Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Sep 1.
Published in final edited form as: Spine J. 2023 May 27;23(9):1389–1399. doi: 10.1016/j.spinee.2023.05.014

A novel rodent model of chronic spinal implant-associated infection

Trevor DeMourdant a, Christian J Rajkovic a, Jovanna A Tracz a, Alexander Perdomo-Pantoja a,b, Brendan F Judy a, Vaughn N Hernandez a, Jessica Lin a, Julianna L Lazzari a, Dustin A Dikeman c, Nathan K Archer c, Kimberly M Davis d, Oren Gordon e, Timothy F Witham a,*
PMCID: PMC10530089  NIHMSID: NIHMS1919658  PMID: 37247639

Abstract

BACKGROUND CONTEXT:

Bacterial infection of spinal instrumentation is a significant challenge in spinal fusion surgery. Although the intraoperative local application of powdered vancomycin is common practice for mitigating infection, the antimicrobial effects of this route of administration are short-lived. Therefore, novel antibiotic-loaded bone grafts as well as a reliable animal model to permit the testing of such therapies are needed to improve the efficacy of infection reduction practices in spinal fusion surgery.

PURPOSE:

This study aims to establish a clinically relevant rat model of spinal implant-associated infection to permit the evaluation of antimicrobial bone graft materials used in spinal fusion.

STUDY DESIGN:

Rodent study of chronic spinal implant-associated infection.

METHODS:

Instrumentation anchored in and spanning the vertebral bodies of L4 and L5 was inoculated with bioluminescent methicillin-resistant Staphylococcus aureus bacteria (MRSA). Infection was monitored using an in vivo imaging system (IVIS) for 8 weeks. Spines were harvested and evaluated histologically, and colony-forming units (CFUs) were quantified in harvested implants and spinal tissue.

RESULTS:

Postsurgical analysis of bacterial infection in vivo demonstrated stratification between MRSA and phosphate-buffered saline (PBS) control groups during the first 4 weeks of the 8-week infection period, indicating the successful establishment of acute infection. Over the 8-week chronic infection period, groups inoculated with 1 × 105 MRSA CFU and 1 × 106 MRSA CFU demonstrated significantly higher bioluminescence than groups inoculated with PBS control (p = 0.009 and p = 0.041 respectively). Histological examination at 8 weeks postimplantation revealed the presence of abscesses localized to implant placement in all MRSA inoculation groups, with the most pervasive abscess formation in samples inoculated with 1 × 105 MRSA CFU and 1 × 106 MRSA CFU. Quantification of CFU plated from harvested spinal tissue at 8 weeks post-implantation revealed the 1 × 105 MRSA CFU inoculation group as the only group with a signifycantly greater average CFU count compared to PBS control (p = 0.017). Further, CFU quantification from harvested spinal tissue was greater than CFU quantification from harvested implants across all inoculation groups.

CONCLUSION:

Our model demonstrated that the inoculation dosage of 1 × 105 MRSA CFU exhibited the most robust chronic infection within instrumented vertebral bodies. This dosage had the greatest difference in bioluminescence signal from control (p < 0.01), the lowest mortality (0% compared to 50% for samples inoculated with 1 × 106 MRSA CFU), and a significantly higher amount of CFUs from harvested spine samples than CFUs from control harvested spine samples. Further, histological analysis confirmed the reliability of this novel rodent model of implanted-associated infection to establish infection and biofilm formation of MRSA for all inoculation groups.

CLINICAL SIGNIFICANCE:

This model is intended to simulate the infection of instrumentation used in spinal fusion surgeries concerning implant locality and material. This model may evaluate potential antimicrobial and osteogenic biomaterials and investigate the relationship between implant-associated infection and failed fusion.

Keywords: Implant-associated infection, Osteomyelitis, Rodent model, Spinal arthrodesis, Spine infection, Spinal instrumentation

Background

The incidence of infection after spinal instrumentation is estimated between 0.7%−20% across all spinal instrumentation procedures [1], with most implant-associated infections acquired during surgery and presenting within 6 weeks of the procedure [2]. Infection of spinal instrumentation may not only lead to life-threatening bacteremia but also interferes with osteogenesis required to restore the structural integrity of the spine [3]. The most common pathogen identified in spinal implant-associated infections is Staphylococcus aureus (20%−84% of all spinal infections), followed by Streptococci, Enterococci, and anaerobic microorganisms [4]. Methicillin-resistant S. aureus (MRSA) and bacterial formation of biofilms further complicate treatment. Chronic or persistent infection is correlated with the establishment of biofilms on spinal instrumentation, as the metal surface is an ideal surface for bacterial attachment and biofilm formation [5]. Additionally, chronic spinal implant-associated infection may necessitate neurosurgical reintervention for debridement and removal of instrumentation, which is associated with increased health-care costs, morbidity, and length of hospital stay, in addition to decreased patient satisfaction [6].

A promising avenue for providing local and sustained therapeutic concentrations of antibiotics within the instrumented vertebra is the use of antibiotic-loaded bone graft materials [7]. However, the lack of standardized rodent models for establishing spinal implant-associated infections complicates the evaluation of such materials. Rats are commonly used for preclinical spinal fusion models due to their reliability and relatively low cost compared to larger models [8]. Further, intervertebral disks of rats and humans have been shown to have similar mechanical performance, indicating the conservation of tissue properties across the two species [9,10].

To the best of our knowledge, there are three existing murine or rodent models of spinal implant-associated infection. The rat model of spinal implant-associated infection developed by Ofluoglo et al. evaluated infection over an acute period of only 2 weeks postoperatively with a titanium implant and inoculation at only one vertebra unilaterally [11]. The mouse model described by Dworsky et al. proposed a similar infection model where the L4 vertebra was implanted with a stainless steel wire and inoculated unilaterally with different CFUs of MRSA, and infection establishment was quantified using in vivo bioluminescent imaging and post-mortem CFU quantification from implants and adjacent spinal tissue [12]. This model was used in subsequent studies by Zoller et al. and Kelley et al. to investigate infection imaging with fluorescent monoclonal antibodies and infection resolution with vancomycin respectively [13,14]. While this model accommodated these novel imaging and antimicrobial modalities, its complexity was relatively limited compared to the materials and technique used in spinal fusion procedures due to its stainless-steel unilateral implantation at a single vertebra. Therefore, we sought to produce a model that investigated multilevel bilateral implantation of an inoculated titanium implant to simulate clinical spinal fusion implantation procedure and infection more faithfully.

