Influence of Staphylococcus epidermidis biofilm on the mechanical strength of soft tissue allograft

Abstract We sought to determine the impact of bacterial inoculation and length of exposure on the mechanical integrity of soft tissue tendon grafts. Cultures of Staphylococcus epidermidis were inoculated on human tibialis posterior cadaveric tendon to grow biofilms. A low inoculum in 10% growth medium was incubated for 30 min to replicate conditions of clinical infection. Growth conditions assessed included inoculum concentrations of 100, 1000, 10,000 colony‐forming units (CFUs). Tests using the MTS Bionix system were performed to assess the influence of bacterial biofilms on tendon strength. Load‐to‐failure testing was performed on the tendons, and the ultimate tensile strength was obtained from the maximal force and the cross‐sectional area. Displacements of tendon origin to maximal displacement were normalized to tendon length to obtain strain values. Tendon force‐displacement and stress‐strain relationships were calculated, and Young's modulus was determined. Elastic modulus and ultimate tensile strength decreased with increasing bioburden. Young's modulus was greater in uninoculated controls compared to tendons inoculated at 10,000 CFU (p = 0.0011) but unaffected by bacterial concentrations of 100 and 1000 CFU (p = 0.054, p = 0.078). Increasing bioburden was associated with decreased peak load to failure (p = 0.043) but was most significant compared to the control under the 10,000 and 1000 CFU growth conditions (p = 0.0005, p = 0.049). The presence of S. epidermidis increased elasticity and decreased ultimate tensile stress of human cadaveric tendons, with increasing effect noted with increasing bioburden.

reconstruction is used in certain circumstances and includes the same types of tendons harvested from donors, as well as others, including hamstring, patellar, quadriceps, Achilles, and anterior/posterior tibialis tendons. 4 ACL reconstruction is generally safe and effective, and infection after ACL reconstruction is a very rare complication, with current literature suggesting a rate of approximately 0.14%-1.7%. 5 While rare, the occurrence of a significant postoperative infection after ACL reconstruction may directly compromise the mechanical strength of the graft. Infection rates are mostly attributed to members of the Staphylococcus species, with Staphylococcus epidermidis being of particular clinical interest, as it has emerged as one of the most common organisms seen in prosthetic joint infections. Its predominance may be related to its ability to form biofilm infections on tissue grafts and implants, making antibiotic efficacy and bacteria removal more difficult. 6 Previous work by this study group has shown that bacterial DNA is commonly found on failed ACL reconstruction grafts in the form of biofilms, as its presence on failed graft tissue and monofilament suture was visually confirmed with fluorescence microscopy. 7 While bacterial DNA was detectable in torn graft tissue in most revision ACL cases, the degree to which biofilm formation affected the mechanical strength and stability of the ACL reconstruction graft remained unclear. 8 Furthermore, it is important to note that ACL failure due to subclinical infection may be underestimated as tissue and synovial cell cultures fail to accurately detect all biofilms. The tube method and congo red agar methods compared to tissue culture plates methods were associated with greater false-negative rates when detecting clinical biofilm. 9 Also, while the presence of bacterial DNA does not necessarily cause clinically relevant infection symptoms, biofilm presence is known to be associated with other issues such as tunnel widening. 8 Previous studies have shown that increased tunnel widening could lead to graft failure, joint laxity, and increased revision surgery requirements. 10 Preliminary work in the lab has shown that S. epidermidis can develop biofilms on human tendon grafts and that growth conditions for governing bioburden to mimic clinical infection can be controlled.
Increasing incubation time is associated with greater bioburden and with increased exposure time, greater unraveling of the tendons making up the graft occur. Mechanical integrity seems to be impacted with increased exposure; however, inoculated allograft tendons have not been tested to confirm this hypothesis. The impact of bacterial colonization on the mechanical integrity of soft tissue ACL grafts as a potential cause of graft failure is concerning and deserves further investigation.
The purpose of this study was to assess the influence of S. epidermidis bacterial biofilm presence on tendon mechanical strength of soft tissue tendon allografts used in ACL reconstruction.
We hypothesized that an increase in S. epidermidis concentration will compromise the mechanical strength of the soft tissue tendon allograft. Our approach to explore this hypothesis was to use mechanical testing protocols to observe a tendons ability to elongate and fail under axial loads. Quantitative measurements for tendon mechanical strength include an analysis of Young's modulus, peak stress, and strain.

