In recent years the increasing use of implantable medical devices, both temporary and permanent, combined with the growing number of immunocompromised individuals being treated, has lead to a greater number of nosocomial infections.1, 2 Several reports discuss device-associated infections in terms of economic and clinical consequences,3–5 and some studies suggest that Staphylococcus spp. can be isolated from deep wound infections in 70–90% of elective orthopaedic surgery patients.6 This issue increases in importance as the recent rise in infections associated with methacillin/oxacillin-resistant organisms that respond poorly to traditional therapies is considered. Although advances in surgical technique, implant sterilization, and infection control have helped decrease the likelihood of implant-associated osteomyelitis, most would argue that bacterial contamination and infection still occur at an unacceptable level.
Bacterial biofilms have been ascribed a central role in implant-associated osteomyelitis due, in large part, to their resistance to antimicrobial therapy and clearance by the host immune system.3, 7–9 The development of a biofilm is dependent on several bacterial, substrate, and host factors and is a process that is initiated by adherence of planktonic organisms.4, 10 Regardless of their source or location, bacterial biofilms contribute significantly to increased morbidity and mortality, represent a therapeutic challenge, and are an area of tremendous interest in the research community. An important need in this area of research is an animal model that effectively mimics the clinical situation. While numerous implant-associated osteomyelitis models exist, most require creation of a bone defect,11 involve creation of an open fracture,12 result in chronic osteomyelitis13 or require the use of a sclerosing agent to enhance infection.13 At the present time there is not a model that can be used to appropriately mimic implant-associated acute osteomyelitis following repair of closed fractures.
The use of an intramedullary pin or nail is a typical treatment method for many closed fractures. Therefore, creating an animal model that mimics this clinical scenario would render the model useful for translational studies that are designed to evaluate methods to prevent and/or treat osteomyelitis and biofilm formation.
The objective of this study was to modify an existing femur fracture model in the rat to one that could be used as a model of osteomyelitis associated with a closed fracture. We hypothesized that an injection of 104 colony-forming units (CFU) of Staphylococcus aureus would result in acute osteomyelitis and impaired fracture healing in a rat femur fracture model. The development of osteomyelitis was verified by determining the number of the CFU per femur, serial radiographic evaluations of osteomyelitic changes, and postmortem histopathological appearance of the tissues.
Thirty male Sprague–Dawley rats (250–300 g) were randomly assigned into three groups (n = 10/group): Control (pinned femur fracture); S. aureus (pinned femur fracture and bacteria); and S. aureus + ceftriaxone (pinned femur fracture and bacteria and antibiotics). The appetite, general attitude, and surgical sites of the rats were monitored every 12-h for the first 7 days after surgery, after which they were monitored every 24 h. The body weight of each rat was measured prior to surgery and every 7 days after surgery, until the end of the study. Opioid analgesia (buprenorphine 0.03 mg) was given via subcutaneous injection every 12 h for the first 3 days after surgery. The rats were humanely euthanized when they developed radiographic evidence of severe osteomyelitis with systemic illness or 3 weeks after surgery, whichever occurred first. The development of osteomyelitis was documented via serial radiographs (all animals) and postmortem microbiologic (n = 8 per group) and histopathologic (n = 2 per group) analyses. The study was approved by the Institutional Animal Care and Use Committee.
The organism used was a S. aureus strain that was isolated from a patient with an infected total hip arthroplasty, was known to have maintained virulence as assessed by its ability to cause osteomyelitis, and was known to have an in vitro sensitivity to ceftriaxone.11, 14–16 Prior to use in the inoculation, a pure culture of the S. aureus was grown for 24 h on blood agar plates. The bacterial inoculum was prepared such that there was 104 CFU in 50-µl of phosphate-buffered saline (PBS). Tryptic soy broth (TSB) was used as the culture media for the growth of the bacteria from harvested specimens. Tryptic soy agar (TSA) plates were used to grow and quantify organisms recovered from the implant and femora. All organisms and culture samples were incubated at 37°C in ambient air.
316L stainless steel pins (SS) (n = 30), 1.4 mm in diameter and 26 mm in length, were used to internally stabilize the femur fracture. One end of the pin was hand milled so it had a smaller diameter (0.8 mm) and could be easily seated in the proximal aspect of the femur. All pins were sterilized by steam autoclave sterilization before being used.