The goal of this study is to develop a rat model of spinal implant-associated infection that mimics the intraoperative placement of pedicle screws and rods in spinal fusion surgery and to evaluate the course of chronic implant-associated infection over the standard fusion period of 6−8 weeks. A 2018 study at eleven hospitals in South Korea showed that 50% of spinal implant-associated infections were late-onset (occurring greater than 30 days after implantation), with 75% of all cases caused by MRSA [15]. Thus to reflect the temporality of the establishment of chronic infection in the clinical setting, we chose an infection period of eight weeks with in vivo violescent imaging of infection establishment and postmortem analysis of implant and tissue bacterial burden. Establishing a standardized, clinically relevant, and reliable rat model of chronic spinal implant-associated infection will allow for the evaluation of novel antimicrobial and osteogenic bone graft materials for use in spinal fusion as well as the investigation of the relationship between implant-associated infection and failed fusion.

Methods

Experimental design

All experiments were performed following Johns Hopkins University School of Medicine Animal Care and Use Committee (Protocol #RA21M151) and Institutional Bio-safety Committee (IBC Registration #P2109210101) approved protocols and procedures. Thirty-two syngeneic female Lewis rats (Charles River Laboratories, Wilmington, MA), 6−8 weeks of age, weighing 100−150 g, were randomly assigned to four experimental groups according to the concentration of MRSA CFU. Titanium implants were inoculated with either 1 × 104 CFU, 1 × 105 CFU, or 1 × 106 CFU of MRSA to determine the concentration of bacterial inoculation necessary to establish chronic implant-associated infection (Table 1).

Table 1.

Experimental grouping

Experimental Group Implant Inoculation No. Animals Mortality Rate During or Within 72 Hours of Surgery
PBS (Control) 1X PBS 8 25%
S. aureus 1 × 104 CFU S. aureus 1 × 104 CFU 8 25%
S. aureus 1 × 105 CFU S. aureus 1 × 105 CFU 8 0%
S. aureus 1 × 106 CFU S. aureus 1 × 106 CFU 8 50%
Open in a new tab

Design of spinal implant

Surgical-grade titanium wire (22-AWG/.34mm2, Grade 1, TEMCo, Fremont, CA) was shaped to mimic pedicle screws with interconnecting rods used in spinal fusion surgery, permitting insertion into adjacent L4 and L5 vertebrae (Figure 1). Prior to initiating surgeries, three 6-week-old Lewis rat spines were harvested and a CT scan was used to determine the optimal size and shape of the implant in addition to the correct angle of entry in order to place the implant in the vertebral body through the pedicle. The size of the implant was determined by measurement of the average distance between the L4−L5 vertebrae, and the angles were measured by referencing CT scans of the spines of these rats to determine optimal placement in the vertebral bodies. The final design of the titanium implant was staple-like in shape with each prong measuring 3 mm in length and the base measuring 7 mm in length. The prongs were then bent 35 degrees clockwise from the base to match the angle of the intended pedicles from the entry point on the base of the transverse processes in line with vertebral bodies in the rat model.

Fig. 1.

Fig. 1.

Open in a new tab

Titanium implant anchored across L4 and L5 vertebral bodies in rodent model. The implant served to mimic pedicle screw placement in clinical spinal fusion surgeries and provided an abiotic surface for bacterial growth and biofilm development after inoculation with S. aureus.

Preparation of bioluminescent methicillin-resistant S. aureus

All work with bacteria was performed in BSL-2 certified facilities using aseptic technique in fume hoods. The bioluminescent USA300 S. aureus strain SAP231 (generously provided by the Nathan K. Archer Lab at the Johns Hopkins Department of Dermatology), which is derived from the community-acquired MRSA strain NRS384 isolated from an outbreak in the Mississippi prison system and contains a stably integrated luxABCDE operon from Photorhabdus luminescens that confers bioluminescence to metabolically active bacteria, was used in all the experiments [16]. This strain has been reported to develop biofilms and adherence to titanium orthopedic implants [17].

Three bacterial colonies were placed in tryptic soy broth (TSB) and cultured for 16 hours at 37°C at 240 rpm. The bacterial overnight culture was then diluted 1:50 into fresh TSB to subculture under the same conditions for another 2 hours before harvesting mid-logarithmic phase bacteria for quantification by absorbance. The solution was diluted until an optical reading of ~ 0.6 was reached, equivalent to 2 × 108 CFU. This concentration was then diluted to obtain cultures of 1 × 108, 1 × 107, and 1 × 106 CFU per milliliter (liquid suspension in phosphate-buffered saline [PBS]) for implantation. Inocula of 10 microliters (μl) were used to achieve 1 × 106 CFU, 1 × 105 CFU, or 1 × 104 CFU per inoculum respectively. 100 μl of a 1 × 103 CFU/mL dilution were plated for CFU verification. The inocula preparations were placed in separate tubes on ice to preserve the approximate CFU concentration for the duration of the surgical procedures.

