Design of composite scaffolds and three-dimensional shape analysis for tissue-engineered ear

Thomas M. Cervantes, Erik K. Bassett, Alan Tseng, Anya Kimura, Nick Roscioli, Mark A. Randolph, Joseph P. Vacanti, Theresa A. Hadlock, Rajiv Gupta, Irina Pomerantseva, Cathryn A. Sundback


Engineered cartilage is a promising option for auricular reconstruction. We have previously demonstrated that a titanium wire framework within a composite collagen ear-shaped scaffold helped to maintain the gross dimensions of the engineered ear after implantation, resisting the deformation forces encountered during neocartilage maturation and wound healing. The ear geometry was redesigned to achieve a more accurate aesthetic result when implanted subcutaneously in a nude rat model. A non-invasive method was developed to assess size and shape changes of the engineered ear in three dimensions. Computer models of the titanium framework were obtained from CT scans before and after implantation. Several parameters were measured including the overall length, width and depth, the minimum intrahelical distance and overall curvature values for each beam section within the framework. Local curvature values were measured to gain understanding of the bending forces experienced by the framework structure in situ. Length and width changed by less than 2%, whereas the depth decreased by approximately 8% and the minimum intrahelical distance changed by approximately 12%. Overall curvature changes identified regions most susceptible to deformation. Eighty-nine per cent of local curvature measurements experienced a bending moment less than 50 µN-m owing to deformation forces during implantation. These quantitative shape analysis results have identified opportunities to improve shape fidelity of engineered ear constructs.

1. Introduction

Complete reconstruction of the external ear remains a surgical challenge for both congenital and acquired auricular defects. True hallmarks of a successful reconstruction include recreation of the complex three-dimensional contours of the outer ear, as well as the considerable flexibility of auricular cartilage. Current reconstruction strategies have used carved costal cartilage or rigid polymer implants [13]. However, neither approach truly meets the functional requirements of shape fidelity and flexibility. Carved costal cartilage retains some compliance but lacks flexibility, and cosmetic results are somewhat unpredictable [4]. Polymer implants are capable of excellent cosmetic results, but the constructs are inflexible and can fracture, and possess a long-term risk of extrusion through the skin [4].

Tissue-engineered ear cartilage presents a promising alternative that can overcome the drawbacks of existing methodologies [5]. A proof-of-concept study in immunocompromised mice demonstrated the development of flexible neocartilage within a composite ear-shaped biodegradable collagen scaffold [6]. A key feature of this scaffold was the presence of an embedded titanium wire framework, which had sufficient rigidity to maintain the shape of the ear despite the compressive forces of implantation and contractile forces exerted during neocartilage formation. The wire framework also had sufficient flexibility to permit natural elastic bending of the ear structure. A two-dimensional profile analysis, commonly used to assess basic shape changes [79], showed minimal shrinkage of the engineered ears containing a wire framework (2.0%) compared with those without the wire framework (16.5%). These two-dimensional measurements indicated a successful proof-of-concept prototype but offered limited information on the three-dimensional behaviour of the complex structure. Additionally, the ear scaffold size (half-size of an adult human ear) was designed to fit on the back of a mouse and the master was hand carved without consideration of feature definition loss when placed under the skin. The technology is now under development for clinical trials, and thus we have scaled-up and redesigned the prominent features of the scaffold to match the size of an adult human ear and to preserve the aesthetic appearance after implantation. We also employed more rigorous methods to analyse the fidelity of the ear geometry after in vivo implantation.

Few attempts have been made to objectively quantify three-dimensional size and shape changes of engineered ear-shaped constructs. An approach, modified from Tanzer's classifications of auricular deformities [10], graded ear morphology based upon the shape of specific ear landmarks [11]. However, this method is subjective and provides no quantitative benchmarks. A three-dimensional laser scanning system was used to capture the surface geometry of engineered ear constructs [12,13]. Local and global shape similarity to the pre-implantation geometry was determined by comparing the volumetric pixels with variation of less than 1 mm from the designed ear geometry. This approach is contingent upon careful removal of all surrounding connective tissue from the surface of ear construct. It is suitable for in vitro studies because the surface geometry is unhindered and readily accessible.

