting in zero force, whereas wrist extension led to ligament lengthening and increasing tensile forces, which ultimately drove the accessory carpal bone into the neighboring ulnar carpal bone.

Contact forces between articulating surfaces were computed using an on-board implementation of a rigid-body-spring-model within OpenSim using a Kelvin-Voigt model:

$$F=kx+b\dot{x}$$

where F is the contact force, k is the spring constant, x is the penetration depth, b is the damping value, and x is the penetration velocity along the surface normal. Values for k, and b were assigned to values of stiffness (1e11) and dissipation (1) within the OpenSim model, which was within the range of a previously published study.[49] Static, dynamic, and viscous friction coefficients were set to be 0.2, 0.2, and 0.01 respectively[50] in order to stabilize motion in the simulation. Contact forces between the accessory and ulnar carpal bones were estimated throughout the course of a controlled wrist extension using forward dynamic techniques. Compressive normal forces, normal to the bone surface, and shear forces, in the medial-lateral and proximal-distal directions, were calculated.

To validate the model, thin film pressure sensors (Tekscan iScan 6900) were used to measure contact mechanics in three adult minipig forelimbs. Sensors were equilibrated and calibrated no more than 24 hours before testing using previously established methods.[51] A 3 cm incision was made between the accessory carpal and ulnar carpal bones and sensors were securely placed into the joint space. The carpus was manually extended through a range of angles from 90 degrees of flexion to full extension in 15-degree increments. Trials were repeated three times and average contact force, area, and pressure were recorded (See Figure 3D).

Figure 3:

(A) Diagram of porcine AC (yellow) and its articulation with the ulnar carpal (blue) and ulna (light blue). (B) μCT rendering with AC identified. (C) Plot of contact forces computed in OpenSim model. (D) Example TekScan pressure map of the AC contact. (E-G) Plots of force, contact area, and stress with respect to flexion angle computed with TekScan. * denotes statistical differences to 45 degrees colored by subject.

2.5. Estimation of Human TMC Contact Forces During Daily Activities

To determine TMC contact forces during activities of daily living, we measured external thumb forces in 4 young-adults (2 male, 2 female) using a Tekscan iScan 6900 sensor. Data collection was approved by the Institutional Review Board of the University of Pennsylvania (#824466) and subjects provided written informed consent. Subjects opened a jar and wrote their name with a pen (3 trials for each condition) with the sensor held between their thumb and the object. TMC contact force was estimated by scaling the external thumb forces by the gear ratio of thumb.[52] (See Supplementary Figure 1)

2.6. Assessment of Cartilage Mechanics across the Accessory Carpal Bone

A custom indentation testing setup[53] was used in conjunction with an Instron 5948 Universal Testing System (Instron Inc, Norwood, MA) to evaluate cartilage mechanics along the midline of the ACB articular surface using a creep indentation test.[54] Lengths of three-quarter inch acrylic tubing were cut to approximately one inch in length, and the bottom was sealed with tape. An accessory carpal bone was placed in each tube with the cartilage surface facing up, and the bone braced against the bottom. A low-melting temperature bismuth alloy was then poured into the tube to support the ACB in the upright position with the cartilage surface exposed. After the metal solidified, the remainder of the tube was filled with phosphate-buffered saline to maintain hydration during testing. The tubes were placed in a custom rig equipped with an XY positioning stage and a goniometer to ensure that the cartilage surface was perpendicular to the indenter. The saddle-shaped articular cartilage surface was indented with a 2 mm diameter spherical indenter in three locations (superior, middle, and inferior). A load of 0.1N was applied at an initial rate of 0.1N/s and held through a displacement creep for 900 seconds after the target load was reached. To measure the cartilage thickness at each testing location, each sample was cut along the midline with a fine-toothed hacksaw and high-resolution photos of the resulting cross section were taken with a Canon SX720HS digital camera. Photos were imported into ImageJ (National Institutes of Health) and the built-in measurement tool was used to determine the cartilage thickness at each indentation location (See Supplemental Figure 2A-B). The material parameters of compressive modulus (E_y−), tensile modulus (E_y+), and zero-strain permeability (k₀) were computed by fitting the creep data to a Hertzian Biphasic analytical model, as in Moore et al.[54]