Surgical procedure

Posterolateral instrumentation surgery at the L4−L5 level was performed on thirty-two female Lewis rats divided into four experimental groups: [1] PBS [Control] (n=8); [2] S. aureus 1 × 104 CFU (n=8); [3] S. aureus 1 × 105 CFU (n=8).; and, [4] S. aureus 1 × 106 CFU (n=8). As previously described, rats were initially weighed, and anesthesia was induced via a single intraperitoneal injection of a solution containing ketamine hydrochloride (75 mg/kg; 100 mg/ml), and xylazine (7.5 mg/kg; 100 mg/ml) in a sterile saline [18]. The paw-pinch test was then performed to determine the level of sedation. The dorsal lumbar region was shaved, prepared with alcohol and povidone-iodine, and draped under sterile conditions. Sterile gowns, gloves, caps, and masks were used by all surgical personnel. All surgical procedures were performed under an operating microscope. A dorsal midline skin incision was made centered over the L4−L5 spinous process, and a self-retaining retractor was utilized to retract the skin edges. Two paramedian fascial incisions were made, and an intermuscular plane was established between the multifidus and longissimus muscles to expose the transverse processes (TPs) of L4 and L5 as well as the intertransverse membrane. TP cannulation was performed with a 0.6-mm motorized burr (Ideal Microdrill, Cellpoint Scientific Inc, Gaithersburg, MD, USA), following insertion of a 26-gauge needle to verify the creation of small channels inside the TPs, into which the titanium implant was placed. Rats were then taken inside a fume hood, and implants were inoculated with 10 mL of either PBS or MRSA at varying concentrations (Table 1, Fig. 1). The wound was closed in a layered fashion of the fascia and skin using absorbable sutures (Poly-sorb, Covidien, Dublin, Ireland).

In vivo imaging of bioluminescent S. aureus

An in vivo Imaging System (IVIS) was utilized to noninvasively evaluate the presence of infection (Lumina III IVIS system, PerkinElmer, Hopkinton, MA, USA). The MRSA strain utilized contains a stably integrated lux-ABCDE operon which confers luminescence to metabolically active bacteria [16]. Bioluminescent bacteria within animal subjects can be observed using the Lumina III IVIS. The Lumina III IVIS software produces an image with a color scale corresponding to the bioluminescence produced by metabolically active bacteria in each animal after an imaging scan. Living image software is then utilized to quantify the photons/second expressed by the bacteria. This method of measuring bacterial load in vivo is relatively common and was used in the spinal implant infection models by Dworsky et al., Zoller et al., and Kelley et al. [12−14] Rats were imaged for 2 minutes immediately following suture closure, on postoperative day 3, and weekly thereafter. Bacterial burden was quantified using Living Image software where a region of interest (ROI) with uniform dimensions was defined to encompass the thoracolumbar region of each rat. Total flux was evaluated in each ROI.

CFU quantification from spinal implant, bone, and soft tissue

Two rats from each experimental group were euthanized and spines were harvested at the end of postoperative week 8. Implants were removed from the spine and placed in 0.3% Tween TSB solution, vortexed for 2 minutes, then sonicated for 10 minutes. After 1 minute of additional vortexing, 100 uL of solution was plated on tryptic soy agar (TSA) overnight for CFU quantification.

The remaining spinal tissue was placed in PBS on ice until implant processing was complete. These spines were then manually homogenized using sterile bone rongeur forceps. After homogenization the average volume of each spine sample (tissue + 10 mL PBS) was 15 mL, from which 100 μL was taken and serially diluted to 1:10 and 1:100. The 1:10 and 1:100 dilutions for each rat were plated overnight for CFU quantification.

Histological processing

Spines were harvested 8 weeks postoperatively and fixed in 4% paraformaldehyde until dissection was performed. Paraspinal muscle and soft tissue were removed while preserving the infected ROI for histological examination. Following dissection, the spines were immersed in Rapid Bone Decalcifier solution (Fisher Scientific, Waltham, MA, USA) for 72 hours. Samples were then cut to approximately 3.5 cm in length for histological processing. Each spine was embedded in paraffin before sectioning and staining with hematoxylin and eosin (H&E) and Masson Trichrome. Gold slides were used to improve adherence of the osseous tissue to the 10 μm slides throughout sectioning and fixation. After slides were stained, they were scanned at 40x magnification for analysis.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 9 software (GraphPad, La Jolla, CA, USA). One-way ANOVA was performed with pairwise comparison between the PBS Control and other experimental groups. A p-value of less than 0.05 was considered statistically significant.

Results

Surgical outcomes

After surgery, rats were closely monitored for any sign of hemiparesis, nerve palsy, or infection as well as any changes in general condition. Eight rats from our sample of 32 did not survive the 8-week study. Of the eight nonsurviving rats, four were inoculated with 1 × 106 MRSA CFU, our highest inoculation dosage. Five of the eight total rats that died prematurely died during surgery, and the remaining three rats died approximately 72 hours postoperation, all of which were inoculated with 1 × 106 MRSA CFU. The rats that died approximately 72 hours postsurgery were not fixed in paraformaldehyde in a timely fashion, thus these specimens were deemed to be not reportable due to sample degradation. Further, two of the six surviving animals inoculated with PBS were found to be contaminated postsurgery. Contamination was defined as the unintentional presence of enhanced MRSA signal within the vertebrae of the two control rats, as indicated by increased MRSA bioluminescent signal. Contaminated animals were omitted from analysis and thus are not reflected in statistical results.

Establishment of infection in vivo

In vivo bioluminescence, post mortem tissue CFU quantification, and histological analysis 8 weeks postoperatively demonstrated that MRSA-infected experimental groups exhibited significant bacterial burden localized to the implanted hardware, indicative of successful establishment of chronic infection (Fig. 2–4, ).

Fig. 2.

Fig. 2.

Open in a new tab

In vivo Bioluminescent flux by experimental group. Panel A quantifies bioluminescent flux by IVIS scan over an 8-week infection period for all inoculation groups. Total flux is graphed on a logarithmic scale for ease of comparison between groups. Error bars are obtained from standard deviations of average luminescence scans per group (n = 4 for the PBS Control inoculation group, n = 6 for the 1 × 104 inoculation group, n = 8 for the 1 × 105 inoculation group, n = 4 for the 1 × 106 inoculation group). Contaminated PBS controls from the first surgery day and the week 2 outlier from the third surgery group were omitted from this graph for ease of comparison. Rats inoculated with PBS exhibited less luminescence than infected groups over weeks 2−4 and 6−8 of the infection period. Panel B depicts the average luminescence value for each rat and the mean + SEM of each experimental group. One-way ANOVA pairwise comparison analysis revealed the 1 × 105 CFU inoculation group had significantly higher average luminescence per scan than the PBS control group, as denoted by a double asterisk (p = 0.009). The 1 × 106 inoculation also showed significantly higher average luminescence per scan than the PBS control group after one-way ANOVA pairwise comparison denoted by a single asterisk (p = 0.041). The 1 × 104 inoculation group did not show significantly higher average luminescence per scan than the PBS control group (p = 0.673).

