Scott J. Hollister

Indirect solid free form fabrication of local and global porous, biomimetic and composite 3D polymer-ceramic scaffolds

Precise control over scaffold material, porosity, and internal pore architecture is essential for tissue engineering. By coupling solid free form (SFF) manufacturing with conventional sponge scaffold fabrication procedures, we have developed methods for casting scaffolds that contain designed and controlled locally porous and globally porous internal architectures. These methods are compatible with numerous bioresorbable and non-resorbable polymers, ceramics, and biologic materials. Phase separation, emulsion-solvent diffusion, and porogen leaching were used to create poly(L)lactide (PLA) scaffolds containing both computationally designed global pores (500, 600, or 800 microm wide channels) and solvent fashioned local pores (50-100 microm wide voids or 5-10 microm length plates). Globally porous PLA and polyglycolide/PLA discrete composites were made using melt processing. Biphasic scaffolds with mechanically interdigitated PLA and sintered hydroxyapatite regions were fabricated with 500 and 600 microm wide global pores. PLA scaffolds with complex internal architectures that mimicked human trabecular bone were produced. Our indirect fabrication using casting in SFF molds provided enhanced control over scaffold shape, material, porosity and pore architecture, including size, geometry, orientation, branching, and interconnectivity. These scaffolds that contain concurrent local and global pores, discrete material regions, and biomimetic internal architectures may prove valuable for multi-tissue and structural tissue interface engineering.

Optimal design and fabrication of scaffolds to mimic tissue properties and satisfy biological constraints

Bone tissue engineering scaffolds must shape regenerating tissue, provide temporary mechanical support and enhance tissue regeneration. These requirements result in conflicting design goals. For example, increased temporary mechanical function requires a dense scaffold while enhanced cell/gene delivery requires a porous scaffold. This paper demonstrates an image-based homogenization optimization approach that can design scaffold microstructure, scaffold material and regenerate tissue microstructure to meet conflicting design requirements. In addition, constraints to ensure adequate cell/gene delivery can be introduced using a minimum porosity threshold. Homogenization theory was used to compute relationships between scaffold microstructure and effective stiffness. The functional relationships were used in the MATLAB optimization toolbox to compute optimal pore dimensions and scaffold material such that the scaffold and regenerate tissue effective stiffness matched that of native bone stiffness. The scaffold design was converted into STL format for solid free-form fabrication. Scaffolds were designed that matched mandibular condyle trabecular bone properties. Results showed excellent agreement between native bone properties and designed scaffold properties (all R2 > 0.89). Finally, example scaffolds were built from hydroxyapatite using a SFF casting technique.

Mechanical and in vivo performance of hydroxyapatite implants with controlled architectures

Internal architecture has a direct impact on the mechanical and biological behaviors of porous hydroxyapatite (HA) implant. However, traditional processing methods provide minimal control in this regard. To address the issue, we developed a new processing method combining image-based design and solid free-form fabrication. We have previously published the processing method showing fabricated HA implants and their chemical properties. This study characterized the mechanical and the in vivo performance of designed HA implants. Thirteen HA implants with orthogonal channels at 40% porosity were tested on an Instron machine. The compressive strength and compressive modulus measured were 30+/-8 MPa and 1.4+/-0.4 GPa, comparable to coralline porous HA. Twenty-four cylindrical HA implants with two architecture designs, orthogonal and radial channels, were implanted in the mandibles of four Yucatan minipigs for 5 and 9 weeks. Normal bone regeneration occurred in both groups. At 9 weeks, bone penetrated 1.4mm into both scaffold designs. The percent bone ingrowth in the penetration zone was higher in the orthogonal channel design but not statistically different due to the low number of samples. However, the overall shape of the regenerated bone tissue was significantly different. In the orthogonal design, bone and HA formed an interpenetrating matrix, while in the radial design, the regenerated bone formed an intact piece at the center of the implant. These preliminary results showed that controlling the overall geometry of the regenerated bone tissue is possible through the internal architectural design of the scaffolds.

A homogenization sampling procedure for calculating trabecular bone effective stiffness and tissue level stress

A homogenization sampling procedure is introduced which allows computation of effective trabecular bone stiffness and individual trabecula level stress based on precise models of trabecular bone architecture. Three-dimensional digitized images of 53 trabecular bone specimens with a resolution of 50 microns per voxel were directly converted into three-dimensional finite element meshes by making each voxel an 8-node isoparametric brick element. Owing to the large mesh of 8000 elements, an element-by-element preconditioned conjugate gradient (EBEPCG) program was written to solve the local homogenization finite element equations. Predicted effective stiffness measures correlated well with experimental results (R2 > 0.73). The predicted effective stiffness tended to under estimate the experimental values. Average absolute errors in effective stiffness estimates ranged between 31 and 38% for the sampling procedure compared to a range 49-150% for a regression fit to volume fraction squared. Trabecula level stress ranged between -200 and +300 times that predicted by analyzing trabecular bone as a continuum. Both tensile and compressive tissue stresses were engendered by a continuum compressive stress. Trabecula level strain energy density (SED) ranged between 0 and 100 times the continuum SED value for two trabecular specimens. In conclusion, the homogenization sampling procedure consistently predicted the influence of trabecular bone architecture on effective stiffness. It can also provide trabecular tissue stress and strain estimates for arbitrary global loading of whole bones. Tissue stresses and strains showed large variations compared to corresponding continuum level quantities.