2.7. Fabrication of Anatomic Replacements for the Accessory Carpal

Implants matching the articulating cartilage surface and first third of the ACB were designed in Solidworks using the μCT data. The implant design included a 2mm thick by 5mm deep keel for fixation. To fabricate implants, positive molds were first 3D printed in an ABS-like photopolymer. To generate elastomeric negative molds, Sylgard 184 (polydimethylsiloxane, PDMS) was prepared at a 10 parts monomer to 1 part curing agent ratio. This solution was poured over the 3D printed designs, degassed, and allowed to cure at 40°C overnight. Two molds were created, with the first including the bone only portion of the implant and the second including the composite bone plus cartilage implant. To fabricate the boney portion of the anatomic implants, poly(ε-caprolactone) (PCL) was dissolved in chloroform at 20% wt/vol and mixed with NaCl crystals, sieved to ~106 μm, as previously described.[55–57] Zirconium nanoparticles were included for radioopacity at a density of 3% wt/vol.[58] The slurry was poured into the mold and the solvent was evaporated overnight. Boney units were demolded, and the salt was leached for two days in a bath of distilled water.

Next, as a proof of concept for the cartilage layer of the anatomic mold, a 5 wt% agarose solution doped with red food coloring for visualization was poured into the composite mold, and the bone component was added to the mold to shape the cartilage portion. In a second iteration, a 1% wt/vol methacrylated hyaluronic acid (MeHA) solution with 20 million juvenile bovine mesenchymal stem cells (MSCs) per mL was dispensed into the composite mold and the bone component placed into the mold to form the final shape of the cartilage. For these HA-based gels, Lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP)[59] was included to enable crosslinking. The MeHA solution was polymerized through the transparent PDMS using UV light for 10 minutes. After 24 hours in culture, construct viability was assessed using the Live Dead Assay kit. Briefly, constructs were incubated in a solution containing calcein AM and ethidium homodimer-1, staining living-cell cytoplasm green and dead-cell nuclei red, respectively. Gels were then imaged on a Nikon A1R Confocal Microscope at 2x magnification.

2.8. Ex Vivo Orthotopic Implantation of Patient-Specific Engineered ACB

To validate the feasibility and of this workflow for anatomic resurfacing, clinical CT images of cadaveric forelimbs from three additional skeletally mature Yucatan minipigs were obtained with a portable 8-slice CT scanner (CereTom, Samsung Neurologica). In each, the ACB was segmented using ITK-SNAP and a 3D model was developed and opened in Solidworks as described above. An axis was defined parallel to the long axis of the bone and through the center of the articulating surface. A plane was then defined orthogonal to this axis and ~3mm from the articulating surface. This reference plane was used to remove the bottom portion of the bone. The bone model was copied and translated 0.5 mm up along the previously defined axis, with the resulting “shell” representing the cartilage portion of the implant. A 2mm thick by 5mm deep fixation keel parallel to the longest axis of the implant was added to the model.

Based on this model, positive and negative molds were prepared as described above and the PCL foam bone integrating portions of each implant were fabricated. An acellular stiff hydrogel cap was then formed using 20% by weight polyethylene glycol diacrylate (PEGDA), whose compressive modulus is comparable to that of native articular cartilage.[60] To form the cartilage cap, the PEGDA solution along with 0.05% LAP as a photoinitiatior was pipetted into each cartilage mold, followed by placement of the corresponding PCL bone construct, and polymerization via UV light for 10 minutes before demolding.

To test whether the implants recapitulated normal load transfer across the ACB, the patient-specific implants were implanted orthotopically. In each of the three forelimbs, a longitudinal skin incision was made at the level of the accessory carpal bone joint. The subcutaneous fascia was bluntly dissected, and the ACB was identified within the flexor carpi ulnaris lateralis tendon. A TekScan pressure sensor was placed into the joint. As described above, the carpus was manually extended through a range of angles from 90 degrees of flexion to full extension in 15-degree increments. Flexion was repeated three times and average contact force and area across the joint surface were recorded.