Fig. 4.

Fig. 4.

Open in a new tab

Gross specimen depicting osteomyelitis in 1 × 105 CFU implanted rat. A) Ventral aspect of the spine with transverse processes numbered according to corresponding vertebrae. Arrows indicate abscesses protruding from vertebral bodies into the transverse processes at L4 and L5. B) Dorsal aspect of the spine of a rat inoculated with 1 × 106 MRSA CFU. Circle indicates the area where infection and implant migrated from the intervertebral space to pierce the epidermis.

Both acute and chronic infections were established in our model among animals inoculated with 1 × 105 MRSA CFU as indicated by the significantly greater bioluminescence exhibited by that inoculation group compared to the group inoculated with PBS control over the acute infection period 2 weeks (p = 0.008) and over the entire 8-week postoperative period, independent of time (p = 0.009) (Fig. 2A). Animals inoculated with 1 × 106 MRSA CFU exhibited significantly greater bioluminescence compared to the group inoculated with PBS control over only the chronic infection period (p = 0.041), indicating the establishment of chronic infection. Bioluminescence in PBS controls remained significantly lower than in infected rats for the first 5 weeks and during week 8. During weeks 5−8, a categorical drop in luminescence was noted in all inoculation groups (Fig. 2A).

To confirm the sustained greater establishment of infection of MRSA inoculated groups compared to the PBS control group, the average luminescence per scan was calculated for each rat using the IVIS data, and these values were organized into experimental inoculation groups for analysis. The highest bioluminescent signal as measured by total flux (photons/second) was noted in samples inoculated with 1 × 105 CFU when compared to PBS control (p < 0.01) in the first 4 weeks, but by the end of the 8-week period, the differences between inoculation groups was less pronounced (Fig. 2A). The 1 × 106 inoculation also showed significantly higher average luminescence than the PBS control (p = 0.041). However, rats inoculated with 1 × 105 CFU demonstrated the greatest sustained infection over the 8 weeks, independent of time (p = 0.009) (Fig. 2B).

Histological evaluation

Sample spines and adherent soft tissue were harvested, and implants were retrieved prior to histological examination, which revealed the presence of abscesses with vertebral involvement in all MRSA-inoculated spines. The PBS inoculated sample depicts mild myeloid and lymphoid cellular infiltration in the target vertebrae as demonstrated by darker staining caused by greater cell density during wound healing, however there is no evidence of abscess formation (Fig. 3). No significant difference in abscess formation was noted between spines inoculated with 1 × 105 CFU and 1 × 106 CFU. Visual ventral inspection of a gross specimen inoculated with 1 × 105 CFU revealed confirmed osteomyelitis at the L4 and L5 vertebrae observed in histology (Fig. 4A). Visual dorsal inspection of a gross specimen inoculated with 1 × 106 MRSA CFU revealed migration of infection and spinal implant through the skin (Fig. 4B).

Fig. 3.

Fig. 3.

Open in a new tab

Histological and gross examination of lumbar spines of rats with S. aureus inoculated implants 8 weeks postoperatively. Histological analysis was performed using Masson’s Trichrome Stain on harvested spines representative of bacterial growth in each experimental group. A) PBS Control: The L4 (left) and L5 (right) vertebrae depict cell infiltration (white arrows), likely due to the presence of a foreign body (spinal implant) and trauma of implantation. However, the absence of abscesses in this sample is indicative of a lesser establishment of infection compared to samples inoculated with MRSA, which is consistent with this sample’s noninfectious inoculation dose. B) MRSA 1 × 104 CFUs: Sample demonstrates presence of established infection through the formation of abscesses (white arrows). C) MRSA 1 × 105 CFUs: Sample demonstrates larger and more numerous abscesses (white arrows). D) MRSA 1 × 106 CFUs: Abscesses indicative of severe vertebral infection are apparent throughout the spine sample (white arrows).

CFU quantification of harvested implants and spinal tissue

Harvested sonicated implants and homogenized spinal tissue were plated separately on TSA for CFU quantification, which verified bacterial presence in MRSA-infected experimental groups and bacterial absence in the PBS control group (Fig. 5). CFUs grown from harvested implants and tissues from animals inoculated with PBS were not statistically significant from zero (p = 0.344 and p = 0.095 respectively). While a significant difference in CFU grown on TSA plates was noted between implants retrieved from 1 × 104 MRSA CFU and PBS control groups (p = 0.014), there was no significant difference in CFU yield from implants when comparing groups inoculated with 1 × 105 and 1 × 106 MRSA CFU to PBS control (Fig. 5A). The quantity of CFU grown from spinal tissue was higher in the 1 × 105 MRSA CFU than the quantity of CFU grown from spinal tissue in the PBS control group (p = 0.017). However, there was no significant difference in CFU yield from spinal tissue when comparing groups inoculated with 1 × 104 and 1 × 105 MRSA CFU to PBS control (p = 0.121 and p = 0.087 respectively, Fig. 5B).

Fig. 5.

Fig. 5.

Open in a new tab

Implant and spinal tissue CFU quantification 8 weeks postoperatively. Error bars are derived from standard deviations of inoculation groups (n = 2). Panel A depicts the average number of CFU plated and grown from the implants retrieved from each inoculation group. After one-way ANOVA analysis, the quantity of CFU grown from implants retrieved from the 1 × 104 MRSA CFU group was significantly higher than the number of CFU plated from implants retrieved from the PBS control group (p = 0.014). However, the quantity of CFU grown from implants retrieved from the 1 × 105 and 1 × 106 MRSA CFU groups did not differ significantly from the number of CFU grown from implants retrieved from the PBS control group (p = 0.173 and p = 0.171 respectively). Panel B depicts the average quantity of CFU plated and grown from the spinal tissue retrieved from each inoculation group. The number of CFU grown from spinal tissue of the 1 × 105 MRSA CFU group was significantly higher than the quantity of CFU grown from spinal tissue retrieved from the PBS control group (p = 0.017). The number of CFU grown from spinal tissue retrieved from the 1 × 104 and 1 × 106 MRSA CFU groups did not differ significantly from the number of CFU grown from spinal tissue retrieved from the PBS control group (p = 0.121 and p = 0.087 respectively).

