Preparation of Allograft-Derived Scaffolds
Fifty-two human freeze-dried Achilles tendon allografts (provided by the Musculoskeletal Transplant Foundation) were prepared as previously described (see Appendix)20. The allografts underwent washing in hypotonic aqueous solutions followed by trypsin digestion and treatment with a combination of an oxidizing agent (peracetic acid) and a detergent (Triton X-100; Union Carbide). The product, referred to as scaffold, was rinsed, freeze-dried, and stored at −80°C until further use (Fig. 1).
In Vitro Histologic Analysis
Midsubstance portions of the allografts (n = 10) and scaffolds (n = 10) were fixed, processed, and embedded for histology. Sections were stained with hematoxylin and eosin (Sigma-Aldrich) or 4,6-diamidino-2-phenylindole (DAPI; Vector Laboratories) to identify cellular and nuclear components, respectively. Representative light (hematoxylin and eosin) and fluorescence (DAPI) micrographs were made (×200).
DNA Content
Samples of allograft (n = 10) and scaffold (n = 10) were weighed and placed in sterile 1.5-mL microcentrifuge tubes. Total DNA was isolated from each sample (DNeasy; Qiagen). The total DNA content was calculated from the absorption at 280 nm analyzed using a spectrophotometer (Spectronic; Thermo Scientific) and normalized to the initial dry weight of the sample.
Scanning Electron Microscope (SEM) Imaging
Samples of allograft (n = 10) and scaffold (n = 10) were processed and cross-sectional SEM images were obtained at 15.0 kV, 50 Pa, and ×150 magnification (S-2600; Hitachi High Technologies America).
Mercury Intrusion Porosimetry
Samples of allograft (n = 3) and scaffold (n = 3) were weighed and placed in a porosimeter, which was evacuated and filled with mercury (initial pressure, approximately 207 kPa [30 psi]) (Micromeritics Instrument). The porosimeter was transferred into a high pressure chamber (345 MPa [50,000 psi]) and the intruded volume was recorded. The volume of intruded mercury per gram of sample was assumed to be equal to the pore volume. The percent porosity was calculated as (1 − envelope density/skeletal density) × 100.
Analysis of Residual Peracetic Acid
Commercially available peracetic acid (PAA) test strips (EMD Chemicals) were used to measure the PAA concentration (in parts per million) in the initial processing solution and after successive wash steps at 24, 48, and 72 hours (n = 10).
In Vitro Cytocompatibility of Scaffolds Assessed by Direct Contact
Scaffolds (n = 10) were removed from storage and placed in 70% ethanol for twelve hours at 4°C. The scaffolds were then removed and washed briefly three times in 50 mL of cell culture medium containing antibiotic and antimycotic agents (Sigma). Negative controls consisting of sterile latex fragments (n = 10) and positive controls consisting of media only (n = 10) were treated similarly. Cell proliferation and metabolic activity (CellTiter 96 AQueous One Solution Cell Proliferation Assay [MTS]; Promega) and cell viability (release and uptake of neutral red stain) were assayed in NIH 3T3 cells as previously described20,23-25.
Assessment of Virus-Inactivating Efficacy During Scaffold Production
Four additional allografts were used to demonstrate the ability of the processing to inactivate viruses. The central portion of the tendon was cut into 1 × 2-cm sections and immersed in 20 mL of Dulbecco modified Eagle medium (DMEM; Gibco) supplemented with 2% fetal bovine serum (FBS; Valley Biomedical). Three tissue sections from each donor were spiked with 108 plaque-forming units (PFU) of vesicular stomatitis virus (VSV, an enveloped virus), simian virus 5 (SV5, enveloped), or adenovirus (nonenveloped). Spiked fragments were incubated on a shaker at 160 rpm at 37°C for twelve hours. After incubation, the titer of remaining virus was determined at four different steps in the scaffold production process by standard viral plaque assay on monolayers of CV-1 cells (for SV5 and VSV) or 293 cells (for adenovirus) to determine the virus-inactivating efficacy of each processing step.
