Modular Lattice Constructs for Biological Joint Resurfacing

Production of building blocks

Cubic (1.5 × 1.5 × 1.5 mm) and triangular (4 × 4 × 2 mm) building blocks with and without integrated cylindrical (2 × 5 mm) anchoring pins were manufactured by injection transfer molding. Five types of customized feedstocks were prepared from materials that are already used for medical purpose: alumina (Al₂O₃, CT 3000 SG; Almatis, Ludwigshafen, Germany), hydroxyapatite (HAp; Sigma-Aldrich Corp., St. Louis, MO), β-tricalcium phosphate (β-TCP, Tricafos 13-73; Chemische Fabrik KG, Budenheim, Germany), biphasic calcium phosphate (BCP, 60/40 wt.% mixture of HAp/β-TCP), and synthesized bioactive glass 13-93 (BG). The raw powders were hydrophobized in dry hexane with stearic acid and then dispersed in molten paraffin wax (Granopent P; Carl Roth GmbH, Karlsruhe, Germany) at 120°C for 12 h under continuous stirring up to a solid content of 57 Vol.%. Afterward, the building blocks were debinded at 500°C to burn out all organic components and sintered at 665°C for BG, 1150°C for HAp and β-TCP, 1200°C for BCP, and 1700°C for Al₂O₃ with 2 h dwell time. The flexible lattice structures were generated by bonding of individual building blocks with well-defined dots of biocompatible UV curing adhesive (Vitralit^®4731; Panacol-Elosol GmbH, Steinbach, Germany). A detailed description of the building block fabrication can be found elsewhere.^11,12

Cell isolation and bone grafts

Human adult subchondral bone grafts were obtained from knee joints of 10 patients undergoing total knee arthroplasty for osteoarthritis (mean age 66.4 years, range 57–80 years). Each patient gave informed consent before surgery, and the Institutional Ethics Committee approved the study (Ref. No. 3555). Bone grafts from the femoral condyles were prepared by removing the articular cartilage layer using a scalpel and curette. Bone specimens were cut into rectangular bone blocks (1.5 × 1.5 × 1.0 cm) and thoroughly rinsed in phosphate-buffered saline (PBS) to remove debris and kept in Dulbecco's modified Eagle's medium (Gibco Life Technologies, Grand Island, NY), 10% fetal calf serum (FCS; Gibco Life Technologies), 1% penicillin/streptomycin (Gibco Life Technologies), or frozen at −80°C until further use.

Isolation and characterization of human mesenchymal stem cells

Human bone marrow-derived mesenchymal stem cells (hBMSCs) were provided by Dr. Farida Djouad (Institute for Regenerative Medicine and Biotherapy, Centre Hospitalier Régional Universitaire de Montpellier, France) and were isolated using the published protocols. ¹³ Mesenchymal stem cells (MSCs) were characterized according to the surface expression of CD44, CD29, and CD105 and the nonexpression of hematopoietic markers, CD45 and CD11b. Until use, the cells were thawed and cultured in alpha-minimum essential medium (α-MEM) (Gibco Life Technologies, Grand Island, NY), 10% FCS (Gibco Life Technologies), and 1% penicillin/streptomycin (Gibco Life Technologies).

Cellular attachment and repopulation of different ceramic materials

Cell attachment to the five materials (HAp, β-TCP, BCP, Al₂O₃, and BG) was investigated using cubic building blocks cultured under static conditions in 48-well plates surrounded by a monolayer of bone marrow-derived mesenchymal stem cells (BMSCs) (4 × 10 ⁴ cells/well) in α-MEM (Gibco) supplemented with 10% FCS, 1% penicillin/streptomycin, and 0.1% amphotericin. Optionally, platelet-rich plasma (PRP) was added (10% v/v). Three independent experiments were performed for each material (n = 3). The PRP was prepared as described previously. ¹⁴ After 14 days, the building blocks with the attached and overgrown cells were documented by phase-contrast photos, followed by fixation with 100% ethanol and staining with 4′,6-diamidino-2-phenylindole. The amount of cells per mm ² was counted under fluorescence microscope by automatic cell counting feature of the program ImageJ (Version 1.51p Rasband, 2017). ¹⁵

Dynamic cell culture of 3D constructs on bone explants (3D Spinner flask)