Discussion

To maximize the translational capacity of our research on spinal implant-associated infection, we chose to create a model of spinal implant-associated infection that adapted as many clinical characteristics of cases of spinal implant-associated infection present in the literature as possible. While it is common practice for spinal instrumentation to be implanted bilaterally, reviews of spinal implant-associated infection described by Köoderet al., Dapant et al., and Margaryan et al. revealed a median of four, six, and three stabilized segments respectively in patients with infection, indicating the necessity to include implantation of multiple spine vertebrae for a clinically translatable model of spinal implant-associated infection [19−21]. Further, these articles reported that cases of infection most commonly occurred in cases of lumbar stabilization (55.9%, 71.7%, and 45.2% of cases respectively) [19−21], and a retrospective case control study of surgical site infections (SSI) after laminectomy or spinal fusion procedures conducted by Olsen et al. reported that a posterior approach had significantly higher rates of SSIs compared to an anterior approach (OR = 8.2, 95% CI = 2−33.5) [22]. Titanium was chosen for the implant material given both its extremely common use for spinal instrumentation and exceptional bio-compatibility, mechanical properties, corrosion resistance, and a smaller risk of infection compared to stainless steel and polyethyletherketone [6,23−25]. Therefore, our model of spinal implant-associated infection is uniquely suited to investigate clinically translatable antimicrobial therapies against infection due to its posterolateral surgical procedure and bilateral multilevel titanium implantation at the L4 and L5 vertebrae.

In our assessment of in vivo imaging of infection establishment through bioluminescence, we sought to optimize our model to a single inoculation dosage against which future antimicrobial therapies could be compared. When averaging bioluminescence signals across groups for the entire 8-week infection period, animals inoculated with 1 × 105 MRSA CFU and animals inoculated with 1 × 106 MRSA CFU both exhibited average bioluminescence signals statistically significant from average bioluminescence signals from control animals inoculated with PBS (Fig. 2B). Of note, the luminescence of PBS control animals was not 0 photons/s, likely due to background sample luminescence. However, the average bioluminescence signal from animals inoculated with 1 × 105 MRSA CFU had a p-value less than 0.01 after a one-way ANOVA comparison with average bioluminescence signals from control animals (p = 0.006), while the average bioluminescence signal from animals inoculated with 1 × 106 MRSA CFU only had a p-value less than 0.05 after a one-way ANOVA comparison with average bioluminescence signals from control animals (p = 0.041). In our time-dependent analysis of infection establishment in-vivo (Fig. 2A), animals inoculated with 1 × 104 MRSA CFU and animals inoculated with 1 × 106 MRSA CFU exhibited all four phases of biofilm formation, and animals inoculated with 1 × 105 MRSA CFU exhibited faster adherence and proliferation stages which occurred prior to week one scans [26,27]. Further, animals inoculated with 1 × 106 MRSA CFU had a postoperative mortality rate of 50% compared to 25% for animals inoculated with 1 × 104 MRSA CFU and 0% for animals inoculated with 1 × 105 MRSA CFU. Therefore, we recommend an inoculation dosage of 1 × 105 CFU for our model given its statistically significant difference in average bioluminescence signal from animals in the control group, exhibition of faster biofilm formation in vivo, and low postoperative mortality.

The potential of our model to investigate biofilm dynamics is particularly clinically relevant given its effect on the prognosis and treatment options for patients with spinal implant-associated osteomyelitis. The formation of biofilms in clinical cases of implant-associated osteomyelitis is associated with challenges in infection treatment due to the innate protection of MRSA biofilms from patient immune response and antimicrobial treatment [28,29]. A 2020 cohort study of 93 patients with spinal implant-associated infection performed by Köder et. al reported that patients treated with biofilm-active antibiotics had significantly better outcomes (HR = 0.23; p = 0.017) and two-year infection-free survival rates [84% (95% CI 71−93%) vs. 49% (95% CI 28−61%)] than patients treated without [19]. However, bioluminescent imaging has been more strongly correlated as a measure of the bacterial proliferation stage of biofilm development rather than the bacterial growth stasis exhibited during biofilm maturation[27]. Therefore, our study is limited by the in vivo bioluminescent imaging technique used to quantify infection establishment. Future studies involving our model could more robustly investigate biofilm development and biofilm-active antimicrobial therapies using more sophisticated imaging modalities such as scanning electron microscopy (SEM) or positron emission tomography (PET).

Histological examination revealed differences in the severity of vertebral osteomyelitis as observed by abscess formation and cell infiltration in the L4 and L5 vertebral bodies. Cell infiltration localized to the implant site was noted in histological samples from PBS control and all inoculation groups, associated with the recognition of damage-associated molecular patterns (DAMPs) via pattern recognition receptors (PRRs) [30]. In surgical cases of trauma, damaged tissue cells have been shown to express DAMPs to promote inflammation while causing patient pain even in cases of sterile surgery [31]. The 1 × 105 and 1 × 106 inoculation groups had larger quantities of abscesses indicative of more severe infection. The greater magnitude of expression of bacterial pathogen-associated molecular patterns (PAMPs) in an inoculation dosage with a greater bacterial load and subsequent immune response may explain the reduced average luminescence of the 1 × 106 CFU group overall as the infection elicited high immune reactivity [32]. However, the limited data from this group could also be responsible for this discrepancy.