Next, the original ACB surface was removed. For this, starting proximally and moving distally, the tendon was sharply lifted off the ulnar aspect of the accessory carpal bone to expose approximately 5mm deep to the articular surface. An osteotomy was then performed using a microsagittal saw parallel to the articular surface approximately 3.5mm in thickness. Using a Dremel-powered bur and hand-held curette, a trough was created in the midline of the accessory bone along its long axis to a depth of 5mm. The fabricated biphasic implant (specific to each animal) was implanted by placing the implant keel into the trough in the parent accessory carpal bone, with the base of the boney PCL surface flush with the osteotomized surface. Two 1.0mm compression screws measuring 8mm in length were then inserted perpendicular to the keel to stabilize the PCL implant during testing. The accessory carpal bone was then repositioned into its articulating surfaces and the TekScan pressure sensor was returned to the joint. The joint contact load measurement experiment was then repeated with the carpus manually extended through a range of angles from 90 degrees of flexion to full extension in 15-degree increments. Flexion was repeated three times and average contact force and area across the restored joint surface were recorded. Cross-correlation analysis was used to measure the similarity between the effect of change in angle on force and contact area, before and after anatomic resurfacing of the ACB, where a Pearson correlation coefficient, r_xy, of one indicates a perfect linear correlation between the curves before and after intervention.[61]

2.9. Statistical Analysis

Analyses were performed using GraphPad Prism 7 (GraphPad Software, San Diego, CA). For the loading patterns of the ACB, a two-way analysis of variance was performed with a Bonferroni post-test to determine differences between subjects and across angles, as well as a Tukey’s multiple comparisons test to assess significance between angles for each subject. Correlations between animal weight and ACB volume or cartilage surface area were calculated using a linear regression analysis. For indentation testing data, a D’Agostino-Pearson test indicated that the data was not normally distributed so a Kruskal-Wallis test was performed to compare between testing locations for each of the three material properties reported (E_y−, E_y+, and k₀). Unless otherwise specified, p<0.05 indicated significance. Using MatLab 2020b (Mathworks, Natick, MA), cross-correlation analysis was performed to compute the Pearson correlation coefficient, r_xy, to compare the joint loading response as a function of flexion angle before and after resurfacing of the ACB.

3. Results:

3.1. Anatomy and Histological Appearance of the Porcine ACB relative to the Human TMC.

The porcine ACB was similar in both size and shape of the major articulating surface to the human trapezium (Figure 1). Both possessed a saddle-shaped articulating surface — the ACB where it meets the ulnar carpal, and the trapezium where it articulates with the first metacarpal. Both the human trapezium and porcine accessory carpal bone showed strong staining for proteoglycans on their cartilage surfaces (Figure 2A), and collagen throughout the tissue depth (Figure 2B). Type II collagen staining intensity was high throughout the cartilage for both (Figure 2C).

Figure 1:

(A) Diagram and gross view of the human trapezium and porcine AC bone (in yellow). Dotted line represents sectioning plane. Scale = 3mm. (B) μCT slices in ITK-SNAP showing cartilage segmentation in red and 3D renderings. Scale = 3mm.

Figure 2:

Histology and immunohistochemistry of the human trapezium (left) and porcine accessory carpal bone (right). (A) Safranin-O/fast green. (B) Picrosirius Red. (C). Immunohistochemistry for collagen II. Scale = 3mm (1mm for insets)

3.2. Loading Patterns of the Yucatan ACB

The OpenSim model of the pig forelimb was used to predict the motion and contact forces of the Yucatan ACB joint. The model (Figure 3C) demonstrated that as the unloaded hoof extends, the ulnaris lateralis tendon embedding the ACB goes into tension and causes the ACB to contact and articulate with the ulnar carpal and ulna. The contact force remained ~0N as the carpus was extended, until the joint reached ~20° flexion. After this point, force increased rapidly and reached a peak of 67N at full extension. Similarly, in the ex vivo experiment (Figure 3D), force across the joint remained close to 0 N until 15° of flexion, and then rose rapidly reaching its maximum of 29 ± 11N at 0°. Contact area followed the same general pattern, reaching 17 ± 11 mm² at 15° and 35 ± 7mm² at 0°. Contact stress showed a similar pattern, remaining flat and close to zero throughout much of the range of motion, before rapidly increasing to 0.8 ± 0.3 MPa at 0° (Figure 3E-G). Two-way analysis of variance indicated that the majority of the variation in this experiment (87%, 87%, and 90% for force, contact area, and stress respectively) was due to the angle. However, for both force and contact area, variation due to subject number was significant. Bonferroni post-hoc tests revealed that, at 0 degrees, pig 1 and pig 3 had significantly different force results. Contact areas between pigs 1 and 2 and 1 and 3 were significant at both 15 degrees and 0 degrees. Additionally, the contact area at 15 degrees was significantly different between pigs 2 and 3. Finally, the contact stresses calculated at 0 degrees were significantly different between pigs 1 and 2 and pigs 2 and 3.