In a clinical setting, this method may be successfully employed to compare the reconstructed ear to the target geometry of the contralateral ear. A similar approach to quantify three-dimensional shape changes has been used in other craniofacial repair procedures. Results of neonatal cleft lip repair were evaluated by comparing surface contours of the upper jaw [14]. Plaster moulds of the jaw were obtained and digitized with a three-dimensional laser scanning system. Finite-element scaling analysis software was used to compare local changes in size and shape. In these two clinical approaches, the surface of interest is readily exposed and accessible. However, for the purposes of this in vivo study, the implanted ear must be compared directly with the original scaffold geometry, which would require removal of all surrounding tissue. In this case, a surface imaging approach would pose significant challenges for maintaining consistency and accuracy.

Other approaches to shape comparison have also been demonstrated in addition to surface comparisons. Algorithms for patient-specific craniofacial reconstruction in forensic anthropology were validated using a quantitative tool termed a Euclidian distance matrix (EDM) [15]. An EDM is a numerical matrix consisting of pairwise distances between a set of pre-determined landmarks upon a surface [16]; a higher correlation between EDM ‘signatures’ indicates similar surfaces. In that study, robust results were achieved by using 52 defined landmarks on the skull; achieving a similar number of landmarks for the ear would be difficult, as no precedent currently exists. A rigorous quantitative method for evaluating breast plastic surgery was developed, using  three-dimensional photography to create a model of the patient's breast [17]. A generalized reference shape was fitted to the  three-dimensional model using a set of pre-defined clinically relevant deformation parameters. We were unable to identify any publications that define a similar set of common deformation parameters to evaluate aesthetic outcomes for auricular reconstruction.

Instead of using a surface-based approach, we chose to quantify shape fidelity of the engineered ear constructs by monitoring the deflections of the titanium wire framework. High-resolution models of the titanium frameworks can be obtained non-invasively from CT scans. Several  three-dimensional parameters can then be measured using computer software. This method was contrasted with a conventional surface topology approach, using software to analyse geometries captured with a  three-dimensional photography system.

The goal of this study was twofold: (i) to design and analyse an adult human ear scaffold with better aesthetic appearance than demonstrated in our proof-of-concept study and (ii) to evaluate the non-invasive methodology for assessment of  three-dimensional shape fidelity. A digital model of an adult human external ear was acquired, evaluated by a facial plastic surgeon, and modified to warrant adequate aesthetic appearance after implantation under the skin of a rat. The dimensions of this ear-shaped scaffold were significantly altered to closely resemble adult human auricular cartilage. Shape fidelity of the engineered ears was assessed by comparison of  three-dimensional photography and computed tomography (CT) images of the scaffolds before implantation and after 12 weeks in vivo in a nude rat model.

2. Material and methods

2.1. Ear geometry design

The ear geometry was designed to have the size and features of an adult human outer ear. A  three-dimensional digital model of a human ear was acquired (3D CAD Browser, Hampshire, UK) and adapted to an ear-shaped scaffold model (figure 1a) using SolidWorks CAD software (SolidWorks Corp, Waltham, MA, USA). Contours and key landmarks such as the tragus, antitragus, lobule, fossa triangularis, helix and antihelix were accentuated to ensure that the features retained sufficient definition when covered with a skin layer. Key features were adjusted in conjunction with a facial plastic surgeon to ensure that the shape was aesthetically pleasing following implantation (figure 1b). The resulting model (figure 1c) was three-dimensionally printed using stereolithography technology (Realize, Inc., Noblesville, IN, USA). The three-dimensionally printed ear was cast in polydimethylsiloxane (PDMS) (Dow Corning Corp, Midland, MI, USA) to create a mould which was then split along the outer contour, resulting in a two-piece cavity mould (figure 1d).

Figure 1.

Ear shape design and composite scaffold fabrication process. (a) Initial ear CAD image and (b) the corresponding prototype with plastic surgeon notes. (c) Revised ear prototype with accentuated features to better visualize ear landmarks upon implantation. (d) PDMS negative mould used to cast collagen slurry with embedded wire framework. (e) Titanium wire framework with outer coil sheath; close-up depicts outer coil sheath and an intersection point of skeleton inner wires. (f) Line drawing of inner skeleton path. Dashed lines are overlap regions required for facile construction. (g) Composite collagen ear scaffold with embedded wire framework.

2.2 Wire framework fabrication

The wire framework was designed in the shape of the prominent ear contours. The framework consisted of two components, an inner core and an outer coil (figure 1e). The inner core was a continuous piece of 0.38 mm diameter titanium wire (Ti6Al4V ELI, annealed) (Small Parts Inc, Miramar, FL, USA). The inner core was manipulated using hand tools to match the contours of the CAD model. Because the core was constructed from a single wire, there were four regions where the wire overlapped upon itself (figure 1f). The outer coil had an inner diameter of 0.40 mm and an average pitch of 0.76 mm, and was hand wound with 0.24 mm titanium wire (Ti6Al4V ELI, annealed) (Small Parts Inc.) using a drill to coil the wire around a rod.

Prior to use, the wire frameworks were sonicated (Ultrasonic Cleaner FS30, Fisher Scientific Inc, Waltham, MA, USA) in chloroform (Sigma-Aldrich Co., St Louis, MO) for 5 min and then allowed to air dry on aluminium foil. The chloroform wash was repeated twice and followed by a detergent wash. Detergent (Sparkleen, Thermo Fisher Scientific Inc.) was dissolved in deionized water according to the manufacturer's instructions and the wires were sonicated for 15 min. The wires were then thoroughly rinsed under running deionized water before sonicating in clean deionized water for 5 min.

2.3. Composite scaffold fabrication

Wire frameworks were embedded within fibrous collagen matrix (bovine dermis type I collagen, Kensey Nash Corp, Exton, PA, USA) to create the composite ear scaffolds. The PDMS mould halves were filled with aqueous collagen slurry and debubbled manually. The wire frameworks were dipped into slurry and placed within one side of the mould. The mould halves were brought together and excess slurry extruded through the mating surfaces of the mould. Scaffolds then underwent lyophilization and cross-linking, resulting in an ear-shaped porous collagen matrix with the embedded wire framework (figure 1g). Prior to cell seeding, the scaffolds were sterilized using ethylene oxide gas (55°C).

2.4. Chondrocyte isolation and culture

Chondrocytes were isolated from auricular cartilage of adult Polypay sheep as previously described [6]. Isolated chondrocytes were plated into roller bottles (Corning Inc., Corning, NY, USA) at 3 × 103 cells cm–2. Culture medium consisted of Ham's F12 medium (Invitrogen Co., Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS, Sigma-Aldrich Co.), 100 U ml–1 penicillin, 100 μg ml–1 streptomycin, 292 μg ml–1 l-glutamine and 0.1 mM non-essential amino acids (all from Invitrogen). After reaching confluence, the chondrocytes were trypsinized with 0.05% trypsin-EDTA and used for the study.

2.5. Cell seeding and construct culture

Chondrocytes were suspended in 3 ml of culture medium at 50 × 106 cells ml–1 and seeded onto ear-shaped scaffolds, as previously described [6]. Constructs were cultured in wide mouth, 500 ml polypropylene containers with screw-top lids (Thermo-Fisher Scientific). For gas exchange, 4 mm diameter holes were drilled into the lid and the bottom of the containers and covered with a silicone film which was glued to the cap and bottom of the container. An approximately 150 µm thick silicone film was created by applying MDX4-4210 (Dow Corning Corp) silicone pre-polymer to a silicon wafer coated with a sacrificial substrate of polyacrylic acid [18] and spinning it at 1500 r.p.m. for 45 s using a plate spinner (Brewer Scientific, Rolla, MO, USA). Constructs were cultured in standard incubator conditions for two weeks in 75 ml of culture medium additionally supplemented with 50 μg ml l-ascorbic acid (Wako Pure Chemical Industries Ltd., Osaka, Japan) on a Talboys standard orbital shaker (Henry Troemner LLC, Thorofare, NJ, USA) at 55 r.p.m. The culture medium was changed twice a week.