Postmortem, the mechanism of infection and optimal inoculation dosage were further investigated through the quantification of CFUs grown from harvested spinal implants and spinal tissue. Across all inoculation groups, a greater number of CFUs grew from harvested spinal tissue than from harvested implants. This was evident despite the inoculation of MRSA directly onto the implant during surgery, indicating a migration of infection away from the implant to adjacent spinal tissue. Biofilms commonly spread to uninfected areas of a host through the detachment of microcolonies under shear stress or a genetically programmed response [29]. Single planktonic cells from core microcolonies in biofilms may also migrate by similar mechanisms to infect new areas in a host, and these microbes have enhanced resistance to host immune responses and antimicrobial therapies due to the enhanced genetic variation of interior microcolonies in biofilms [33]. Infection and implant migration is also commonly seen in clinical cases of vertebral osteomyelitis. In a retrospective cohort study performed by Leitner et al., patients who received spinal implant explanation or revision surgery were significantly more likely to have preoperative radiological signs of screw loosening if infection screening positively identified a germ [34]. Migration and loosening of orthopedic implants during infection involve the increased rate of bone resorption caused by biofilm formed in cases of chronic infection [35]. The incidence of spinal surgical site infections and implant loosening is also strongly correlated with the number of prior spinal operations a patient has received [36]. In addition to the disparity between harvested spinal tissue and implants producing CFUs after plating, evidence of implant and infection migration, in some cases breaking the skin (Figure 4B), in our model was evident on gross inspection of harvested spines from animals inoculated with MRSA. Therefore, in optimizing our model to a single inoculation dose, the case for using 1 × 105 MRSA CFUs was further supported as it was the only inoculation dose whose harvested spinal tissues gave a statistically significant different number of CFUs grown compared to the number of CFUs grown from spinal tissues harvested from control animals.

This study is limited by a small sample size of eight rats per group for proof of concept, with group sizes not determined by power calculations. Given the challenge of implanting titanium instrumentation into the L4−L5 pedicles and inoculation of animals with infectious agents, mortality rates among experimental animals further limited the sample size of animals that survived the entire 8-week infection. However, this limitation was mitigated as greatly as possible by maximizing the consistency of surgical procedure through the sterile technique used throughout the surgical and inoculation procedure and microCT analysis of the accuracy of implant placement in practice surgeries prior to the actual experiments. Other limitations include the contamination of two rats inoculated with PBS solution, however, these animals were removed from data analysis. The contamination rate in control animals of 25% seems to exceed the baseline level of infection risk in humans of 0.7%−20%, [1] however it is difficult to determine whether the proximity of bacteria within the same fume hood versus a baseline level of infection risk in this group contributed to such contamination. In clinical settings, the etiology of contamination during spinal fusion procedures is complex and multifactorial. An analysis of the microbiology of spine surgery infection (SSI) using PCR in 351 patients performed by Long et al. revealed an anatomic gradient of pathogens correlating with a higher incidence of gram-positive infection in cervical and thoracic procedures and a higher incidence of gram-negative infection in lumbosacral procedures [37]. This anatomical gradient may guide future work involving our lumbar fusion model in the testing of unique bacteria strains or antimicrobial agents. Long et al. also observed that the majority of cases with MRSA SSI tested negative for MRSA nasal carriage preoperatively, indicating the potential acquisition of MRSA intraoperatively, which is depicted in our animal model [37]. Instrumentation may also serve as a source of contamination in clinical settings. Exposure to open air and a lack of sterile covering for the implant have been correlated with increased rates of implant contamination [38,39], which may also have a role in our animal model during inoculation within the fume hood. There are also limitations inherent to the use of bioluminescent imaging concerning susceptibility to changes in bacterial metabolic activity and resolution for low-density bacterial colonies. Bernthanl et al. reported that inoculations of 500 CFUs or lower did not result in visible bioluminescence signal twenty-four hours postinoculation [40]. Lastly, histological changes noted in vertebrae outside of abscess formation (ie, cell infiltration in the postsurgical implant site) could be a result of surgery and/or trauma caused to tissue by the instrumentation rather than infection, which would also be expected in humans after spinal instrumentation.

The model of chronic spinal implant-associated infection described herein permits further investigation into the pathogenesis of implant-associated vertebral osteomyelitis and the evaluation of novel therapeutic agents such as bactericidal bone graft materials to prevent such infection. Future work may include the (1) investigation of the use of other bacteria implicated in chronic spinal implant-associated infection using this model and (2) investigation into the broader application of such model to evaluate antibiotic-laden implants in other areas of the body, not limited to spine or orthopedic devices. Future studies may also utilize this model to examine methods of antimicrobial delivery for penetration of deep tissue infection, disruption of biofilm formation, and to investigate the relationship between osteomyelitis and failed fusion.

Conclusion

We established a rodent model of chronic spinal implant-associated infection with clinically relevant placement of spinal instrumentation inoculated with methicillin-resistant S. aureus bacteria (MRSA). Our experiments were successful in establishing a rat model of chronic, spinal implant-associated infection, as represented by the presence of persistent chronic infections in all bioluminescent MRSA inoculated groups over the entire 8-week period. Infections were verified and quantified through in vivo bioluminescence of MRSA, CFU quantification from harvested tissue and implants, and histological analysis of harvested sample spines. These infections were simulated via a MRSA inoculated titanium implant mimicking pedicle screws anchored in the vertebral body and spanning a vertebral level. Bioluminescent flux, CFU quantification from harvested spinal tissue, and histological features confirmed the reliability of this model, with groups inoculated with 1 × 105 CFU and 1 × 106 CFU exhibiting higher bacterial colonization in vivo and in postsurgical analysis. Across all samples, CFU quantification was greater from spinal tissue samples rather than spinal implant samples, indicating the migration of infection away from the implants into adjacent spinal tissue. This was evident on gross examination of samples with infections visibly leaving the intervertebral space or breaking the epidermal layer. However, rats inoculated with 1 × 106 CFU more commonly exhibited slower wound healing, scab formation, and wound dehiscence, and a higher mortality rate of 50% compared to 0% in rats inoculated with 1 × 105 CFU. Further, only the 1 × 105 inoculation group exhibited a significantly higher average CFU count from plated spine tissue than the average CFU count from the PBS control spine tissue (p = 0.017), indicating greater development of osteomyelitis throughout the spinal tissue. Additionally, bioluminescence in vivo imaging from animals inoculated with 1 × 105 MRSA CFU indicated faster infection establishment consistent with the phases of biofilm formation, a prevalent issue in clinical cases of vertebral osteomyelitis. Therefore, we recommend inoculation with 1 × 105 CFU for the establishment of chronic spinal implant-associated infection in the model described.