The surgical procedure was a modification of that outlined in detail in Skott et al.17 Briefly, each rat was anesthetized using intraperitoneal injections of ketamine (100–200 mg/kg) and xylazine (2.0–4.0 mg/kg) and maintained using isoflurane delivered in oxygen via a mask. Breathing rate and depth, mucous membrane color, and jaw tone were used to measure the depth of anesthesia. The right femur was aseptically prepared and an approach to the distal femur was made via a medial stifle arthrotomy. An 18-gauge needle was used to create an entry port into the distal aspect of the medullary canal of the femur and ream the canal for placement of the intramedullary pin. An inoculation dose of 50-µl of bacterial suspension was slowly injected into the medullary cavity via an 18-gauge polypropylene catheter that was left in place for 2-min following inoculation. In the Control group PBS was injected instead of the bacterial suspension. After the bacteria or PBS were injected, the pin was inserted (narrow portion first) into the medullary canal and seated into the cortical bone in the proximal aspect of the femur. The opening in the distal femur was sealed with bonewax to prevent leakage of the bacterial inoculum from the medullary canal. The surgical site was lavaged with sterile saline and the soft tissues and skin were closed. A mid-shaft closed fracture of the right femur was then created using a specifically designed fracture apparatus.17, 18 The femur was radiographed to document the fracture and the rat was recovered from anesthesia. Rats in the S. aureus + ceftriaxone group received ceftriaxone (50 mg/kg) every 24 h, starting 4 h after inoculation, via a subcutaneous injection for the duration of the study.
Radiographic Assessment of Osteomyelitis
Lateral radiographs of the right hind limb were obtained immediately after surgery and at weeks 1, 2, and 3 postoperatively using a digital dental radiographic system (Scan X Digital Imaging System; Air Techniques, Inc., Melville, NY). Two individuals (DAR and MGC), blinded to study group, evaluated the radiographs focusing on three regions of interest (ROI): (1) proximal metaphyseal area where the implant was seated in cortical bone; (2) diaphyseal region involving the site of the fracture; and (3) distal metaphyseal area where access to the medullary canal was made. During each evaluation each radiograph was assessed based on a system used by Lucke et al.6, 19 The following radiographic changes were evaluated for each ROI: (1) osteolysis; (2) soft tissue swelling; (3) periosteal reaction; (4) general impression; and (5) deformity. The changes were given a score corresponding to the following scale: 0—absent; 1—mild; 2—moderate; or 3—severe. For the general impression evaluation a 0 represented a normal appearing femur/fracture and a 3 represented severe changes were present overall. In addition, sequestra formation (6) and spontaneous fracture (7) were evaluated for each femur as a whole and were given a score of 0—absent or 1—present. The scores were then summed, with highest possible total score being 47. The score from two evaluators was averaged for statistical evaluation.
Recovery of Bacteria
Three weeks after surgery, the rats were sedated for radiographic evaluation of the operated femur after which a blood sample (approximately 2.0 ml) was collected aseptically and transferred to a blood culture container (BBL Septi-Chek TSB 20 ml, BD Company, Franklin Lakes, NJ) for aerobic culture, according to the manufacturer's directions. After incubation for 7 days samples were aseptically retrieved from the blood culture container and streaked onto blood and MacConkey agar plates which were then incubated for 48 h at 37°C, after which the plates were evaluated for bacterial growth (methods described below).
After euthanasia a sample of synovium was retrieved from both the operated (right) and unoperated (left) stifle joint from each animal for aerobic culture. The tissue samples were placed in 10.0 ml of TSB and incubated at 37°C for 48 h at which time each sample was classified as culture positive or negative based on the presence of turbidity in the culture vial.
Both femurs were aseptically retrieved and used for bacterial quantification. The SS pins were aseptically retrieved from the operated femurs prior to snap freezing. Under sterile conditions, the pins were placed in 1.0 ml of sterile, chilled PBS, and then sonicated (67 kHz), vortexed, and centrifuged (∼11,000 rpm) to dislodge adhered bacteria. Samples were then collected for microtiter dilution and the results were used to calculate the CFU/pin (methods described below). Each femur was snap frozen and ground to a powder under sterile conditions.11 The resulting powder was suspended in 3.0 ml of chilled TSB, which was kept on ice until sampled (<10 min) for microtiter dilutions and calculation of the CFU/femur (see methods below).