Bone grafts (n = 12) with removed overlying cartilage tissue were prepared by drilling holes (diameter 1.7 mm, depth 7 mm) in a trigonal pattern at a distance of 5 mm. Press-fit fixation of the 3D lattice construct was achieved by impacting the triangular building blocks with integrated anchoring into the drill holes under sterile conditions. After washing, 3 × 10 ⁶ human mesenchymal stem cells (hMSCs) within 12.5 μL α-MEM (with 10% v/v PRP) were applied into the cavities in between the building blocks, followed by application of 100 μL of a commercially available liquid type I collagen (Col1) hydrogel (ChondroFiller liquid; Amedrix, Esslingen, Germany). The two-component collagen hydrogel (Col1 concentration of 8 mg/mL) was provided in a two-chamber syringe. It was stored at −20°C and had to be defrosted to 30–35°C in an incubator before application with a provided syringe adaptor that allows mixing of the components. Gelation was then achieved at 37°C within 3–5 min. After gelation of the collagen matrix, the constructs were kept in static conditions in α-MEM, 10% FCS, and 1% penicillin–streptomycin at 37°C [5% carbon dioxide (CO₂)] for 7 days. Optionally, 100 ng/mL transforming growth factor-β (TGF-β) (R&D Systems, Minneapolis, MN) or 100 ng/μL bone morphogenetic protein-2 (BMP-2) (R&D Systems) were supplemented for that period, followed by transfer to spinner flasks (CELLSPIN; Pfeiffer, Lahnau, Germany) (37°C, 5% CO₂) by alternating rotation (50 rpm) of the pendula for another 3 weeks. Additionally, as a further control for cell differentiation, 1 × 10 ⁶ hBMCs were suspended in the collagen hydrogel (100 μL) and cultured under the same conditions and analyzed by histology.

The samples were scanned by a high-resolution microcomputed tomography (μCT) using a Skyscan 1172 (Skyscan B.V., Kontich, Belgium) equipped with an 11 Megapixel detector and tungsten X-ray tube. Operating conditions were 80 kV voltage and 100 μA current with an Al/Cu 0.2/0.1 mm source filtering. Scanning conditions for 6.78 μm/resolution were a rotation step of 0.25° over 180°, a random movement of the detector by 10 pixel, and an exposure time of 9145 ms. The raw data sinograms were reconstructed with the tomographic reconstruction software (NRecon Client and Server 1.7.0.1 with GPU support; Skyscan, Kontich, Belgium), which calculates the two-dimensional cross-sections after adjusting gray value levels. The gray value adjusting enables the reconstruction with and without the cells to determine the amount of cell volume.

After μCT scanning, the cell-loaded constructs were either prepared for RNA isolation or were fixed by 4% PFA for tissue processing for histological analysis.

Quantitative reverse transcription PCR analysis

Total RNA was isolated from cells using peqGOLD TRIFastTM (peqlab, Erlangen, Germany). The quality of isolated RNA was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA), and the concentration was determined with the NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies, Rockland, DE). One microgram was used for the first-strand complementary DNA synthesis (Amersham Biosciences, Little Chalfont, UK), which then was used for SYBR Green–based quantitative reverse transcription polymerase chain reaction (RT-PCR). Four replications were performed according to the manufacturer's instructions. The following RT-PCR primer sequences were used: Col1a1 (forward) CAGCCGCTTCACCTACAGC, (reverse) TTTTGTATTCAATCACTGTCTTGCC; Col2a1 (forward) CCGAGGCAACGATGGTCAGC, (reverse) TGGGGCCTTGTTCACCTTTGA; Aggrecan (forward) TGAGGAGGGCTGGAACAAGTACC, (reverse) GGAGGTGGTAATTGCAGGGAACAC; and glyceraldehyde 3-phosphate dehydrogenase (forward) TCC TGT TCG ACA GTC AGC CGC, (reverse) CGC CCA ATA CGA CCA AAT CCG T.

Histological analysis

Following μCT scanning, the constructs were washed in PBS and fixed in 4% PFA for 24 h. Tissue blocks were then dehydrated in a graded series of increasing ethanol concentration and embedded in methyl methacrylate (Technovit 9100; Kulzer, Wehrheim, Germany). Serial sections of methacrylate blocks were produced perpendicularly to the constructs using a diamond saw (DWS 175; Diamond WireTec, Weinheim, Germany). The sections were glued onto polished plexiglas object holders, ground and polished to a thickness of 80–120 μm, and then surface stained with Toluidine Blue. For histological analysis a selection of tissue sections was characterized using light microscopy.