Acknowledgments

Bioluminescent NRS384 (SAP231) S. aureus bacteria were generously provided by the Archer Lab in the Department of Dermatology, Johns Hopkins University School of Medicine.

Funding disclosures

This study was funded in part by grants R01AR073665 (NKA) from the United States National Institutes of Arthritis and Musculoskeletal and Skin Diseases and R01AI146177 (NKA) from the United States National Institute of Allergic and Infectious Diseases.

Declarations of competing interests

Dr. Witham is a medical advisory board member and investor in Augmedics, Inc. He has received grant support from the Gordon and Marilyn Macklin Foundation. Dr. Archer has received previous grant support from Pfizer and Boehringer Ingelheim and is a paid consultant for Janssen Pharmaceuticals. One or more of the authors declare financial or professional relationships on ICMJE-TSJ disclosure forms.

TD: Nothing to disclose. CR: Nothing to disclose. JT: Nothing to disclose. AP: Nothing to disclose. BJ: Nothing to disclose. VH: Nothing to disclose. JL: Nothing to disclose. JL: Nothing to disclose. DD: Nothing to disclose. NA: Grant: NIAMS (R01AR073665) (H, Paid directly to institution), NIAID (R01AI146177) (H, Paid directly to institution); Consulting: Janssen Pharmaceuticals (A); Grants: Pfizer (F, Paid directly to institution). KD: Nothing to disclose. OG: Nothing to disclose. TW: Grant: The Gordon and Marilyn Macklin Foundation, Grant for Research conducted at the Spinal Fusion Laboratory at Johns Hopkins University (F, Paid directly to institution).

Abbreviations:

CFU

colony-forming unit

DAMP

damage-associated molecular pattern

IVIS

in-vivo imaging system

MRSA

methicillin-resistant Staphylococcus aureus

PBS

phosphate-buffered saline

PET

positron emission tomography

PRR

pattern recognition receptors

SEM

scanning electron microscopy

SSI

spine surgery infection

Footnotes

Submission statement

This manuscript is original and has not been submitted elsewhere in part or in whole.