Microtiter Dilution and Viable Bacterial Counts
Microtiter dilutions were performed using a modification of a previously described technique.20, 21 The CFU in each tube was determined in quadruplicate by aseptically collecting a sample. Tenfold dilutions were made (10−1 to 10−6) using PBS in 96-well round bottom microtiter plates. Twenty microliters was collected from each well and streaked across a TSA (Beckton Dickinson Diagnostic Systems, Sparks, MD) plate in a uniform manner. The plates were incubated aerobically at 37°C for 24 h at which time the number of colonies were counted. Dilutions with up to 30 colonies present were used to calculate the median CFU/pin or CFU/femur.
The operated femurs from two rats in each group (n = 6 total) were used for histopathological evaluation. After removal of soft tissues, the intact femurs were fixed in 10% neutral buffered formalin for 48 h, after which they were transferred to 70% ethanol. After decalcification in 10% EDTA, the femurs were bisected midsagittally and the implant was removed. The bisected femurs were then processed for histology and embedded, longitudinally, with the cut surface down, in paraffin. Two 5-µm-thick sections were obtained from one block from each femur and were stained with hematoxylin and eosin. The sections were then evaluated by a veterinary pathologist (CSC) who was blinded to treatment group. Histopathologic descriptions were provided for each section and then combined to provide a summary description for each of the three groups.
The results of the radiographic scoring across the three treatment groups were compared using one-way analysis of variance (ANOVA). The recovered CFU/pin and CFU/femur were reported as median CFU and were compared across all three groups using a Kruskal–Wallis one-way analysis of variance (chi-square analysis), with significance set at p < 0.05. The distributions of radiographic scores were compared for each pair of treatment groups using the Student's t-test with Bonferroni correction. The distributions of CFU/pin or CFU/femur were compared for each pair of treatment groups using the pairwise Wilcoxon rank sums test for nonparametric data with Bonferroni correction, with significance set at p < 0.017 (Bonferroni correction = 0.05/3 = 0.017).
One rat in each of the Control and S. aureus + ceftriaxone groups had to be euthanized within 24 h of surgery due to incisional dehiscence/self-mutilation. As a result of this, only seven rats were available for the culture-related outcome measures in both the Control and S. aureus + ceftriaxone groups versus eight rats in the S. aureus group. Similarly, only nine rats had radiographs available for evaluation in the Control and S. aureus + ceftriaxone groups versus 10 rats in the S. aureus group. No other rats had any obvious health issues and all remaining rats gained weight in a similar manner over the course of the study.
Radiographs were available for all rats that survived the study period (Fig. 1). The fracture site was most evident in the S. aureus group, followed by the S. aureus + ceftriaxone group, but was only faintly evident in the Control group. The radiographs from the Control group were characterized by the formation of a normal bridging fracture callus and minimal periosteal reaction. There was a radiolucent area within the fracture callus that was typical of what would be expected at this stage of healing. The S. aureus group was characterized by a severe periosteal reaction extending along the entire length of the femur, accompanied by significant osteolysis. The periosteal new bone did not bridge the fracture site and osteolytic areas of radiolucency extended proximally and distally from the fracture site. The S. aureus + ceftriaxone group was characterized by a small amount of periosteal reaction and new bone formation. The facture callus did not bridge the fracture site and the areas of radiolucency were present to a lesser degree when compared to the S. aureus group. The mean ± SE radiographic scores for the radiographs taken immediately postoperatively were similar in each of the three treatment groups (ANOVA, p = 0.77). The mean ± SE radiographic scores were significantly different at week 1 (p = 0.02), week 2 (p < 0.0001), and week 3 (p = 0.0005). When the treatment groups were compared as pairs at all three time points, the radiographic scores for the S. aureus group were significantly higher than those in the Control group at each postoperative time point. The radiographic scores for the S. aureus + ceftriaxone group were significantly higher than those in the Control group at week 2 but not at week 1 or 3. The scores for the S. aureus group were also significantly higher than those of the S. aureus + ceftriaxone group at weeks 2 and 3, but not at week 1 (Table 1).