| Tibialis posterior graft preparation
Cadaveric human posterior tibialis tendons were surgically released from surrounding tissues, maintaining the proximal end of the tendon to its muscular attachment. Fresh frozen specimens were stored in a −20°C freezer and removed the day before mechanical testing. For tendon preparation, tendons were submerged in 50-ml of sterile water in a clear tempered glass pie dish (dimensions, 9″ 23) and trimmed to remove any remaining muscular attachment. Following graft composition, tendons were imaged, and calculation of surface area was approximated using the analysis function of NIH ImageJ (https://imagej.nih.gov/ij/). The length of the tendon was measured and six data points along the width of the tendon were averaged, and cross-sectional area (CSA) was approximated assuming an elliptical shape ( Figure 2).

| Mechanical testing and data collection
Following biofilm growth, the allograft tendons underwent mechanical testing using a servohydraulic materials test frame (MTS 858 Bionix; MTS Corp.). Testing parameters were set to apply an axial load at a defined strain rate of 10 mm/min. The tendon specimen was gripped by two custom-fabricated, corrugated cryo-clamps, reinforced using dry ice (to improve adhesion between the tissue and clamp), and further clamped using two C-clamps to mitigate the risk of slipping ( Figure 3A). The texture of the corrugated surface gives a good method of monitoring slippage of the specimen; displacement was demonstrated to be tissue extension and not specimen sliding in the grip. Force-deformation curves were generated during elongation to failure of the tendon.
An assessment for graft displacement and load were recorded across the length of the experiment. To find the greatest stress the tendon could withstand, the ultimate tensile strength, the maximum load on the tendon was divided by CSA to obtain stress values. The displacements of the allograft tendon origin to maximal displacement were normalized to tendon length to obtain strain values. From the data obtained, the tendon force-displacement and stress-strain relationships were calculated, and Young's modulus were determined.

| Interpretation of stress-strain curves
A total of 40 allograft tibialis tendons (10 with each inoculum concentration) underwent biomechanical testing undergoing uniaxial loading to failure (rate of 10 mm/s) using the MTS 858 Bionix Test system. This displacement rate was chosen because the recorded quantitative data on the characteristics of soft tissue was most favorable to observe tissue behavior before tendon failure, usually about 2 min of applied axial load.
A general stress-strain relationship for tendons was modeled and included calculations for ultimate tensile strength, peak tendon strain, and Young's modulus. Young's modulus of tendon tissue was calculated by dividing yield stress by the corresponding tendon strain.

| Environmental scanning electron microscopy preparation
All tendon components were soaked in prefixing agents containing 2.5% glutaraldehyde in 0.2 M cacodylate buffer (pH 7.4) for 24 h at room temperature in a 12-well plate. These components were then rinsed with cacodylate buffer twice. After the final rinse, the specimens were dehydrated by placing in increasing concentrations of ethanol series (70%, 90%, and 100%) two times each for 5 min.
F I G U R E 1 Semitendinosus bioburden by incubation time and initial inoculum concentration. Increasing incubation time was associated with greater bioburden. In contrast, initial inoculum concentration had minimal effect on bioburden. CFU, colony-forming unit.
F I G U R E 2 Tibialis posterior graft preparation. For tendon preparation, tendons were submerged in sterile water in a clear tempered glass dish and trimmed to remove any remaining muscular attachment Components were lastly dehydrated in 100% hexamethyldisilazane for 5 min, twice, then depressurized before a sputter coating procedure. These components were dried at room temperature before imaging via Quanta 200 imaging system.

| Statistical analysis
Data were analyzed using standard statistical software on Microsoft Word. The association between bacterial presence and modulus of elasticity (Young's modulus) and peak stress to failure (Sigma) were assessed by analysis of variance (ANOVA). To assess for an association between bacterial concentration and changes in both elasticity and peak stress, paired t-tests assuming equal variance were conducted across each inoculum concentration. Differences were considered significant for p < 0.05.

| Interpretation of stress-strain curves
The recorded data were observed as a force-displacement curve and a stress strain curve was calculated. Ultimate tensile strength was dependent on level of inoculation and CSA of the tendon. Across 10 replicates, the uninoculated experimental control displayed an

| Association between bioburden and mechanical strength
Young's elastic modulus and ultimate tensile strength decrease with increasing bioburden. Young's modulus was greater in the uninoculated control group compared to tendons inoculated at 10,000 CFU (p = 0.0011) but was unaffected by bacterial concentrations of 100 and 1000 CFU (p = 0.054, p = 0.078) as detected by t-test: two-sample assuming equal variances ( Figure 5). Increasing F I G U R E 3 Mechanical testing of the tibialis posterior tendon. The tendon specimen was gripped by two clamps, reinforced using dry ice, and further clamped using two C-clamps to mitigate the risk of slipping (A). Specimens underwent an axial strain until peak load was obtained or until the graft snapped (B). To account for tendon size differences, initial length and tension values were included in calculations (C). Tendons underwent mechanical testing using the MTS 858 Bionix Test system (D) SORENSEN ET AL. | 469 bioburden was also associated with a decreased peak load to failure (p = 0.046, Welch ANOVA). The difference was most significant when compared to the control under 10,000 and 1000 CFU conditions (p = 0.0005, p = 0.049). No significant differences were observed regarding the experimental control compared to the 100 CFU inoculum concentrations as assessed by a paired t-test assuming equal variance (p = 0.072). Therefore, no adjustment or stratification was performed based on these variables in subsequent analyses.
There was no association between bacterial concentration and strain as measured by the MTS Bionix testing system.

| DISCUSSION
While rare, infection after ACL reconstruction exists and subclinical bacterial colonization of orthopedic graft material has proven to be clinically relevant. 11 Subclinical biofilm formation and its impact on the mechanical integrity and success of soft tissue grafts has rarely been studied.
The most important finding of the present study is that increasing levels of bacterial colonization are associated with a weakening of the mechanical properties of the ACL soft tissue F I G U R E 4 Stress-strain curves. Tendons underwent an axial load at a defined strain rate of 10 mm/min until tendon failure. Representative stress-strain curves were generated across each experimental replicate. Peak load to failure of the control specimen was significantly greater than that of the 10,000 CFU inoculum. This trend was observed when comparing the control to the 100, 1000, and 10,000 CFU concentration but was statistically significant in comparison to both the 1000 and 10,000 CFU inoculum. BHI, brain heart infusion; CFU, colony-forming unit.
F I G U R E 5 Box and Whisker plots of calculated values for Young's modulus (A) and peak stress (B). (A) There was no difference in bacterial concentration and change in elasticity when comparing the experimental control to 100 and 1000 CFU. There were statistically lower values for Young's modulus among tendons inoculated under 10,000 CFU. (B) There was no statistical difference between increasing bioburden and peak load to failure when comparing the experimental control to 100 CFU. However, there were statistically lower values for peak stress among tendons inoculated under 10,000 and 1000 CFU. CFU, colony-forming unit.
allograft. In particular, despite short-term exposure and relatively low-inoculation time, the presence of increasing bioburden showed evidence of increased elasticity and decreased ability to tolerate axial load.
Despite short-term exposure and relatively low inoculation time during our experiments, we believe that postsurgical infection can harbor an appropriate environment for bacterial growth and biofilm formation regardless of starting inoculum concentration. Perez and Patel 12 found that S. epidermidis was capable of invading osteoblasts and fibroblasts in vitro. It is known that bacterial colonization is associated with transosseous tunnel widening after ACL reconstruction, most likely due to the same mechanism observed above. 8 There is evidence that allograft tendons are safe and effective with no higher risk of infection and equivalent failure rates compared to autografts. 13  Clinically, failures of primary ACL reconstructions are mainly due to poor technique leading to graft laxity, traumatic reinjury, and poor biology that manifests as either failure of graft ligamentization or subclinical infection. 14 A loose graft at 6 months after ACL reconstruction has been shown to increase the risk of later ACL revision surgery, reduce patient return to sport and ADLs, and cause permanent increased anterior laxity. 15 Evidence from our studies suggest that an increase in tendon laxity and a decreasing load to failure in the presence of infection may play a role in graft failure.
Increased tendon laxity can lead to reinjury either traumatically or nontraumatically by altering the biomechanics of the knee.
Because of low ACL reconstruction infection rates, there is no clear consensus regarding the appropriate management of infection-related complications. Current research suggests that presoaking grafts in vancomycin may lead to decreased deep infection rates. [16][17][18] Some studies, including a systematic review by Xiao et al, found that soaking ACL tendon grafts with vancomycin before implantation was associated with a nearly 15 times decrease in infection compared with grafts not soaked. 19 Another study suggested that through reducing hematoma occurrence postsurgery using a Hemovac drain, deep surgical site infections could be reduced when using hamstring tendon allografts. 20 Given the results of our study, it is important to consider the role of ACL infection prevention given the favorable outcomes presented in several studies performed.

| Future directions
Mechanical integrity of the tendonous materials seems to be impacted with increased exposure; however, inoculated tendons will need to be tested to confirm this hypothesis. The exact physiologic mechanism mediating a reduction in mechanical strength remains to be elucidated. Future studies will aim to assess bioburden quantitatively using CFU counts (possible contamination during this experiment) and observe unravelling of soft tissue specimens by S. epidermidis. On a subset of tendons, we plan to complete scanning electron microscopy imaging to get finer resolution of the unraveling and if it is related to the presence of biofilm.

| Limitations
There are several limitations to the current study. There is no prior reported evidence of impaired mechanical strength of soft tissue allografts inoculated with S. epidermidis; due to a lack of available data on the topic, we could not perform a power analysis. The current study could potentially be underpowered which would increase the risk of beta error (i.e., failure to detect a relationship that does exist).
The current study demonstrates a link between bacterial colonization and a decrease in mechanical strength of soft tissue allografts, but it does not prove causation between bacterial colonization and graft failure. Our choice of control included a cadaveric posterior tibialis tendon which effectively controlled for an absence of bacterial colonies, mimicking ideal surgical conditions. However, the possibility of contamination and bacterial colonization during mechanical testing was not controlled for on the surface of the MTS clamps. Whether bacterial colonization occurred outside of the inoculation protocol cannot be determined by the current study design. Additionally, although the measurement of peak stress was recorded, the measurement of CSA needed to be approximated to account for differences in tendon thickness. This could have an impact on the calculated values for stress and thereby Young's elastic modulus.

| CONCLUSIONS
The presence of S. epidermidis increased the elasticity and decreased the ultimate tensile stress of human cadaveric tendons, with increasing effect noted with increasing bioburden.

AUTHOR CONTRIBUTIONS
Hanna Sorensen, Paul Stoodley, Steven D. Swinehart, and David C.
Flanigancontributed to data acquisition, analysis, and drafting, while all authors contributed equally to conception and design, interpretation of data, and revising; they gave their final approval and agreed to be accountable for all aspects of this study. SORENSEN ET AL. | 471