Immunohistochemistry

Immunohistochemistry was performed with samples embedded in paraffin. Therefore, the cell-containing matrix was isolated from the respective constructs and fixed in 4% paraformaldehyde. Immunohistochemical detection of Col1 and type II collagen (Col2) was performed as described previously in detail. ¹⁶ Briefly, deparaffinized sections were pretreated with hyaluronidase (2 mg/mL) (Sigma-Aldrich, Munich, Germany) for 60 min and subsequently with pronase (1 mg/mL) (Sigma-Aldrich) for 60 min. Sections were then exposed overnight to anti-human Col1 antibody (MP Biomedicals, Aurora, OH) diluted 1:100, or anti-human Col2 antibody (MP Biomedicals) diluted 1:250. After incubation with a biotinylated donkey anti-mouse secondary antibody (Dianova, Hamburg, Germany), a complex of streptavidine and biotinylated alkaline phosphatase was added. The sections were developed with fast red and counterstained with Hematoxylin.

Optical clearing and light sheet fluorescence microscopy of bone grafts

Human bone grafts were fixed in 4% PFA for 12 h at 4–8°C. After fixation, whole-mount staining of entire bone grafts was performed. Samples were stained with 5 μg Propidium Iodide Solution (Biolegend; cat. no. 421301) in 1% Tween20/1% dimethyl sulfoxide/PBS for 3 days at 4–8°C. Optical clearing of tissue samples was performed as described previously. ¹⁷ In detail, samples were dehydrated by ethanol, transferred to ethyl cinnamate (ECi; Sigma-Aldrich; cat. no. 112372), and incubated at room temperature until they became transparent. Light sheet fluorescence microscopy (LSFM) of optically cleared samples was performed with an UltraMicroscope II (LaVision BioTec, Bielefeld, Germany), including Olympus MVX10 zoom microscope body, LVBT Laser module, Andor Neo sCMOS camera with a pixel size of 6.5 × 6.5 μm ² , NA of 0.5, and detection optics with an optical magnification range of 1.26 × – 12.6 × . General tissue autofluorescence was generated by exciting the samples with a 488 nm optically pumped semiconductor laser (OPSL) and signals were detected at 525/25 nm. For Propidium Iodide excitation, a 561 nm OPSL and a 595/40 nm filter for signal detection were used. A 3D LSFM data reconstruction and analysis were performed with Imaris 9.1 software (Bitplane, Switzerland).

Pin pull-out testing

Pull-out tests were performed using a universal testing machine (Instron 5565; Instron GmbH, Pfungstadt, Germany) with Al₂O₃ pins (diameter 2 mm, length 4.8 mm, conical tip; Fig. 1b) anchored in drill holes of different diameters (ranging from 1.5 to 2.0 mm). We used polyurethane foam blocks as standardized artificial bone substitutes. The blocks were 10 pcf (pounds per cubic foot), laminated with a 3 mm-thick 20 pcf layer to mimic the bone of femoral condyles (Sawbones P/N 1522-633; Pacific Research Laboratories, Vashon, WA). The pins were inserted perpendicularly into the holes within the bone substitutes and pulled perpendicularly at a crosshead speed of 5 mm/sec. The failure tensile strength was recorded for each pull-out test. In total, 25 pull-out tests were performed (n = 5 per drill diameter).

OPEN IN VIEWER

FIG. 1.

Modular lattice constructs as a fixation concept for joint resurfacing by tissue-engineered constructs: (a) Scheme for a modular lattice constructs of different building blocks, anchoring modules, and interposed areas for cell therapy. (b) Transfer molded Al₂O₃ building blocks with various shapes as basic elements for the 3D lattice structures. (c) Exemplary lattice structure of trigonally arranged building blocks connected by a biocompatible polymer glue. (d) Scheme of a 3D arranged lattice construct. High magnification demonstrated the bonding area by polymer glue. (e, f) Fixation modules (pins) serve to anchor the modular constructs to the joint surface (model) and the versatile concept of lattice structures allows to adapt to the contours of joint surfaces. Bars in (c, e, f) = 1 cm. 3D, three-dimensional; Al₂O₃, alumina.