3.3. Morphology and Mechanics, of the Yucatan ACB

Across the eight donors, the ACB had the same basic shape and geometry, but showed a high degree of inter-subject variation in both shape (Figure 4A) and size (Figure 4B). Neither ACB volume (Pearson r=0.014) nor cartilage surface area (Pearson r=−0.257) had a significant correlation with animal weight (Figure 4B).

Figure 4:

(A) Solidworks models of eight accessory carpal bones and their corresponding cartilage surfaces (in red). Detail at left. Scale = 5mm. (B) Correlation analysis of animal weight and cartilage volume and surface area for eight skeletally mature male castrated pigs.

The average thickness of the ACB articular cartilage ranged from 350–500 μm within the contour of the main articulating surface, trending towards thicker in the middle of the surface (Figure 5A). There was more variation in the size and shape of the cartilage surface than in the thickness (Figure 5B).

Figure 5:

(A) Average (left) and standard deviation (right) thickness maps of the main articulating surface overlaid on top of a profile indicating the average cartilage perimeter (black line). Indentation test location indicated by red circles. Superior, Inferior, Medial, and Lateral aspects are labeled as S, I, M, and L respectively. In situ the superior label is proximal and inferior is distal in the forelimb. (C-E) Compressive modulus (E_y−), tensile modulus (E_y+), and permeability (k₀) of the articular cartilage along the midline ACB (red circles in A). N = 6, Error bars = mean + SEM

The cartilage mechanical testing data was fit by the Hertzian Biphasic Theory analytical model very well, with an average R² of 0.997±0.002. The compressive modulus in the superior, middle, and inferior regions was 1.4 ± 1.0, 1.4 ± 0.8, and 1.3 ± 0.8 MPa, respectively. The tensile modulus in the superior, middle, and inferior regions was 6.2 ± 5.2, 3.9 ± 3.9, and 4.9 ± 3.1 MPa respectively. The zero-strain permeability in the superior, middle, and inferior regions was 2.1×10-³ ± 2.2×10⁻³, 2.2×10⁻³ ± 1.1×10⁻³, and 1.6×10⁻³ ± 1.8×10⁻³ mm⁴/Ns respectively. No region was significantly different from any other region for any of these parameters (Figure 5C-E). Thickness measures taken at each of the testing locations agreed with the thickness maps produced from μCT data (Supplementary Figure 2).

3.4. Anatomic Molds of the ACB:

Using anatomic renderings generated from μCT scans, an implant was designed in silico to replace the articular surface of the porcine ACB (Figure 6A). A fabrication process (Figure 6B) was developed, consisting of first 3D printing a positive mold, allowing for a negative to be cast from a solvent resistant elastomer (PDMS) (Figure 6C). A slurry of PCL dissolved in chloroform with a suspension of NaCl crystals sieved to 106 μm was then poured into this mold, and the chloroform was allowed to evaporate. Demolding 24 hours later and soaking in distilled water for 24 hours resulted in a porous PCL implant shaped like the original bone (Figure 6D). Next, a photocrosslinkable hydrogel solution (methacrylated HA, MeHA) was deposited into the composite mold and the bone portion of the implant was added to form the final shape of the cartilage (Figure 6E). When the hydrogel was seeded with mesenchymal stem cells (MSCs) and polymerized, after 24 hours in culture, MSCs remained viable (Figure 6F).

Fig. 6.

(A) Design of composite implant with keel, bicortical screws, parent bone, and engineered cartilage surface (in red). (B) Flowchart of workflow from CT data to a cell-seeded implant. (C) ABS positive mold (top) and PDMS negative mold (bottom) used to create a PCL foam implant (D). (E) Composite construct with PCL foam and hydrogel (red) (F) Live (green)/Dead (red) image of MSCs in MeHA in the cartilage layer. Scale = 200μm.

3.5. Patient-Specific Implants and Contact Restoration.

The molding technique described above was next used to generate patient-specific anatomic constructs for three different minipig forelimbs. Starting with clinical CT data, each accessory carpal was segmented in silico (Figure 7A). Implants, and positive molds were designed (Figure 7B-C), and positive molds were 3D printed (Figure 7D). PDMS negative molds were created (Figure 7E) and finally, a patient-specific PCL implant was molded for each forelimb (Figure 7F). For each of these, an acellular, stiff PEGDA hydrogel (with cartilage-like mechanical properties) formed the cartilage surface.

Figure 7:

A) Renderings of 3 ACBs from CT data. (B) Designed ACB implants with cartilage in red. (C) Design of positive molds of the three implants. (D) 3D printed positive and (E) cast PDMS negative molds. (F) Resulting PCL foam constructs before addition of hydrogel.

The force and contact area between the accessory carpal bone and the ulnar carpal bone was measured through a range of flexion angles in each of these three limbs. The loading pattern was similar to the results described above—with force (Figure 8A-C) and contact area (Figure 8D-F) remaining close to zero until around 15 degrees of flexion, before ramping up and reaching a maximum at full extension. The maximum force was 23 ± 6, 24.2 ± 0.7, and 17.5 ± 2.5N for pigs 1, 2, and 3 respectively and the contact area was 107 ± 12, 75 ± 7, and 41 ± 5 mm². When the articulating surfaces of the ACBs were removed and our anatomic patient-specific constructs were implanted orthotopically, this loading pattern was restored. After implantation, the maximum force detected was 21 ± 1, 21 ± 1 and 23 ± 1 N for pigs 1, 2, and 3 respectively and the contact area was 93 ± 12, 60 ± 9, and 62 ± 8 mm². Loading curves before and after ACB resurfacing showed a strong correlation, with Pearson correlation coefficients >0.96 for all three animals for both force and contact area as a function of flexion angle.

Figure 8:

A-C: Force measurements across flexion angles for each of 3 pigs before and after implantation of composite implant. D-F; Contact area of same. r_xy is the Pearson correlation coefficient between preoperative and postoperative trials.

4. Discussion

Over the past two decades, the field of chondral and osteochondral tissue engineering has progressed substantially, while at the same time advances in imaging and rapid prototyping technologies have allowed for the creation of patient-matched, anatomically-shaped constructs. However, despite this progress, there is not yet a tissue-engineered anatomic implant approaching viability for clinical translation. Arthritis of the trapeziometacarpal joint is a debilitating disorder without a clear solution. Surgical interventions are limited at best and new treatment modalities are sorely needed. This study coupled recent innovations in tissue engineering and anatomic fabrication to address the unmet clinical need of restoration of the trapeziometacarpal joint.

Resurfacing of the trapezium presents a unique opportunity that is tailor-made for testing the function of anatomic osteochondral engineered constructs. First, the geometry of this articulation is complex and irregular, necessitating a custom, image-based approach for replacement. Second, the cartilage is quite thin[62] (~0.5mm), decreasing required path-lengths for nutrient diffusion and reducing the previously reported maturation barriers in the center of very thick tissue-engineered constructs.[63] However, without an animal model to test the efficacy of biologic joint resurfacing strategies and to refine fabrication techniques, it is difficult to make progress towards clinical translation. In this study, we established the porcine accessory carpal bone as a model for the human trapezium. We characterized the ACB in terms of loading environment, geometry, material properties, and chemical content. Furthermore, we developed a method to reproducibly manufacture a patient-specific articulating surface for the ACB using biomaterials conducive for osteochondral tissue engineering. Finally, we established the ability of these engineered ACBs to restore native loading patterns when implanted orthotopically.

Findings from this work demonstrate that the porcine accessory carpal bone is remarkably similar to the human trapezium. Most notably, the saddle-like shape of the ACB’s major articulating surface closely matches the shape of the articulation between the trapezium and first metacarpal. The articulating surface area of the trapezia samples from the two female human donors and two male donors were ~2 and 3 times the average size of the cartilage surface of the porcine ACB, respectively. Furthermore, the magnitude of forces acting across ACB and both reported[64,65] and measured (Supplementary Figure 1) forces across the trapezium are of similar magnitude. We found that the average TMC contact forces for opening a jar and writing with a pen were 144 ± 77 N and 42 ± 25 N, respectively, which compares favorably to those measured across the ACB joint.

Another important finding of this study was that the porcine ACB is unloaded throughout most of the forelimb range of motion. Only at joint angles close to full extension, such as when the pig is standing or in the stance phase of ambulation, does the ACB experience load. This suggests that a simple flexion splint might be used during post-operative management in future in vivo studies, to provide early protection for healing and integration of an engineered replacement. This will need to be validated via additional cadaveric studies, but it is similar to the post-operative care possible in hand and upper extremity surgery in humans. Together, these findings support the porcine ACB as a viable testbed for anatomic tissue engineering of living osteochondral trapezium replacements in humans.