Table 1. Radiographic Scores Immediately and at 1, 2, and 3 Weeks after Surgery
Scores were compared at each time point (within each column). A lower score represents a radiograph that is more similar to normal healing without osteomyelitis. Cells (within a given column) with a different letter were significantly different when a Bonferroni correction was applied (p < 0.017).
1.17 ± 0.48a
6.00 ± 2.23a
9.05 ± 1.03a
10.20 ± 1.55a
0.75 ± 0.45a
11.94 ± 0.89b
21.86 ± 1.36b
20.50 ± 1.67b
S. aureus + ceftriaxone
0.78 ± 0.40a
12.60 ± 1.52a,b
14.50 ± 0.96c
12.20 ± 1.71a
Recovery of Bacteria
No bacteria were recovered from the left (unoperated) femur of any of the rats from any of the treatment groups or from the right (operated) femur in rats in the Control group. Chi-square analysis revealed a significant difference in the CFU/right femur (p < 0.0001) and CFU/pin (p = 0.0002) across all three treatment groups. When the results were compared using the pairwise Wilcoxon rank sums test for nonparametric data, the CFU/right femur was significantly higher in the S. aureus (p = 0.0006) and the S. aureus + ceftriaxone (p = 0.0008) groups compared with the control group; the CFU/right femur was significantly higher in the S. aureus group compared with the S. aureus + ceftriaxone group (p = 0.0026). The CFU/pin results followed an identical pattern to CFU/femur (Fig. 2).
Blood cultures were negative for all 28 rats. Stifle cultures were positive in 8/10 of the knees of rats that were in the S. aureus group, 3/9 of the knees of rats that were in the S. aureus + ceftriaxone group, and 0/9 of the knees of the rats in the Control group. None of the unoperated (left) knees had positive culture results.
Specimens within each treatment group were similar histologically. Specimens from the Control group were characterized by the presence of a callus that was composed of hyaline cartilage and periosteal new bone that bridged the fracture site in both specimens. No areas of inflammation were noted. In contrast, specimens from the S. aureus group were characterized by severe suppurative osteomyelitis with extension of the inflammation to the joint space and periosteum accompanied by extensive myelofibrosis. Although an extensive callus was present in both specimens, involving the diaphysis, metaphysis, and epiphysis, it contained extensive areas of fibrous connective tissue and inflammation (primarily composed of neutrophils) in the area of the fracture site, resulting in the lack of a bony bridge over the site. In both specimens, there also was nearly complete loss of cortical bone the majority of the diaphysis, with only necrotic fragments of this tissue remaining. Specimens from the S. aureus + ceftriaxone group contained a few small foci of suppurative inflammation; however, the marrow cavity, which in the specimens from the S. aureus group contained extensive areas of inflammation, was largely filled with fibrous connective tissue. Both specimens contained a bridging callus that was composed primarily of new bone but included a small component of hyaline cartilage (Fig. 3).
Orthopaedic implants related infections occur in nearly 112,000 human patients annually and create an estimated $1.8-billion burden to the healthcare industry.3 Among categories of orthopaedic implants, fracture fixation devices are particularly susceptible to infection as they are often used on devitalized tissue that may or may not have accompanying wounds that have been contaminated with bacteria.3, 4 The physical and emotional consequences are more difficult to measure; however, infections also often result in revision surgeries, prolonged antibiotic therapy, impaired functional outcome, and lengthy hospital stays.6 The overall outcome is increased patient morbidity and in some cases mortality.22
An animal model that reproduces the salient features of the clinical situation is critical to advance scientific investigations in this field. While there are other implant associated infection models in the rat,11–13 none effectively mimic osteomyelitis that occurs following repair of closed fractures. The model presented here is based on a modification of a previously described fracture model, where a mid-diaphyseal closed fracture of the femur was created using a specifically designed fracture apparatus.17, 18 In the present study, the femur was inoculated with a S. aureus strain that was known to be capable of producing osteomyelitis, as demonstrated by its use in other studies.11, 14–16 Because the fracture site itself was not approached to insert the intramedullary nail, to inoculate the femur or to create the fracture, this model may better reflect the clinical situation of bacterial contamination of a closed fracture. The inoculation of this S. aureus strain at 104 CFU was effective in producing osteomyelitis without having untoward systemic effects on the rats. All the rats gained weight in a similar manner over the course of the study, none had positive blood cultures at the end of the study, and none of the unoperated femurs or synovial membrane samples had positive culture results.