Craniofacial Tissue Engineering by Stem Cells

Craniofacial tissue engineering promises the regeneration or de novo formation of dental, oral, and craniofacial structures lost to congenital anomalies, trauma, and diseases. Virtually all craniofacial structures are derivatives of mesenchymal cells. Mesenchymal stem cells are the offspring of mesenchymal cells following asymmetrical division, and reside in various craniofacial structures in the adult. Cells with characteristics of adult stem cells have been isolated from the dental pulp, the deciduous tooth, and the periodontium. Several craniofacial structures—such as the mandibular condyle, calvarial bone, cranial suture, and subcutaneous adipose tissue—have been engineered from mesenchymal stem cells, growth factor, and/or gene therapy approaches. As a departure from the reliance of current clinical practice on durable materials such as amalgam, composites, and metallic alloys, biological therapies utilize mesenchymal stem cells, delivered or internally recruited, to generate craniofacial structures in temporary scaffolding biomaterials. Craniofacial tissue engineering is likely to be realized in the foreseeable future, and represents an opportunity that dentistry cannot afford to miss.

A comparison of homogenization and standard mechanics analyses for periodic porous composites

Composite material elastic behavior has been studied using many approaches, all of which are based on the concept of a Representative Volume Element (RVE). Most methods accurately estimate effective elastic properties when the ratio of the RVE size to the global structural dimensions, denoted here as ν, goes to zero. However, many composites are locally periodic with finite ν. The purpose of this paper was to compare homogenization and standard mechanics RVE based analyses for periodic porous composites with finite ν. Both methods were implemented using a displacement based finite element formulation. For one-dimensional analyses of composite bars the two methods were equivalent. Howver, for two- and three-dimensional analyses the methods were quite different due to the fact that the local RVE stress and strain state was not determined uniquely by the applied boundary conditions. For two-dimensional analyses of porous periodic composites the effective material properties predicted by standard mechanics approaches using multiple cell RVEs converged to the homogenization predictions using one cell. In addition, homogenization estimates of local strain energy density were within 30% of direct analyses while standard mechanics approaches generally differed from direct analyses by more than 70%. These results suggest that homogenization theory is preferable over standard mechanics of materials approaches for periodic composites even when the material is only locally periodic and ν is finite.

Bioresorbable Airway Splint Created with a Three-Dimensional Printer

An infant with localized bronchial malacia was treated with a computer-printed bioresorbable three-dimensional splint. Placement of the splint resulted in improved ventilation.

Hydroxyapatite implants with designed internal architecture

Porous hydroxyapatite (HA) has been used as a bone graft material in the clinics for decades. Traditionally, the pores in these HAs are either obtained from the coralline exoskeletal patterns or from the embedded organic particles in the starting HA powder. Both processes offer very limited control on the pore structure. A new method for manufacturing porous HA with designed pore channels has been developed. This method is essentially a lost-mold technique with negative molds made with Stereolithography and a highly loaded curable HA suspension as the ceramic carrier. Implants with designed channels and connection patterns were first generated from a Computer-Aided-Design (CAD) software and Computer Tomography (CT) data. The negative images of the designs were used to build the molds on a stereolithography apparatus with epoxy resins. A 40 vol% HA suspension in propoxylated neopentyl glycol diacrylate (PNPGDA) and iso-bornyl acrylate (IBA) was formulated. HA suspension was cast into the epoxy molds and cured into solid at 85 °C. The molds and acrylate binders were removed by pyrolysis, followed by HA green body sintering. With this method, implants with six different channel designs were built successfully and the designed channels were reproduced in the sintered HA implants. The channels created in the sintered HA implants were between 366 μm and 968 μm in diameter with standard deviations of 50 μm or less. The porosity created by the channels were between 26% and 52%. The results show that HA implants with designed connection pattern and well controled channel size can be built with the technique developed in this study.

Trabecular bone remodeling: An experimental model

An experimental model, capable of inducing controlled stress fields to the distal femoral metaphyses of large dogs, is presented. This model utilized an implantable hydraulic device incorporating five loading cylinders and platens in direct contact with an exposed plane of trabecular bone. A microprocessor controls the loading characteristics, and finite element models were created to calculate the induced stress and strain fields. The trabecular remodeling response is measured using serial in vivo computed tomography, in vitro microcomputed tomography, and histologic analysis. The results of the experiment indicate that significant remodeling can be induced by the activated implant. An increase in trabecular orientation toward the loaded platens was observed, and a statistically significant decrease in connectivity was documented. The greatest effect was associated with a change in the loading rate. A fast rise time (70 ms) loading waveform induced significant bone ingrowth at the implant interface when compared to a slow rise time waveform (700 ms), and demonstrated high correlations with the calculated stress fields as remodeling approached an equilibrium state.

The Accuracy of Digital Image-Based Finite Element Models

Digital image-based finite element meshing is an alternative approach to time-consuming conventional meshing techniques for generating realistic three-dimensional (3D) models of complex structures. Although not limited to biological applications, digital image-based modeling has been used to generate structure-specific (i.e., non-generic) models of whole bones and trabecular bone microstructures. However, questions remain regarding the solution accuracy provided by the digital meshing approach, particularly at model or material boundaries. The purpose of this study was to compare the accuracy of digital and conventional smooth boundary models based on theoretical solutions for a two-dimensional (2D) compression plate and a 3D circular cantilever beam. For both the plate and beam analyses, the predicted solution at digital model boundaries was characterized by local oscillations, which produced potentially high errors within individual boundary elements. Significantly, however, the digital model boundary solution oscillated approximately about the theoretical solution. A marked improvement in solution accuracy was therefore achieved by considering average results within a region composed of several elements. Absolute errors for Von Mises stress averaged over the beam cross section, for example, converged to less than 4 percent, and the predicted free-end displacement of the cantilever beam was within 1 percent of the theoretical solution. Analyses at several beam orientations and mesh resolutions suggested a minimum discretization of three to four digital finite elements through the beam cross section to avoid high numerical stiffening errors under bending.