2.6. Construct implantation and harvest

Ear-shaped constructs, both with and without an embedded wire framework, were implanted subcutaneously on the backs of male nude rats (Charles River Laboratories, Wilmington, MA, USA), one construct per rat. General anaesthesia was achieved with intramuscular injection of ketamine (75 mg kg–1) and dexmedetomidine (0.5 mg kg–1). Under aseptic conditions, a 4 cm incision was made on the lower dorsum of the animal and a subcutaneous pocket was created with blunt dissection. After insertion of the ear-shaped construct, the skin was closed with absorbable sutures. Reversal of anaesthesia was achieved with 0.5 mg kg–1 antisedan. All constructs were harvested at 12 weeks and the surrounding connective tissue was carefully dissected. Select constructs were repeatedly bent and twisted along the long axis to demonstrate flexibility.

2.7. Three-dimensional surface imaging

Three-dimensional surface imaging technology was used to compare shape differences between unseeded composite scaffolds and the same ear constructs harvested at 12 weeks. Surface renderings were captured with the Vectra imaging system (Canfield Imaging Systems, Fairfield, NJ, USA), which uses stereophotogrammetry technology. A three-dimensional volume of the ear was constructed from images taken by a network of cameras positioned at different angles. Scaffolds and engineered constructs were placed on the matching PDMS ear mould half during image capture to ensure consistent orientation. PDMS moulds were painted with a matte black coating to reduce reflection artefacts.

2.8. CT imaging and segmentation

CT technology was used to capture three-dimensional images of the engineered ear constructs. A Siemens SOMATOM Definition Flash CT machine was used (Siemens, Muenchen, Germany) with the following parameters: peak tube voltage = 120 kV, tube current = 196 mA, revolution time = 1 s, section thickness = 16 × 0.3 mm and spiral pitch factor = 0.85. Scans were taken of the composite scaffolds before cell seeding and of the same engineered construct immediately after the harvest and prior to bending manipulation (figure 2a). The wire framework volume was segmented from the raw CT data (figure 2b) using Mimics software (Materialise, Leuven, Belgium) by Biomedical Modeling, Inc (Boston, MA). The centreline of the framework was exported as a compilation of curves for analysis (figure 2c).

Figure 2.

Curvature analysis methods. (a) High-resolution dual-energy CT scan of titanium wire framework. (b) CAD model was segmented from raw CT data. (c) Centreline curves of the wire framework were generated to calculate distances and curvatures and (d) were labelled with wire section numbers and the minimum intrahelix (IH) distance. (e) The arc length and the chord length, the distance between the endpoints, were measured for section 7, ear 2. The arc length : chord length ratio describes the overall section curvature. (f) An osculating circle was constructed along the segment arc to measure the radius of curvature at evenly spaced points (shown as spheres). The radius of the circle equals the radius of curvature at each point of intersection. (Online version in colour.)

2.9. Shape fidelity analysis

Shape analysis of the segmented wire framework models was conducted using SolidWorks CAD software. Changes in overall shape of the framework were assessed as well as local changes in the curvature profile of individual sections (figure 2d). Overall three-dimensional length, width and depth were measured for each framework as well as the minimum intrahelical distance (min. IH). Additionally, the ratio of arc length to chord length for each framework section was measured (figure 2e). This parameter is indicative of the overall curvature of a section; a decrease in the ratio indicates that the section flattens, while an increase suggests that the section has become more curved. This method was adapted from approaches to analyse vascular tortuosity [19].

Changes in the local curvature profile reflect the bending moment experienced by the framework. The local curvature at single points along each framework section was measured by constructing an osculating circle [20]. Curvature values were measured at evenly distributed points along the length of the framework section, with a minimum density of 0.42 points mm–1 (figure 2f). The set of local curvature values were measured for scaffolds and explanted ears. Differences between the local curvature profiles of analogous sections on the scaffold and explanted ear were used to calculate the bending moments experienced within the framework. Bending moment was estimated using the Euler–Bernoulli relationship [21]Embedded Image 2.1where M is the bending moment, E is the modulus of elasticity, I is the area moment of inertia and Δκ is the change in curvature (all in SI units). The value of E for annealed Ti6Al4V ELI is 113.8 GPa [22]. The value of I is defined asEmbedded Image 2.2where Dwire is the diameter of the titanium wire [21]. For framework sections 4, 5 and 7, where the inner core wire significantly overlaps itself, the value of I is approximated by 2 × Dwire. The Euler–Bernoulli equation is not considered valid for structures with a high degree of curvature because the neutral axis is not aligned with the centroid of the structure geometry. Therefore, equation (2.1) was not considered valid for curvature values whereEmbedded Image 2.3This relationship ensures that the offset of the neutral axis from the centroid is less than 1% of Dwire. Points that did not meet this criterion were not included in the results of the bending analysis.

2.10. Histological analyses

Full-thickness 7 mm diameter biopsies were punched at four locations on explanted constructs with wire in the areas of fossa triangularis, between helix and antihelix, in the conha and lobule regions. Complete cross-sections were obtained from explanted constructs without wire. Part of each biopsy or section was fixed in 10% buffered formalin for histology and another part was snap frozen and stored at –80°C for biochemical analysis. Paraffin-embedded specimens were sectioned at 6 μm thickness. Sections were stained with haematoxylin and eosin (H&E), safranin O, toluidine blue and Verhoeff's elastic stains. Collagen was determined immunohistochemically. Tissue sections were pre-treated with 1 mg ml–1 pepsin in Tris–HCl (pH 2.0) for 15 min at room temperature followed by peroxidase block. Sections were incubated with 1 : 100 mouse anti-human collagen type I antibody (Accurate Chemicals & Scientific Corporation, Westbury, NY, USA) and 1 : 100 mouse anti-human collagen type II antibody (Developmental Studies Hybridoma Bank, Iowa City, IA, USA) for 30 min. The EnVision+ System kit (Dako, Carpinteria, CA, USA) was used to identify the antigens according to manufacturer's instructions. Sections were counterstained with haematoxylin.

2.11. Biochemical analyses

For DNA content, frozen samples were weighed and DNA was extracted and purified with the DNeasy kit (Qiagen Inc., Valencia, CA, USA). Total DNA content was determined using the PicoGreen dsDNA assay (Invitrogen). For glycosaminoglycan (GAG) determination, samples were lyophilized for 24 h and digested overnight at 60°C in 125 µg ml–1 papain. GAG content was determined spectrophotometrically using the Blyscan Glycosaminoglycan Assay Kit (Biocolor Ltd., Carrickfergus, UK) according to manufacturer's instructions.

2.12. Statistical analysis

Biochemical analyses values were expressed as mean ± s.d. Statistical analyses were performed using SPSS v. 11.0 (SPSS, Chicago, IL, USA). Comparison of means was assessed by a one-way analysis of variance and the Tukey multiple comparison test (p < 0.05 was considered significant).

3. Results

3.1. Gross evaluation

All implants were well tolerated and no exposure or extrusions occurred during 12 weeks in vivo. The implants with wire resembled a human ear (figure 3a,b), whereas implants without wire were flattened and distorted (figure 3c); the surfaces of both were white, glistening and grossly resembled cartilage. The constructs with the embedded wire framework had significantly improved size and shape fidelity, compared with those without the framework. The implants were flexible (figure 3d; electronic supplementary material, video S1).

Figure 3.

Gross appearance of the engineered ear with embedded wire framework implanted subcutaneously in a nude rat for 12 weeks (a) before and (b) after explant. (c) Gross image of explanted engineered ear without an embedded wire framework. (d) The explanted engineered ear with wire framework maintained its shape and could be elastically deformed.

3.2. Histological analysis

Histological and immunohistochemical staining of the neocartilage extracellular matrix (ECM) exhibited patterns similar to native cartilage ECM, as demonstrated with safranin O, toluidine blue, collagen type II and elastin (figure 4a). Collagen type I staining was negative in all constructs (not shown). Cartilage ECM was seen throughout the explants as demonstrated by a composite image of safranin O-stained cross-section of the ear-shaped construct without wire (figure 4b).

Figure 4.

Engineered cartilage characterization. (a) Neocartilage formation in ear-shaped scaffolds after 12 weeks' implantation in nude rats demonstrated by H&E, safranin O, toluidine blue, collagen type II and elastin stains. Residual collagen scaffold fibres stained red with H&E. Scale bar, 200 µm. (b) Composite image of the cross-section of the ear-shaped construct without wire framework. Neocartilage formed throughout the construct after 12 weeks in a nude rat. Unstained areas represent residual collagen scaffold fibres. Safranin O staining. Scale bar, 2 mm. (c) DNA content normalized to sample wet weight and GAG content normalized to sample dry weight in ear-shaped constructs explanted from rats at 12 weeks. Both parameters were similar to those of native sheep cartilage. Differences between constructs with wire and control constructs without wire were not significant (p > 0.1).

3.3. Biochemical analyses

Analysis of the DNA and GAG content of engineered cartilage in constructs with and without wire showed values similar to those in native sheep ear cartilage (figure 4c). The DNA content of engineered cartilage from constructs with and without the wire support was 91.6 ± 18.5 and 72.5 ± 21.2 ηg mg–1 wet weight, respectively, compared with 94.6 ± 16.1 ηg mg–1 for native cartilage (p > 0.1). GAG content for engineered cartilage from constructs with and without wire and native cartilage were 93.6 ± 16.4, 79.6 ± 7.7 and 91.4 ± 7.8 μg mg–1 dry weight, respectively (p > 0.1).

3.4. Three-dimensional surface imaging

Three-dimensional surface images of a scaffold and corresponding explanted ear (with embedded wire framework) were superimposed with depth differences coloured in a scaled heat map (figure 5a). A positive depth difference indicates material subtracted from the original scaffold, whereas a negative difference indicates added material. Differences in depth ranged within ±5 mm; 69.3% of the surface had depth changes within ±1 mm and 96.2% within ±3 mm. The largest displacements occurred in the central area of the ear (cavum of concha) as well as in the contours corresponding to framework sections 7 and 8 (tragus and antitragus).

Figure 5.

Analysis of three-dimensional shape changes in engineered ear with embedded wire framework. (a) Heat map comparing depth differences between three-dimensional surface images of ear construct before implantation and after explant. Depth differences range from 5 mm (blue, less material) to –5 mm (red, additional material). The concha appears red indicating added material, likely fibrous connective tissue not fully dissected after explantation. Per cent changes between scaffold and explant are shown for (b) overall curvature (arc length: chord length) and (c) arc length for each beam section in ear 1. Arc length changes imply slippage of the inner wire skeleton within the outer coil sheath. (d) Curvature and bending moment analysis for selected points along section 8 of ear 1 for the scaffold and explanted ear. (e) Distribution of bending moments for ears 1 and 2 after explantation.

3.5. Shape fidelity analysis

Two sets of ear scaffolds and explanted constructs were scanned with a dual-energy CT and segmented to isolate the wire framework. Three-dimensional splines of each framework were constructed from the centreline curves and used to analyse shape fidelity. A comparison of the per cent changes in overall length, width, depth, and the minimum IH distance are shown in table 1. Relative changes in overall curvature (arc length : chord length ratio) for each framework section are shown in figure 5b and relative changes in arc lengths are shown in figure 5c. The total sum of the arc lengths for sections 1–9 differed by less than 2% between scaffold and explant for each ear (table 1), suggesting good accuracy for the segmentation and curve construction methods. The representative local curvature profile for section 8 of the ear 1 framework is depicted in figure 5d. Ninety-one per cent of all measured curvature values in all sections met the inclusion criterion (equation (2.3)). Bending moment magnitudes for all framework sections ranged from 0.04 to 274.80 µN-m, with an average value of 18.6 µN-m (figure 5e). Eighty-nine per cent of all bending moment values were less than 50 µN-m.

View this table:
Table 1.

Comparison of key dimensional parameters between scaffolds and explants.

4. Discussion

Optimal qualities for an engineered replacement ear are to maintain shape fidelity and to demonstrate similar flexibility to native auricular cartilage. The composite scaffold concept demonstrated previously in nude mice [6] was adjusted to more accurately replicate the size and contours of an adult human ear when placed under the skin. The larger nude rat model was used to evaluate the behaviour of the scaled-up ear construct with improved definition. Three-dimensional imaging modalities were introduced as a new tool to quantitatively analyse the shape fidelity and the wire framework behaviour and identify opportunities to further improve the scaffold design and mechanical performance of an engineered ear.

The composite scaffold maintained definition of key ear features during implantation (figure 3a). The result is aesthetically superior to prior iterations of the ear scaffold without accentuated contours [6]. In particular, the regions of the antitragus, lobule, fossa triangularis, helix and antihelix are clearly visible under the skin. During clinical procedures, definition will be improved further, as the skin will be pulled taut over the scaffold through vacuum application, as is standard for current ear reconstruction techniques [4].

The outer coil of the wire framework served multiple purposes. First, it held the inner core wire in the desired shape by enclosing the four regions of overlap. The coil increased the surface area of the framework by roughly threefold, encouraging tissue ingrowth and attachment around the wire, which reduced the risk of extrusion through the skin. As shown previously, neocartilage formation was observed between the wire coils and surrounding the inner core wire [6]. Finally, the coil increased the effective diameter of the framework, which distributed pressure and further reduced the risk of extrusion.

Overall, engineered ear constructs with an embedded wire framework maintained size and features, superior to the constructs without a framework over the 12 week implantation period (figure 3b,c). All constructs were able to withstand substantial bending and twisting and were flexible (figure 3d; electronic supplementary material, video S1) at the time of explant. Traditional mechanical testing, such as a three-point bending test, was not performed on the engineered ear construct; no data currently exist on the bending mechanics of the native ear structure as a whole, so there is no benchmark for comparison. Histological analysis of the explanted constructs showed that neocartilage morphology closely matched that of native cartilage. The DNA and GAG content of engineered ear constructs was similar to that of native cartilage. The embedded wire had no negative effect on neocartilage formation based on the results of histological and biochemical analyses (figure 4).

Three-dimensional image analysis offered further insights to the behaviour of the structure. Length and width exhibited minimal change of less than or equal to 1%. Depth, however, decreased significantly. This result is evident from the surface imaging (figure 5a), overall depth measurements from the segmented CT models (table 1) and the decrease in overall curvature in framework section 6 (figure 5b). This behaviour probably occurs owing to the compressive forces of the surrounding tissue upon subcutaneous implantation and the contractile forces of neocartilage formation within the collagen scaffold matrix. Framework section 6 corresponds to the arch of the cavum region. This feature points towards the muscle (as opposed to the skin) when implanted on the nude rat. Forces exerted on §6 in this configuration likely flattened the cavum region, resulting in decreased overall curvature for section 6 as well as decreased overall depth of the entire framework. The minimum IH distance also decreases significantly; the long unsupported length of framework section 9 is a configuration that is prone to large deflections. Surface contour changes in the tragus region correspond to relatively large changes in overall curvature for framework section 7. The contour of section 7 is one of the most tortuous sections and thus, is susceptible to larger deflections.

Bending moment was estimated from changes in the local curvature profile. Results suggest that the framework experiences on the order of less than 500 µN-m from forces exerted by subcutaneous implantation and neocartilage formation. Approximation of these data can be used to drive a deterministic selection of wire framework properties. Wire diameter and material modulus affect the bending stiffness (defined as EI) of the framework, which determines its behaviour in response to a bending moment. Bending stiffness must be sufficiently high to maintain shape during in vivo growth, yet sufficiently low to allow for elastic deformation consistent with native auricular cartilage. These limits create a target design window for frameworks of different material composition (figure 6). The upper limit of bending stiffness is based on the elastic modulus of auricular cartilage [23] and is constrained so that the flexibility of the cartilage structure is unhindered by the wire framework. The lower limit is based on the bending moment values calculated from the curvature analysis (figure 5d,e). Wire diameter is constrained by the average thickness of auricular cartilage [24,25] and the capabilities of manufacturing technologies used to create the wire material. Using this approach, a target wire diameter for the desired material can be determined.

Figure 6.

Comparison of bending stiffness for wire diameters of different materials. The target design window is bounded by the estimated bending stiffness of the native ear (upper), minimum stiffness for in vivo shape fidelity (lower), fundamental manufacturing limitations (left) and the average thickness of native auricular cartilage (right). (Online version in colour.)

Measurements from the segmented CT data also yielded unexpected insights into the wire framework behaviour. Although the total arc length of the framework was consistent between scaffolds and explants (per cent difference less than or equal to 2%), arc lengths for individual framework sections varied (per cent difference 4.5–24%) as shown in figure 5c. Additionally, curvature profile analysis revealed that the largest differences occurred near the endpoints of each section. We hypothesize that this behaviour is the result of the overlapping wire regions sliding upon each other. Because no welds or glue were used, wires in these regions are able to slip within the outer coil sheath, changing the arc lengths of the inner core wire. This explanation can account for the large changes in overall curvature for sections 3, 7 and 9. The finding suggests that shape fidelity may be improved if the intersections are constrained to prevent sliding, which could be achieved through a weld, glue or alternative fabrication approach. However, elasticity of the overall framework may be limited if these sliding joints are fixed. Future iterations of the framework may incorporate different materials, for instance biodegradable polymers [26,27], to reduce stiffness in a fixed configuration.

Shape fidelity of the engineered ear constructs is challenging to quantify because of the irregular three-dimensional geometry. In this study, three-dimensional imaging methods were used to analyse changes in overall shape and curvature. Results from the three-dimensional surface imaging and segmented CT models suggest that these techniques offer an accurate and robust approach to shape fidelity analysis. This approach avoids difficulties associated with surface imaging, such as dissection of the implant from the host and removal of connective tissue from highly curved regions, for instance the cavum of concha. Other imaging modalities, for example magnetic resonance imaging, may be able to identify the surface profile of engineered cartilage in a non-invasive approach.

5. Conclusions

In this study, we describe an improved design for a composite ear scaffold and a non-invasive method for quantifying three-dimensional shape changes and bending moments in a tissue-engineered ear. Results demonstrated regions of the wire framework that are more susceptible to deformations and can be used to improve the design to minimize distortions. Future iterations of the scaffold may incorporate other framework materials and dimensions based on results from an analysis of bending moment magnitudes. This approach can be adapted for non-invasive monitoring of ear shape fidelity in vivo for future clinical applications.

All procedures were approved by the Institutional Animal Care and Use Committee of the Massachusetts General Hospital and performed according to the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals.

Funding statement

This research was sponsored in part by the Armed Forces Institute of Regenerative Medicine award number W81XWH-08-2-0034. The U.S. Army Medical Research Acquisition Activity, 820 Chandler Street, Fort Detrick MD 21702-5014 is the awarding and administering acquisition office.


The content of the manuscript does not necessarily reflect the position or the policy of the Government  and no official endorsement should be inferred.

  • Received May 7, 2013.
  • Accepted July 10, 2013.


View Abstract