In assessing an orthopaedic model, the use of a radiographic scoring system has several advantages. It is a noninvasive method of collecting data over the course of a study and mimics a diagnostic modality that is used in clinical practice. In our study we used a previously published system6, 19 where ROI were identified and scored using seven different parameters. Significant differences in the radiographic score for osteomyelitis were detected between the Control group and both the S. aureus group by 1 week after surgery/inoculation. When the S. aureus group was compared to the S. aureus + ceftriaxone group a significant difference could also be found starting at week 2. These findings demonstrate that the model was representative of the clinical situation and differences between groups could be monitored noninvasively over time. However, it is important to note that even though reviewers of the radiographs were blinded to group assignment, this remains a subjective outcome measure.
Osteomyelitis is often caused by bacterial infections making the recovery of bacteria an important characteristic of a proposed model. In our study, both the operated and unoperated femurs and pins from operated femurs were cultured and the bacteria recovered quantified as CFU per femur and pin. Blood and tissue cultures were also performed to determine if there was evidence of bacteremia or inoculation of the stifle during surgery. No bacteria were cultured from any of the blood samples or any of the unoperated femurs irrespective of the treatment group suggesting that in this model the infection remains localized to the contaminated femur. Bacteria were consistently retrieved from the inoculated femurs and none were retrieved from the Control group. Significant differences were found when the Control group was compared to both the S. aureus and S. aureus + ceftriaxone groups with respect to the CFU/right femur and CFU/pin. While antibiotic therapy did not eliminate the bacterial infection, significant treatment differences were noted, additional evidence that the model represents the clinical situation. However, one limitation of this model is that the knee (stifle joint) that was used to introduce bacteria into the femur was contaminated and a positive synovial tissue culture was found 3 weeks after surgery nearly 60% of the time. Additionally, this culture assessment did not evaluate the numbers of viable bacteria present, information that would be informative in treatment group comparisons. Because the possibility of septic arthritis exists, future investigations using this model must pay special attention to limiting contamination to the knee when introducing the bacteria, sealing the site where the pin is introduced into the femur and thoroughly lavaging the knee to further limit the contamination.
Although only two specimens/treatment groups were evaluated histologically, the results within each group were remarkably consistent; however, marked differences between the treatment groups were apparent. The marked inflammation present in the sections from the S. aureus group was greatly attenuated in the S. aureus + ceftriaxone group. The quality of the callus was similar in the Control and S. aureus + ceftriaxone groups but was poor in the S. aureus group. Grading schemes are available for use in these types of samples and would provide additional semiquantitative information in studies in which a larger sample size is available for evaluation.6, 19, 23
While there exist other models of implant-associated osteomyelitis11–13, 24–26 the model presented here is unique in that it mimics implant-associated osteomyelitis in the setting of a closed fracture repair. Hematogenous models are excellent examples but they tend to be used to model seeding of an implant (e.g., seeding of an orthopedic implant after a dental procedure) and are thus not well suited for the “acute” setting without the use of trauma or additional foreign objects (e.g., sand).24–26 Hematogenous models also run the risk of systemic consequences and often require higher amounts of inoculum, increasing the risk of morbidity and mortality. Hienz et al.27 reported that in order to achieve an infection rate of 100% an inoculum of 5 × 108 was required and that, without the use of the sclerosing agent, sodium morrhuate, osteomyelitis did not occur consistently. Poultsides et al.23 reported that “variation between 1 × 108 and 4 × 108 CFU/ml induced reproducibly either no infection or acute infection and, consequently septic shock and death, respectively.” Thus, the advantages of our model are that it consistently produced osteomyelitis, did not have any obvious systemic effects and used a smaller inoculum (104 CFU) of bacteria.
The radiographic changes; bacterial isolation from infected femurs and implants; and histopathological changes document the reliable development of osteomyelitis after the inoculation with 104 CFU of S. aureus in this model. Thus, we can accept our hypothesis and further suggest that this model has the potential to be used in evaluating anti-fouling/anti-biofilm implant coatings or other osteomyelitis studies.
This research was supported in part by the AO North America Resident Research Support Program. The authors thank Barb Wicklund, Alexa Hart, Stacy Meola, and Kristina Kiefer for their help in completing this project.