Statistical analysis

Data on the cellular population on the building blocks, pull-out tests, and gene expression by quantitative RT-PCR were analyzed using Student's t-test. p-Values <0.05 were considered significant.

Fabrication of modular 3D lattice structures

The building blocks and anchoring pins for the flexible lattice structures are shown in Figure 1. The accuracies of the sintered structures were determined by automatic optical shape recognition. After sintering, accuracies of <0.1 mm were achieved for all materials (Al₂O₃, β-TCP, HAp, BCP, and BG). The triangular building blocks with edge lengths of 3.4 mm were bonded by a biocompatible glue to yield flexible lattices for joint resurfacing (Fig. 1a, c, d). The flexible lattice was shown to fully adapt to the convex and concave joint surface (Fig. 1d–f). The lattice was stably anchored by cylindrical pins, first applied to 3D-printed model joints as shown in Figure 1f, and afterward to human subchondral bone. Depending on the surgeons' preferences, anchoring modules can either be fabricated as building blocks with integrated pins or as circular ring structures, in which separate pins are fixed by a press-fit (Fig. 1b, e, f).

Cellular attachment and repopulation of building blocks

A high biocompatibility of the scaffold is a prerequisite for successful tissue integration. Therefore, cellular attachment and repopulation of the different ceramic materials were analyzed. Cell attachment could be observed irrespective of the surface structure. Thus, cells attached both on a relatively rough surface, such as that of β-TCP [arithmetic average roughness (R_a) = 3.15 ± 0.53 μm; mean roughness depth (R_z) = 27.16 ± 5.76 μm], and smooth surfaces, such as that of Al₂O₃ (R_a = 1.28 ± 0.08 μm; R_z = 11.53 ± 0.44 μm) (Fig. 2a). However, there were significant quantitative differences between some of the materials. Phase-contrast microscopy revealed a zone of lower cell density around Bioglass building blocks (Fig. 2b). This impression could be confirmed by quantitative cell counting with a significant lower cell population on bioglass compared with that on HAp, β-TCP, BCP, or Al₂O₃ (Fig. 2c). Generally, PRP, as a chemotactic stimulus, significantly increased cell attachment to all materials tested (Fig. 2d).

OPEN IN VIEWER

FIG. 2.

Cell interaction with ceramic building blocks. (a) Two examples of cellular repopulation of a rough surface (β-TCP) and a smooth surface (Al₂O₃) (left image: electron microscopy; right image: DAPI staining). (b) Cells densely grow around building blocks of different ceramic materials placed on monolayer cell cultures. (c) Quantification of cellular repopulation of building blocks (n = 3) of different ceramics using normal culture medium or medium supplemented with PRP. (d) Pooled effect of PRP on the cellular repopulation of all different building blocks. Error bars represent one standard deviation (*p < 0.05). β-TCP, β-tricalcium phosphate; BCP, biphasic calcium phosphate; BMSC, bone marrow-derived mesenchymal stem cell; DAPI, 4′,6-diamidino-2-phenylindole; HAp, hydroxyapatite; PRP, platelet-rich plasma.

Cellular differentiation of hBMSCs

For the purpose of joint resurfacing, repair cells should undergo chondrogenic differentiation that is characterized by a round phenotype and formation of a proteoglycan-rich matrix. In this work, characterized hBMSCs were used as an exemplary cell type to provide a reproducible setting. However, despite their multipotency, the hBMSCs cultured in a 3D collagen hydrogel did not spontaneously undergo chondrogenic differentiation under dynamic culture conditions for 4 weeks (Fig. 3a). The cells typically formed a fibrous extracellular matrix that was positive for Col1, but negative for Col2. Instead, a stimulus was necessary to induce—at least partially—chondrogenic differentiation of the hBMSC. Thus, the application of recombinant BMP-2 induced chondrogenic differentiation with formation of a proteoglycan-rich pericellular matrix as shown by Toluidine Blue staining (Fig. 3a). Some of the cells even showed signs of hypertrophy with increased cell size. Stimulation by TGF-β also resulted in a chondrogenic phenotype, however, without any signs of hypertrophy (Fig. 3a). The—at least partially—chondrogenic differentiation of hBMSC achieved by BMP-2 or TGF-β could be confirmed by immunohistochemistry with formation of an extracellular matrix that was positive for both Col1 and Col2. Cell differentiation was also analyzed by the expression pattern typical of chondrocytes. BMP-2 had no influence on the expression pattern of Col1a1, whereas TGF-β had a moderate, yet not significant, effect on Col1a1 expression (Fig. 3b). Instead, the expression of Col2a1 was significantly increased by TGF-β, whereas the effect of BMP-2 was only moderate for this cell type (Fig. 3c). The ratio of Col2a1/Col1a1, which represents an index for chondrogenic differentiation, was significantly increased by TGF-β (Fig. 3d).

OPEN IN VIEWER

FIG. 3.

Cellular differentation of hBMSCs in collagen hydrogel during a culture period of 4 weeks (3 weeks dynamic). (a) Control cells (CTRL) only adopt a fibroblastic phenotype. Application of recombinant BMP-2 and TGF-β stimulates chondrogenic differentiation of hMSCs with formation of a proteoglycan-rich pericellular matrix. Immunohistochemical detection of Col1 was positive for nonstimulated control cells as well as cells stimulated by BMP-2 or TGF-β. Col2 could not be detected in the matrix formed by control cells, but moderately in the matrix of cells stimulated by BMP-2 or TGF-β. (b) Quantitative real-time PCR for Col1a1 demonstrates no significant regulation by BMP-2 or TGF-β. (c) Instead, the expression of Col2a1 was significantly induced by TGF-β. (d) Stimulation by TGF-β results in the highest chondrogenic index defined by the ratio of Col2/Col1 expression. *p < 0.05. **p < 0.01. Bar = 100 μm. BMP-2, bone morphogenetic protein-2; Col1, type I collagen; Col2, type II collagen; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; hBMSCs, human bone marrow-derived mesenchymal stem cells; hMSCs, human mesenchymal stem cells; mRNA, messenger RNA; TGF-β, transforming growth factor-β.

Application of constructs to bone implants

The fixation of the pins to human subchondral bone by press-fit proved to be a reliable anchoring method. Generally, there is complete freedom for the alignment and arrangement of the pins. As an example, we demonstrated a trigonal/triangular arrangement of the triangular pins. This arrangement provides cavities between the building blocks that allow cell constructs to mature protected from adverse shear forces (Fig. 4a). Cells suspended in a collagen hydrogel were applied into the cavities between the building blocks (Fig. 4b). After a dynamic culture period of 4 weeks exposed to shear forces in spinner flasks, ¹⁸ the cell-loaded matrix still remained stably attached to the building blocks and resisted from being washed out (Fig. 4c). Histological sections from explant cultures demonstrated tight press-fit of the pins within the subchondral bone (Fig. 4d). Interposed cells tightly attached to the building blocks of Al₂O₃ (Fig. 4 e, f). Following the dynamic culture period of 4 weeks, nonstimulated control BMSCs were more likely characterized by a fibroblastic phenotype (Fig. 4e), whereas cells stimulated by TGF-β adopted a round chondrocyte-like phenotype (Fig. 4f).

OPEN IN VIEWER

FIG. 4.

Application of building blocks to bone explants. (a) Macroscopical image of building blocks (each with integrated pin) applied to a subchondral bone explant. (b) Same construct after application of a cell-loaded collagen hydrogel. (c) After a 4-week dynamic culture period with the cell-loaded matrix unchanged localized within the protective cavities of the lattice structure. (d) Histological section of a pin anchored in subchondral bone. (e) Cells (nonstimulated CTRL) in hydrogel attached to the surface of the Al₂O₃ building block. (f) Cells stimulated by TGF-β also attach to the building block and adopt a round chondrocyte-like phenotype. Bars = 10 mm in (a–c); bars = 1 mm in (d); bars = 100 μm in (e, f).

In Figure 5, two cross-sectional images obtained by μCT at different z-axis levels show the inserted building blocks fixed to the bone and the cell-loaded matrix (green), as well as the y-axis and z-axis levels corresponding to the indicated lines (Fig. 5a, b). The building block pins demonstrate a tight press-fit anchoring to the bone structure. Preceding drilling of the holes did not alter the microstructure of the bone trabeculae (Fig. 5b). Graphical images demonstrated the persistence of the cell-loaded matrix (green) between the building blocks after the dynamic culture period of 4 weeks (Fig. 5a, b). A displacement or release from the matrix did not occur.