In carrying out this work, we also came to appreciate that both the size and the shape of the ACB vary a great deal between animals, and neither can be predicted accurately from the weight or size of the animal. However, despite variations in the surface area and shape between individual animals, the thickness of the cartilage remained consistent. Therefore, clinical CT of the bones of the pig carpus yield sufficient information to predict the actual cartilage surface thickness on the ACB. Furthermore, the mechanical properties of the cartilage did not vary substantially across the surface, and so any engineered cartilage could be spatially homogeneous across the joint.

This geometry data informed our workflow for the design of patient-specific ACB implants. Clinical CT was used to create a CAD model of the ACB for each pig, which was then modified to the final shape of the bone-portion of the implant. We demonstrated that the workflow of this fabrication process is feasible and could be applied to clinical translation. The turnaround time for the molding process was just a few days from CT imaging to production of a patient-matched osteochondral replica. These replicas could then be seeded with the patient’s own cells and grown to maturity in vitro over the course of 8–12 weeks, at which point the mechanics approach native tissue levels.[22] The design of the construct itself, inclusive of the boney integrating region, enabled manipulation and surgical implantation without compromising the overlying engineered cartilage. Moreover, the inclusion of the keel allowed for surgical fixation within the parent bone using standard surgical tools and techniques. Importantly, these patient-specific implants effectively restored the normal loading patterns of the intact joint. This study established that a patient-specific engineered ACB construct placed orthotopically restored the load transfer between the ulnar carpal and the ACB.

This study is not without limitations, however. First, the number of accessory carpal bones analyzed may not fully capture the range of inter-subject variability. We used eight specimens to assess geometry and cartilage thickness, six for cartilage mechanics, and three whole limbs for each of the two joint-level biomechanics experiments. Moreover, we only looked at young but skeletally mature animals (15–24 months of age), and so we did not capture changes that may occur to the ACB before the animal reaches skeletal maturity or as the animal ages. Regarding our ex-vivo orthotopic replacement experiment, we did not assess the extent to which anatomic fidelity influenced recapitulation of contact forces. We do not know if a generic ACB implant, or an implant designed from another animal’s imaging data could have restored the function of the joint as the anatomic versions employed here. As with other transplant approaches, for example for osteochondral cylinders and meniscus allotransplantation, establishing guidelines and tolerances for this size matching[66] will be an important feature to investigate in future studies. Shape quantifications such as those described in Ateshian et al [67] may help us evaluate the shape fidelity of living implants. Finally, though we have established this joint as a potential animal model for studying biologic joint replacement, this study did not yet evaluate durability or integration of our designed implant with in vivo load, or how matured living constructs[68] would perform in the in vivo setting. This will be the focus of ongoing studies.

5. Conclusions

In this study, we assessed the joint loading of the porcine accessory carpal bone, its geometric and histologic features, and its cartilage mechanics, and demonstrated the feasibility of a workflow and methods to reproduce a patient-specific anatomic replica from biomaterials conducive to tissue engineering as a model for testing biologic joint resurfacing of the human TMC joint. To further this concept towards clinical translation, a series of in vitro studies to characterize engineered osteochondral tissue maturation in this system as well as orthotopic ex vivo studies to further characterize implant durability to repeated loading and to refine surgical implantation techniques will be performed. Future studies will also include implantation of mature, patient-specific anatomic tissue-engineered constructs in living animals and characterization of long-term integration and function. These advances may one day provide for a reproducible and functional means of creating living joint replacements throughout the musculoskeletal system.

Statement of Significance.

Biologic joint resurfacing, or the replacement of a joint with living tissue as opposed to metal and plastic, is the holy grail of orthopaedic tissue engineering. However, despite marked advances in engineering native-like osteochondral tissues and in matching patient-specific anatomy, these technologies have not yet reached clinical translation. Given its propensity for developing osteoarthritis, as well as its small size and complex shape, the trapezial surface of the trapeziometacarpal joint at the base of the thumb presents a unique opportunity for pursuing a biologic joint resurfacing strategy. This work establishes the porcine accessory carpal bone as an animal model for the human trapezium and presents a viable test-bed for evaluating the function of engineered living joint resurfacing strategies.