Tensile Testing and in Vivo Implantation26-34 (see Appendix)
Allografts (n = 10) and scaffolds (n = 10) were equilibrated in phosphate-buffered saline solution and underwent in vitro tensile testing. Additional allografts (n = 8) and scaffolds (n = 8) were implanted in skeletally mature New Zealand White rabbits (twenty-eight to thirty-two weeks, 3.5 to 4 kg), which were killed at twelve weeks. Ten knees implanted with either allograft (n = 5) or scaffold (n = 5) underwent in vivo tensile testing. The remaining six rabbit knees implanted with either allograft (n = 3) or scaffold (n = 3) were processed for histologic analysis.
In Vivo Histology
The harvested femur-ACL graft-tibia complex in each of the knees undergoing histologic analysis was fixed in 10% neutral buffered formalin, decalcified in formic acid with Immunocal (Decal Chemical), dehydrated with ethanol, and embedded in paraffin. Consecutive 5-μm-thick sections cut perpendicular to the tunnel axis and the intra-articular portion of the graft were stained with hematoxylin and eosin or with Masson trichrome stain. Histologic analysis (×50 and ×200) of three sections from the bone-graft interface of each specimen was performed to assess fibrocartilage formation, new bone formation, graft bonding to adjacent tissue, the cell penetration depth, and the number of cells within the graft. Two reviewers blinded to the group assignments scored each characteristic according to a previously published scoring system35.
Statistical Analysis
The Mann-Whitney U test was used for analysis of nonparametric data. Data from the in vitro experiments (MTS assay, neutral red assay, DNA content, and residual PAA) were analyzed in triplicate. All numerical data were averaged, the standard error of the mean was calculated, and a comparison was made between appropriate groups using a two-tailed Student t test with equal variances. A p value of ≤0.05 was considered significant.
Source of Funding
This study was supported by research grants from the Arthroscopy Association of North America (AANA) and the Orthopaedic Research and Education Foundation (OREF). Allografts were supplied by the Musculoskeletal Transplant Foundation.
In Vitro Histologic Analysis
Abundant nuclear material (DNA and RNA) was evident in longitudinal sections of allografts after hematoxylin and eosin staining (Fig. 2-A) and DAPI staining (Fig. 2-B). Minimal interfascicular and intrafascicular space was present in the hematoxylin and eosin-stained sections of allografts prior to processing (Fig. 2-A). After processing, no nuclear material was evident on hematoxylin and eosin staining of the scaffolds (Fig. 2-C). DAPI staining revealed no DNA or RNA within the scaffolds (Fig. 2-D). After processing, a subjective increase in intrafascicular and interfascicular space was observed in the scaffolds by hematoxylin and eosin staining (Fig. 2-C).
DNA Content
The mean DNA content (and standard error) of the processed scaffolds (n = 10) was 0.11 ± 0.07 μg DNA/mg tissue dry weight compared with 0.31 ± 0.03 μg DNA/mg tissue dry weight for the untreated allografts (n = 10), a significant decrease of 67% (p = 0.00003).
SEM Imaging
SEM images confirmed the dense microarchitecture observed in sections of the allografts (Fig. 3-A) and the increase in pore size and porosity in the scaffolds (Fig. 3-B).
Mercury Intrusion Porosimetry
The mean total porosity measured by intrusion porosimetry was 49.63% ± 0.57% for the allografts (n = 3) and 87.50% ± 8.69% for the scaffolds (n = 3) (p = 0.008). The median pore size was 17.54 ± 3.73 μm for the scaffolds compared with 13.23 ± 3.20 μm for the allografts (p = 0.204). The mean pore size was 14.02 ± 3.06 μm for the scaffolds compared with 9.99 ± 2.47 μm for the allografts (p = 0.151). Although there was a trend toward increased pore size in the scaffolds, this difference was not significant.
Analysis of Residual PAA
The residual PAA concentration was >50 ppm, the upper limit of the test strips, in the initial solution and the 24-hour wash. The PAA concentration was 7.92 ± 1.89 ppm after the 48-hour wash and 0.42 ± 1.20 ppm after the 72-hour wash (n = 10). None of the scaffolds had residual PAA at the detection limit after the fourth and final wash.
In Vitro Cytocompatibility of Scaffolds by Direct Contact Method
Metabolic activity was determined by the MTS assay in NIH 3T3 cells exposed to the scaffolds (Fig. 4-A). The absorbance at 490 nm was 1.01 ± 0.04 (n = 10), which was 94% of the 1.08 ± 0.07 value for the positive control, cells exposed to media only (n = 10). This difference was not significant (p = 0.36). Cell viability was determined using the neutral red stain release and uptake assay in NIH 3T3 cells exposed to the scaffolds (Fig. 4-B). The absorbance at 540 nm was 0.19 ± 0.02 (n = 10), which was 109% of the 0.17 ± 0.02 value for NIH 3T3 cells exposed to media alone (n = 10). This difference was not significant (p = 0.60). The absorbance for the cells exposed to the scaffolds and for the positive controls (cells exposed to culture media only) differed significantly from the absorbance for the latex negative controls (n = 10) in both assays (p < 0.0001). The absorbance for the negative controls was also ≤10% of the absorbance for the positive controls in each assay.
Assessment of Virus-Inactivating Efficacy During Production of Scaffolds (Fig. 5)
The virus-inactivating efficacy of the scaffold production process was evaluated by spiking the initial allograft with 1.0 × 108 PFU of virus. Complete inactivation of the enveloped viruses VSV and SV5 was demonstrated after the first step. For adenovirus, infectious virus was detected after the first step (approximately 1 × 103 PFU) but not after the second step. The limit of detection for the plaque assay was 2 PFU for adenovirus and 5 PFU for SV5 and VSV. Only two of 108 nonenveloped viruses were detected after the final processing step, equivalent to a sterility assurance level of 0.2× 10−7. The data points in Figure 5 indicate the total detectable virus at the time of the initial tissue spiking and immediately following each step of the scaffold production process.
In Vitro Tensile Testing (Table I)
All specimens failed at the midsubstance and none were excluded from analysis. No gross slippage or intermittent loss of load during elongation was observed during testing. The ultimate tensile load at failure was 184.69 ± 32.62 N for the scaffolds (n = 10), which was 100% of the 184.14 ± 24.36 N value for the allografts (n = 10) (p = 0.99). The stiffness of the scaffolds (n = 10) was 65.87 ± 38.66 N/mm, which was 73% of the 90.69 ± 11.13 N/mm value for the allografts (n = 10) (p = 0.15). The ultimate tensile stress at failure was 20.40 ± 3.69 MPa for the scaffolds (n = 10), which was 69% of the 29.47 ± 5.13 MPa value for the allografts (n = 10) (p = 0.17). The calculated elastic modulus of the scaffolds was 166.34 ± 36.09 MPa, which was 59% of the 280.51 ± 52.45 MPa value for the allografts (p = 0.09). The percent elongation at ultimate tensile stress was 33.11% ± 2.79% for the scaffolds, which was 138% of the 23.91% ± 1.57% value for the allografts; this difference was significant (p = 0.01).
In Vivo Tensile Testing (Table II)
None of the allografts or scaffolds failed in vivo. During tensile testing, the failure occurred in the midsubstance in one of five scaffolds, at the femoral end in two, and at the tibial end in two. The failure occurred at the femoral end in three of five allografts and at the tibial end in two. Tensile testing indicated the ultimate tensile load to be 37.74 ± 12.13 N for the scaffolds (n = 5), which was 116% of the 32.49 ± 16.08 N value for the allografts (n = 5) (p = 0.08). The stiffness was 16.49 ± 6.56 N/mm for the scaffolds (n = 5), which was 105% of the 15.66 ± 8.02 N/mm value for the allografts (n = 5) (p = 0.94). The ultimate tensile stress was 9.90 ± 3.12 MPa for the scaffolds (n = 5), which was 97% of the 10.23 ± 7.67 MPa value for the allografts (n = 5) (p = 0.96). The calculated elastic modulus was 26.08 ± 8.56 MPa for the scaffolds, which was 108% of the 24.16 ± 17.36 MPa value for the allografts (p = 0.92). The percent elongation at ultimate tensile stress was 122.75% ± 19.74% for the scaffolds, which was 120% of the 102.27% ± 39.49% value for the allografts; in contrast to the difference observed in vitro, this difference was not significant (p = 0.66).
In Vivo Histology
No significant differences in the formation of new bone, formation of fibrocartilage, or graft bonding to adjacent tissue were observed between the scaffolds (n = 3) and the allografts (n = 3) after implantation in vivo (Fig. 6, Table III). Significantly more infiltrating host cells were observed in the scaffolds (Figs. 6-C and 6-D) compared with the allografts (Figs. 6-A and 6-B) (p = 0.002). A significantly greater host cell penetration depth was observed in the scaffolds (Figs. 6-C and 6-D) compared with the allografts (Figs. 6-A and 6-B) (p = 0.002).
The maintenance of necessary tensile properties, elimination of potential sources of disease transmission, and optimization of ultrastructural architecture to accelerate graft remodeling are critical elements of musculoskeletal scaffold design. This study confirmed that human allograft tissue can be processed to remove infectious viral material and produce a decellularized, architecturally modified, cytocompatible scaffold that promotes cell infiltration in vivo yet retains tensile properties similar to those of allograft in vitro or in vivo.
Use of allograft tissue has the potential for disease transmission from contaminated tissue36. Guelich et al. reported a positive bacterial culture rate of 9.7% (twenty-four of 247 allografts) in a retrospective review of 321 consecutive ACL reconstructions37. Similarly, Diaz-de-Rada et al. found positive bacterial cultures in 13% (twenty-four) of 181 allograft tissues38. Although the high rate of positive bacterial cultures is of concern, it may not translate into an equivalent rate of clinical infection. Few cases of septic arthritis have been reported in which the organism cultured was the same as the one cultured before implantation39,40. The most concerning scenario involves the distribution of human allograft tissue that had been deemed safe as a result of laboratory error or incomplete testing. Such tissue could potentially transmit fatal diseases (hepatitis B, hepatitis C, or AIDS [acquired immunodeficiency syndrome]) to allograft recipients, with devastating consequences41. Although the literature confirms that this risk is low, the confirmed cases represent a major medical and surgical challenge and pose serious medicolegal and ethical issues. The oxidative elimination of infectious viral material in the present study resulted in a level of viral sterility that surpassed that of current industry standards3. Further studies are planned to validate the effect of the scaffold production process on bacterial and fungal sterility.
Scheffler et al. compared the remodeling process and restoration of mechanical function of a free soft-tissue autograft and an identical allograft in an in vivo sheep model42. These authors found a significant delay in remodeling and decreased tensile properties in the allograft group compared with the autograft group. They speculated that delayed remodeling resulted from differences in the extracellular and collagenous matrix and from the immunogenic response to the allograft tissue. Others have noted similar findings in allografts3,4,17-19. In our experiments, a significant decrease in residual cellular material was observed in the processed scaffolds, minimizing the immunogenic and inflammatory response to scaffold implantation. The observed increase in total porosity in conjunction with the removal of cellular material from the original allograft (resulting from the combined oxidative and decellularizing effect of the solution containing PAA and Triton X-100) was associated with earlier and increased host cell infiltration into the implanted scaffold. Infiltration of cells may promote earlier repopulation, remodeling, and integration of scaffold compared with traditional allograft tissue. At harvest, the scaffolds appeared to be in the early maturation phase of healing, with increased cell infiltration and possibly a shorter necrotic phase compared with the allograft, as would be expected after decellularization and oxidation8. Theoretically, this finding may translate into accelerated rehabilitation and earlier return to activity or sport, although further studies are necessary to prove this.
PAA has been successfully used to sterilize human patellar tendon allograft without impacting its mechanical properties43. Scheffler et al. analyzed the remodeling of PAA-sterilized ovine ACL allograft and found that PAA slowed the remodeling process and altered mechanical properties at six and twelve weeks compared with nonsterilized allograft and autograft44. In contrast, we did not detect residual PAA, and the use of PAA in our decellularization protocol did not significantly decrease the tensile properties of the scaffold either in vitro or in vivo, render the scaffolds cytotoxic, or provoke an inflammatory response in vitro. The significantly greater percent elongation of the scaffolds compared with the allografts at ultimate tensile stress in vitro was expected. The oxidative and enzymatic disruption of the allograft is necessary to increase porosity and to remove infectious particles and donor cellular material during scaffold production20. The increased elongation is likely a consequence of the disruption of covalent and noncovalent interactions within the scaffold leading to decreased resistance to deformation.
Limitations of the current study include the use of a single allograft tendon type, freeze-dried human Achilles tendon allograft. However, other types of tissues have responded favorably to the same decellularization protocol45. Furthermore, freeze-dried allografts are an excellent source material because they possess equivalent tensile properties prior to and after implantation in vivo, their incorporation is similar to that of fresh-frozen allografts, they are less immunogenic than fresh or fresh-frozen grafts, and they are not associated with disease transmission in recipients even when the graft came from an infected donor. In contrast, fresh-frozen and fresh grafts from those same donors did result in disease transmissions to recipients46. Another limitation is the time period at which the implanted scaffolds were studied. Further longitudinal in vivo studies are necessary to determine the behavior of implanted scaffolds, with an additional comparison to autograft tendon, at time points beyond twelve weeks to investigate the long-term clinical results of allograft-derived scaffolds. Similarly, tissue processing protocols may require optimization to influence scaffold behavior at time points beyond twelve weeks. Although the scaffold production process may increase the number of usable donor allografts, there is still a limited supply of donors.
The tensile properties of the allografts and the scaffolds were determined in both the in vitro and the in vivo portion of the study using established testing parameters that allowed direct comparison of the scaffolds and the allografts in vitro or in vivo with equivalent error26-34. However, differences in the overall dimensions and fixation between these two portions of the study (and particularly the complexity of the in vivo portion) make a direct comparison of in vitro with in vivo tensile and material properties inappropriate. Thus, no comparison should be made between the in vitro and in vivo results with regard to these specific properties. Similarly, the small specimens used for tensile testing in this study represent only a small portion of an allograft and not an intact paratenon. Therefore, the observed tensile and material properties cannot be directly compared with prior results for freeze-dried human Achilles tendon allografts and other common grafts used for ACL reconstruction. The tensile properties of grafts from human donors were variable and unpredictable (consistent with the results of a previous study in which the mean ultimate tensile stress of human Achilles tendon was 1189 N but the range was 360 to 1965 N47), which would have made statistical analysis challenging for even large (>100) sample numbers. Therefore, a source of musculoskeletal tissue with more homogenous tensile properties and greater availability, such as xenograft tissue, may prove valuable in future studies.
In summary, this study supported the hypothesis that human allograft tissue can be processed to remove infectious material and produce a decellularized, cytocompatible, architecturally modified scaffold with tensile properties similar to those of human allograft tissue in vitro and in vivo. The process also succeeded in removing inflammatory material from the allografts and yielded an increased porosity of the allograft-derived scaffolds that subsequently led to increased cell infiltration in vivo.