OPEN IN VIEWER

FIG. 5.

μCT imaging. Cross-sectional images obtained by μCT at different z-axis levels and corresponding transverse images as indicated by the blue and red lines (a, above subchondral bone plate; b, below subchondral bone plate). The absorbance values of cell-loaded matrix are colored green. μCT, microcomputed tomography.

3D matrix formation in dynamic cultures of constructs on bone explants

During the dynamic cell culture in spinner flasks, the cells were exposed to biomechanical forces that at least partly imitate the strain occuring within the joint. Since chondrogenesis is also influenced by mechanical stimulation, we analyzed the explant cultures histologically and by LSFM. In control specimens, hBMSCs did not undergo chondrogenic differentiation despite the dynamic culture conditions (Fig. 6a). Instead, BMP-2-stimulated cells adopted a round chondrocyte-like phenotype and produced a pericellular proteoglycan-rich matrix as shown by Toluidine Blue staining (Fig. 6b). Some of the cells even tended to adopt a hypertrophic phenotype. Stimulation by TGF-β also induced chondrogenic differentiation with formation of proteoglycan-containing matrix (Fig. 6c). LSFM was performed to provide overview images of 3D cell distribution, which can hardly be provided by conventional histology due to the hardness of the ceramic building blocks. LSFM demonstrated that the cells tightly attached to the Al₂O₃ building blocks, and formed a dense cellular network (Fig. 6d). A 3D data processing also showed merging of the cell–matrix and the building block (Fig. 6e). The constructs can be anchored to any joint surface geometry, such as the convex curvature of bone explants from human femoral condyles (Fig. 6f). Filling of the voids of the constructs with cell-loaded matrices, such as collagen hydrogel, yields a smooth surface (Fig. 6f, g).

OPEN IN VIEWER

FIG. 6.

A 3D matrix formation between the building blocks. Histological sections (a–c) demonstrate that control cells (hMSCs) are distributed within the hydrogel but fail to undergo chondrogenic differentiation (a). Stimulation by rBMP-2 (b) and rTGF-β (c) induces chondrogenic differentiation of hMSCs with a proteoglycan-rich pericellular matrix. LSFM reveals merging of the cell-loaded hydrogel with Al₂O₃ building block in a 3D manner (d, top view; e, 3D image). Toluidine Blue staining in (a–c). *Building block; **Subchondral bone plate; arrow, superficial zone of cell-loaded collagen hydrogel; arrow-head, cells evenly distributed within the hydrogel. Macroscopic view of constructs anchored to human femoral condyles shows adaption to the curvature of the joint surface and filling of the voids with collagen hydrogel to yield a smooth surface (f, g). Bars = 100 μm in (a–c), 500 μm in (d, e), 10 mm in (f, g).

Mechanical pull-out tests

Pull-out testing demonstrated comparable peak pull-out tensile stress of the pins anchored by press-fit in drill holes of 1.5 to 1.8 mm (Fig. 7). Only minimal decrease was observed for fixation in 1.8-mm holes. However, a significant loss of pull-out tensile stress was measured in 2-mm drill holes that only provided exact fitting but not press fitting of the pins.

OPEN IN VIEWER

FIG. 7.

Mechanical pull-out testing. Pull-out tensile stress of pins anchored within drill holes of different diameters in polyurethane bone substitute. **p < 0.01.

Versatility of this concept in fixation to subchondral bone

The main advantage of this concept is its versatility. The modular structure can adopt any shape and allows to cover any type of contour, such as convex or concave areas of joint surfaces. This stands in contrast to existing cartilage repair techniques, such as autologous chondrocyte transplantation, osteochondral autograft or allograft transplantation, which are primarily suited for circumscribed lesions with intact surrounding cartilage. ² The missing adequate fixation of tissue-engineered constructs has counteracted the broad clinical use in large nonconstrained lesions so far. Most cartilage repair techniques rely on a protecting surrounding intact cartilage. Cartilage sutures have long been performed as a fixation technique (e.g., first-generation autologous chondrocyte transplantation),^1,7,8 but this cannot be performed in degeneratively altered cartilage and may interfere with the integrity of surrounding and opposing cartilage. ¹⁹ In other studies, pin fixation was also shown to provide superior stability compared with glueing or suturing. ⁹ However, separate pins were shown to destroy the vulnerable scaffold integrity when being driven through the material. ⁷ The novelity of the current concept is that fixation modules are not disruptive elements but represent an integral part of the lattice structure. We demonstrated that pull-out tensile stress exceeded 6 MPa in perpendicular orientation for 2-mm pins fixed by press-fitting to holes with a diameter of 1.5–1.8 mm. Stability in horizontal orientation was even considerably higher. These data propose sufficient biomechanical stability of the constructs within articulating joints in which shear stress was measured to be <1 MPa. ²⁰

Furthermore, modular assembly facilitates to control the velocity of bone ingrowth and resorption rate, as β-TCP, HAp, BCP, or any other biomaterial (e.g., polymers) can be used simultaneously in one multimaterial scaffold. For later clinical application, the use of resorbable biomaterials will be preferable. We used Al₂O₃ in this proof-of-principle study as an example. Nevertheless, for high-friction areas high-strength Al₂O₃ blocks may be an option.

With modern imaging technologies, such as computed tomography and magnetic resonance imaging with the option for3D reconstruction, the lesions and joint surfaces can be 3D characterized. Based on these data, patient-individualized lattices can be manufactured with ideal anchoring positions considering the biomechanical conditions and the varying bone density.

Bone-sparing technique

Generally, bone-sparing techniques are a favorable therapeutic regime. This novel concept is characterized by complete preservation of the subchondral bone plate. This aspect is of high value considering that any type of revision surgery still remains possible. Furthermore, the subchondral bone plate represents a more stable structure than the underlying cancellous bone and should be preserved from a mechanical point of view. Intact subchondral bone also provides a favorable press-fit fixation of the pins and represents an ideal basis for overlying constructs. The bone preservation achieved by the current concept is one of the main advantages over artificial joint arthroplasty. Also, focal resurfacing by metal implants (Hemicap^®, Episurf^®) requires the removal of a significant amount of subchondral bone impairing a possible revision surgery. ²¹

Biomimetic concept

For a long-term success, fixation of the constructs must not only rely on press-fit principles, but should also be achieved by osseous integration with subchondral bone structures. For this purpose, osteoinductive material is favorable. Thus, the pins should have sufficient hardness and stability to allow impaction into the bone, but should also provide osteointegrative properties.

The current work used monophasic building blocks of Al₂O₃ with the focus on anchoring as a proof-of-principle study. However, in future, biomimetic scaffolds will be required for successful biological repair concepts. In this respect, cartilage resurfacing ideally requires at least a biphasic scaffold to provide a pro-osteogenic environment for adequate bonding to the subchondral bone plate and a prochondrogenic environment for the upper zone. We could demonstrate that a number of ceramic materials allow adequate cell attachment, which confirms previous studies demonstrating that bioactive, bioresorbable, nonporous ceramics induce osteogenesis and osteointegration. ²²

To mimic the complete physiological, mechanical, and structural properties of articular cartilage, the lattice structure would require additional overlying layers that provide a continuous and homogeneous surface. In contrast to bone fixation, such superficial layers should ideally exert prochondrogenic properties or should at least be permissive for chondrogenic differentiation. A number of protein-based polymers, carbohydrate-based polymers, or artificial polymers are suited for this purpose. ¹

Although, hardly any material stimulates chondrogenesis by itself, the use of growth- or differentiation factors seems mandatory for tissue engineering concepts. In this work, we exemplarily used well-characterized hBMSC to provide a reproducable setting. Unfortunately, the multipotent hBMSCs did not spontaneously undergo chondrogenic differentiation, but required at least a transient chondrogenic stimulus for chondrogenic differentiation. Among those, TGF-β and BMP-2 are well-established chondroinductive factors^16,23–25 that were able to induce–at least partially—chondrogenic differentiation of the hBMSC. However, full chondrogenic differentiation without Col1 within the matrix could not be achieved with hBMSC in this setting. Thus, the use of cells that are more specifically committed for chondrogenic differentiation, such as chondrocytes or cartilage progenitor cells, may be preferable in future concepts and may even omit the need for growth factors. However, the limited availability of these cells has to be taken into account.

Having general costs in mind, the use of PRP may be a promising additional part for tissue engineering concepts, due to its broad availability, and efficient proliferative and migratory stimulus. ²⁶ Although, PRP may not exert a prochondrogenic effect, its use may be valuable to promote cell proliferation and scaffold repopulation.