References

  • [1].Schöomig F, Gogia J, Caridi J. Epidemiology of postoperative spinal implant infections. J Spine Surg 2020;6:762–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [2].Lake J, Gordon O. Implant-associated spinal infections in children: how can we improve diagnosis and management? Infect Dis Clin North Am 2022;36:101–23. [DOI] [PubMed] [Google Scholar]
  • [3].Srinivasan D, Terman SW, Himedan M, Dugo D, La Marca F, Park P. Risk factors for the development of deformity in patients with spinal infection. Neurosurg Focus 2014;37:E2. [DOI] [PubMed] [Google Scholar]
  • [4].Tsantes AG, Papadopoulos DV, Vrioni G, Sioutis S, Sapkas G, Benzakour A, et al. Spinal infections: an update. Microorganisms 2020;8:476. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [5].McCarthy H, Rudkin JK, Black NS, Gallagher L, O’Neill E, O’Gara JP. Methicillin resistance and the biofilm phenotype in Staphylococcus aureus. Front Cell Infect Microbiol 2015;5:1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [6].Kasliwal MK, Tan LA, Traynelis VC. Infection with spinal instrumentation: review of pathogenesis, diagnosis, prevention, and management. Surg Neurol Int 2013;4:S392–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [7].Ramey WL, von Glinski A, Jack A, Blecher R, Oskouian RJ, Chapman JR. Antibiotic-impregnated polymethylmethacrylate strut graft as a treatment of spinal osteomyelitis: case series and description of novel technique. J Neurosurg Spine 2020:1–6. [DOI] [PubMed] [Google Scholar]
  • [8].Elliott DM, Sarver JJ. Young investigator award winner: validation of the mouse and rat disc as mechanical models of the human lumbar disc. Spine 2004;29:713–22. [DOI] [PubMed] [Google Scholar]
  • [9].Showalter BL, Beckstein JC, Martin JT, Beattie EE, Espinoza Orías AA, Schaer TP, et al. Comparison of animal discs used in disc research to human lumbar disc: torsion mechanics and collagen content. Spine 2012;37:E900–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [10].Beckstein JC, Sen S, Schaer TP, Vresilovic EJ, Elliott DM. Comparison of animal discs used in disc research to human lumbar disc: axial compression mechanics and glycosaminoglycan content. Spine 2008;33:E166–73. [DOI] [PubMed] [Google Scholar]
  • [11].Ofluoglu EA, Zileli M, Aydin D, Baris YS, Kuҫukbasmaci O, Gonullu N, et al. Implant-related infection model in rat spine. Arch Orthop Trauma Surg 2007;127:391–6. [DOI] [PubMed] [Google Scholar]
  • [12].Dworsky EM, Hegde V, Loftin AH, Richman S, Hu Y, Lord E, et al. Novel in vivo mouse model of implant related spine infection. J Orthop Res 2017;35:193–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [13].Zoller SD, Park HY, Olafsen T, Zamilpa C, Burke ZD, Blumstein G, et al. Multimodal imaging guides surgical management in a preclinical spinal implant infection model. JCI Insight 2019;4:e124813. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [14].Kelley BV, Zoller SD, Greig D, Hori K, Cevallos N, Ishmael C, et al. In vivo mouse model of spinal implant infection. J Vis Exp 2020; 10.3791/60560. 10.3791/60560. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [15].Cho O-H, Bae I-G, Moon SM, Park SY, Kwak YG, Kim B-N, et al. Therapeutic outcome of spinal implant infections caused by Staphylococcus aureus: a retrospective observational study. Medicine 2018;97:e12629. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [16].Plaut RD, Mocca CP, Prabhakara R, Merkel TJ, Stibitz S. Stably luminescent Staphylococcus aureus clinical strains for use in bioluminescent imaging. PLoS One 2013;8:e59232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [17].Wang Y, Cheng LI, Helfer DR, Ashbaugh AG, Miller RJ, Tzomides AJ, et al. Mouse model of hematogenous implant-related biofilm infection reveals therapeutic targets. Proc Natl Acad Sci U S A 2017;114:E5094–102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [18].Perdomo-Pantoja A, Holmes C, Cottrill E, Rindone AN, Ishida W, Taylor M, et al. Comparison of freshly isolated adipose tissue-derived stromal vascular fraction and bone marrow cells in a posterolateral lumbar spinal fusion model. Spine 2021;46:631–7. [DOI] [PubMed] [Google Scholar]
  • [19].Köoder K, Hardt S, Gellert MS, Haupenthal J, Renz N, Putzier M, et al. Outcome of spinal implant-associated infections treated with or without biofilm-active antibiotics: results from a 10-year cohort study. Infection 2020;48:559–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [20].Dapunt U, Bürkle C, Günther F, Pepke W, Hemmer S, Akbar M, et al. Surgical site infections following instrumented stabilization of the spine. Ther Clin Risk Manag 2017;13:1239–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [21].Margaryan D, Renz N, Bervar M, Zahn R, Onken J, Putzier M, et al. Spinal implant-associated infections: a prospective multicentre cohort study. Int J Antimicrob Agents 2020;56:106116. [DOI] [PubMed] [Google Scholar]
  • [22].Olsen MA, Mayfield J, Lauryssen C, Polish LB, Jones M, Vest J, et al. Risk factors for surgical site infection in spinal surgery. J Neurosurg 2003;98:149–55. [PubMed] [Google Scholar]
  • [23].Litak J, Szymoniuk M, Czyżewski W, Hoffman Z, Litak J, Sakwa L, et al. Metallic implants used in lumbar interbody fusion. Materials 2022;15:3650. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [24].Karczewski D, Schnake KJ, Osterhoff G, Spiegl U, Scheyerer MJ, Ullrich B, et al. Postoperative spinal implant infections (PSII)-a systematic review: what do we know so far and what is critical about it? Global Spine J 2022;12:1231–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [25].Arens S, Schlegel U, Printzen G, Ziegler WJ, Perren SM, Hansis M, et al. Influence of materials for fixation implants on local infection. An experimental study of steel versus titanium DCP in rabbits. J Bone Joint Surg Br 1996;78:647–51. [PubMed] [Google Scholar]
  • [26].Foster TJ, Geoghegan JA, Ganesh VK, Höök M. Adhesion, invasion and evasion: the many functions of the surface proteins of Staphylococcus aureus. Nat Rev Microbiol 2014;12:49–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [27].Nishitani K, Sutipornpalangkul W, de Mesy Bentley KL, Varrone JJ, Bello-Irizarry SN, Ito H, et al. Quantifying the natural history of biofilm formation in vivo during the establishment of chronic implant-associated Staphylococcus aureus osteomyelitis in mice to identify critical pathogen and host factors. J Orthop Res 2015;33:1311–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [28].Ha K-Y, Chung Y-G, Ryoo S-J. Adherence and biofilm formation of Staphylococcus epidermidis and Mycobacterium tuberculosis on various spinal implants. Spine 2005;30:38–43. [DOI] [PubMed] [Google Scholar]
  • [29].Brady RA, Leid JG, Calhoun JH, Costerton JW, Shirtliff ME. Osteomyelitis and the role of biofilms in chronic infection. FEMS Immunol Med Microbiol 2008;52:13–22. [DOI] [PubMed] [Google Scholar]
  • [30].Goodman SB, Konttinen YT, Takagi M. Joint replacement surgery and the innate immune system. J Long Term Eff Med Implants 2014;24:253–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [31].Papa A, Di Dato MT, Buonavolonta P, Saracco E, Salzano AM, Casale B. Clinical management of Il-6 driven cytokine storm related to COVID-19 in a patient with recent spinal cord stimulator implants: a case report. Anesth Pain Med 2020;10:e104151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [32].Yang D, Wijenayaka AR, Solomon LB, Pederson SM, Findlay DM, Kidd SP, et al. Novel insights into Staphylococcus aureus deep bone infections: the involvement of osteocytes. MBio 2018;9:e00415–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [33].Sauer K, Camper AK, Ehrlich GD, Costerton JW, Davies DG. Pseudomonas aeruginosa displays multiple phenotypes during development as a biofilm. J Bacteriol 2002;184:1140–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [34].Leitner L, Malaj I, Sadoghi P, Amerstorfer F, Glehr M, Vander K, et al. Pedicle screw loosening is correlated to chronic subclinical deep implant infection: a retrospective database analysis. Eur Spine J 2018;27:2529–35. [DOI] [PubMed] [Google Scholar]
  • [35].Mörmann M, Thederan M, Nackchbandi I, Giese T, Wagner C, Hänsch GM. Lipopolysaccharides (LPS) induce the differentiation of human monocytes to osteoclasts in a tumour necrosis factor (TNF) alpha-dependent manner: a link between infection and pathological bone resorption. Mol Immunol 2008;45:3330–7. [DOI] [PubMed] [Google Scholar]
  • [36].Bratschitsch G, Puchwein P, Zollner-Schwetz I, Sadoghi P, Radl R, Leithner A, et al. Spinal surgery site infection leading to implant loosening is influenced by the number of prior operations. Global Spine J 2022;12:458–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [37].Long DR, Bryson-Cahn C, Pergamit R, Tavolaro C, Saigal R, Chan JD, et al. 2021 young investigator award winner: anatomic gradients in the microbiology of spinal fusion surgical site infection and resistance to surgical antimicrobial prophylaxis. Spine 2021;46:143–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [38].Agarwal A, Lin B, Elgafy H, Goel V, Karas C, Schultz C, et al. Updates on evidence-based practices to reduce preoperative and intraoperative contamination of implants in spine surgery: a narrative review. Spine Surg Relat Res 2020;4:111–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [39].Bible JE, O’Neill KR, Crosby CG, Schoenecker JG, McGirt MJ, Devin CJ. Implant contamination during spine surgery. Spine J 2013;13:637–40. [DOI] [PubMed] [Google Scholar]
  • [40].Bernthal NM, Stavrakis AI, Billi F, Cho JS, Kremen TJ, Simon SI, et al. A mouse model of post-arthroplasty Staphylococcus aureus joint infection to evaluate in vivo the efficacy of antimicrobial implant coatings. PLoS One 2010;5:e12580. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES