There are many surgical models of bacterial osteomyelitis performed in murine models. Models typically utilize long bones, although alternatives such as vertebral models are also reported (57). An extensive review of murine models, including the goal, method, and bacterial inoculum used in each study, was recently published by Guarch-Pérez et al. (58). One approach used by multiple investigators was described in 2008 as a model to assess intramedullary response to titanium particles (59). This surgical approach is accomplished by creating a medial parapatellar arthrotomy to access the femur. Once accessed, a defect extending to the medullary cavity of the femur is created. Kirschner wire (K-wire) is inserted into the femoral medullary canal and penetrated into the patellofemoral joint space. Bacterial inoculation occurs via direct application of a bacterial suspension and the surgical site is closed (48, 60, 61). This model was recently adapted and modified to model shoulder implant infections (62). In this study, investigators were able to reliably induce bacterial osteomyelitis using a bioluminescent strain of S. aureus and were able to track infection with radiographs and bioluminescent imaging (BLI). Another surgical approach that is utilized in various forms by many investigators is described well by Funao et al. (63). Much of this approach is similar to that described above; the distal portion of the femur is exposed surgically, and a 0.5 mm drill hole is created to expose the medullary canal of the femur. Rather than placing an implant, bioluminescent S. aureus is inoculated directly into the defect. The defect is then sealed with bone wax and the surgical site is closed. Another unique model of murine bacterial osteomyelitis is the hematogenous model described by Horst et al. (52). This model does not involve surgical manipulation or placement of foreign material. Instead, mice received one injection of S. aureus in phosphate-buffered saline (PBS) via the lateral tail vein. This model was created to closely mimic both acute and chronic hematogenous bacterial osteomyelitis and is unique in that it does not require additional bone injury. These approaches highlight the various methods available to induce bacterial osteomyelitis and the subtleties between the various models.

While arguably each investigation into bacterial osteomyelitis provides information on pathogenesis, there are experiments designed to evaluate specific questions regarding the pathogenesis of bacterial osteomyelitis (64). One such experiment, described by De Mesy Bentley et al. (65), utilized two murine long bone infection models and captured groundbreaking transmission electron microscopy (TEM) images of S. aureus invading and residing within the osteocyte lacuno-canalicular network (OLCN) of live bone. Staphylococcus aureus cells are thought to be protected while within the canaliculi system, as immune cells are likely too large to successfully access this area of the body. Therefore, these findings offer insight into the ability of S. aureus to evade the host immune system and cause latent and recurrent osteomyelitis. Zoller et al. (40), established and utilized a murine model of bulk allograft infection to expand upon the findings of de Mesy Bentley et al. by investigating the mechanisms of immune system evasion by S. aureus, specifically microarchitecture of implant surfaces as a potential factor in increased bacterial colonization. Staphylococcus aureus was discovered within allograft cortical haversian canals and submicron canaliculi within the native mouse femur. Results indicated that bulk allograft implant material was more susceptible to bacterial infection even at low bacterial inoculums compared to stainless steel implants. This finding suggests that implant microarchitecture is incredibly important and may offer bacteria a submicron reservoir to evade clearance by the immune system. The work of Masters et al. (66) expanded upon these findings by investigating the role of S. aureus cell wall synthesis machinery and surface adhesins in OLCN invasion. The authors established a model of bacterial osteomyelitis by placing stainless steel pins inoculated with various mutant strains of S. aureus into the medial tibia of mice. Results showed significant changes in OLCN invasion, abscess formation and pathogenic bone loss with the deletions of penicillin binding protein 3 and 4 (PBP3, PBP4) and autolysin (Atl), indicating that cell wall synthesis machinery can modulate S. aureus' pathogenesis in osteomyelitis.

While there are multiple reports of utilizing BLI and in vivo micro-CT in murine models (63, 67, 68), these reports are often geared toward improving the in vivo modeling system rather than improving diagnostic capabilities for clinical patients (63). Recently, however, Isogai et al. (69) performed plasma metabolome analysis in a model of murine osteomyelitis caused by S. aureus and identified 12 metabolites as candidate positive biomarkers and two candidate negative biomarkers for osteomyelitis. Novel plasma biomarkers are aimed to improve the early diagnosis of osteomyelitis. Improvement in the early diagnosis of osteomyelitis is of great interest, as there are currently many challenges in obtaining a swift and specific diagnosis in clinical patients.

A major goal of in vivo osteomyelitis work is to evaluate novel treatment strategies and investigate potential efficacy for clinical use. There are many investigations focused on various combinations or applications of antibiotics for clearance of osteomyelitis (48, 70–72). Jørgensen et al. modeled the particularly challenging situation of biofilm presence upon orthopedic implants. They investigated the efficacy of rifampicin-containing combinations of antimicrobials compared with non-rifampicin-containing combinations of antimicrobials in reducing bacterial counts or clearing infection. Results indicated that combinations of antimicrobials that included rifampicin, as well as the combination of daptomycin and linezolid, were more effective in reducing bacterial burden than combinations not containing rifampicin (70). There are also many investigations into novel therapeutics (73–75). Wang et al. utilized a model of S. aureus hematogenous orthopedic implant infection to identify specific virulence factors to be translated into therapeutic targets. This work identified two key pathogenic factors, anti-α-toxin (AT) and anti-clumping factor A (ClfA) and demonstrated markedly improved efficacy in infection treatment utilizing human anti-AT/anti-ClfA combination therapy (50). Similarly, Yokogawa et al. (76) created a novel murine one-stage revision model of methicillin-resistant S. aureus (MRSA) implant-associated osteomyelitis. This model facilitated discovery of synergistic activity of vancomycin and anti-glucosaminidase (Gmd). Identification of alternative therapeutics is important, as medical device implantation continues to increase and antimicrobial resistance (AMR) is becoming increasingly prevalent.

Murine models are particularly helpful to researchers investigating bacterial osteomyelitis. Main attractions of the mouse model include the small size, economics, and genetic and molecular tools that are available to tailor murine strains and facilitate a wide array of investigations. Indeed, mouse strain selection is of paramount importance as strains contain significant differences from one another. Investigators should consider the primary research objective of the model to guide strain selection and ensure research objectives can be accomplished appropriately. This pillar of model development is highlighted nicely by Dworsky et al. (57). These advantages make mice attractive for investigations into pathogenesis and proof of concept models (58, 72). Also, mice allow for certain longitudinal monitoring techniques, such as BLI and in vivo microCT. Longitudinal monitoring is an asset that adds strength and clarity to data collection as individuals can be compared to themselves over multiple timepoints. While mice can mimic the human inflammatory response of osteomyelitis (58), their bone structure and bone remodeling process is less similar to humans than other animal models provide (77). As a result of the mouse's small size, complex and multi-stage surgical procedures are not impossible, but are challenging to perform. This small size also prohibits the investigation and translation of implants intended for human use. Additionally, serial blood collection is limited by volume and frequency. When considering the benefits and limitations of murine models, it can be concluded that mice are an excellent tool for early investigations from in vitro to in vivo modeling and proof of concept work.

Rats provide a variety of models that produce well-characterized and reliable bacterial osteomyelitis. Significant historical developments have previously been described (36, 64, 78, 79). Currently, the most popular rat models are of long bone osteomyelitis and most often utilize the tibia (17, 22, 80–83) or femur (84–90). Long bone models rely on mechanical trauma, placement of foreign bodies, or creation of fractures, all typically with concurrent sealing of the defect area with bone wax to contain bacterial inoculums and prevent concomitant soft tissue infection. Alternative models include mandibular models (91), vertebral models (92), joint prosthesis models (93), and hematogenous models (94, 95). Hematogenous models required additional surgical manipulations to successfully establish osteomyelitis. This may be a result of the rat's ability to respond to acute infection, which can rapidly clear peripheral infection and may complicate infection models (78). Although reports of rat osteomyelitis models exist, a comprehensive review of these models is lacking. In this review, we present a detailed summary of rat osteomyelitis models that were utilized to inform this review (Table 1).

Similar to murine models, rat models can be utilized for investigations into pathogenesis. Rat models have facilitated valuable discoveries, including investigations of virulence factors associated with S. aureus biofilms and the ability of S. aureus to function as an intracellular pathogen. Biofilms are well recognized as a source of recalcitrant bacteria that can impair antibiotic treatment of osteomyelitis and cause persistent or recurrent osteomyelitis, particularly when orthopedic implants are in place (13, 31, 111). Two studies that have pursued the in vivo investigation of biofilm virulence factors and genetic components in rats include the investigation by Wu et al. (107), which demonstrated that overexpression of ASyycG led to a reduction in biofilm formation and in vivo pathogenicity of MRSA in a model of rat tibial osteomyelitis; as well as the investigation by Sahukhal et al. (110) who utilized a model of implant-associated osteomyelitis. This investigation demonstrated that deletion of the msaABCR operon of S. aureus (USA300 LAC) resulted in defective biofilm production and reduced severity of bacterial osteomyelitis. The capability of S. aureus to function as an intracellular pathogen is considered to be a mechanism of immune system evasion and a source of recurrent, persistent osteomyelitis (31) and is supported by in vitro evidence (112, 113). Based on that in vitro evidence, Hamza et al. investigated and confirmed the ability of purely intracellular S. aureus to induce osteomyelitis in a rat model (10).

Similar to murine models, rat models have allowed for improvements in diagnostic or longitudinal monitoring capabilities in experimental models. Examples of these improvements include the findings of Stadelmann et al. (99), who demonstrated the use of in vivo microCT to longitudinally monitor bacterial osteomyelitis in a rat tibial model, thus offering a method to limit numbers of animals needed for experiments and to add strength to collected data. Also, Aktekin et al. evaluated the utility of available scoring systems for the radiographic evaluation of experimental osteomyelitis. Authors utilized a tibial model of osteomyelitis and evaluated serial radiographs throughout their study period, ultimately concluding that it is best to evaluate and report each radiograph individually, rather than appointing a numerical grade from a previously published grading scale (17). This is a valuable report for experimental studies, and with appropriate radiographic interpretation, is likely to add strength to radiograph assessments. An improvement to in vivo studies that holds potential to translate into human medicine is the investigation into various tracers for positron emission tomography (PET) to successfully image osteomyelitis and differentiate between bone infection and bone healing (51). The work investigating PET tracers indicated that Gallium-68 (⁶⁸Ga), did not accumulate in healing bone, only infected bone. This work brings interest to the use of ⁶⁸Ga and PET for clinical patients, although further work is needed to clarify use and safety concerns. Another interesting foray into improving diagnostics for clinical patients was completed by Deng et al. who described the potential use of extracellular vesicles (EVs) as a diagnostic marker for acute osteomyelitis (109).

Rats are recognized to be more resilient than mice and therefore are well suited to investigations into therapeutic strategies, such as antibiotic trials. Indeed, there are many investigations into antibiotic therapies. These include therapeutic efficacy assessments of systemic antibiotics administered solo or in combination (98, 100, 103, 105), investigations of local antibiotic delivery systems (80, 83, 88) and antibiotics in combination with alternative therapies such as omega-3 fatty acid supplementation (108). There also are investigations into novel therapeutic strategies such as the use of photodynamic therapy (PDT) to treat contaminated orthopedic implants and minimize reliance on antibiotic therapy to clear implant associated bacterial osteomyelitis (13). Recently, Cobb et al. (106) investigated the feasibility of utilizing a bacteriophage to mitigate bacterial osteomyelitis, biofilm, and soft tissue infection.

Rat models are a valuable animal resource in the study of osteomyelitis. They provide similar benefits to mice, including small size, economics, ease of housing and handling, and well-characterized strains that provide appropriate uniformity and enable study of disease pathophysiology relevant to that seen in people (97). Rats have the ability to tolerate sustained, high dose antibiotic therapy (97). While larger than mice, rats remain too small for assessment of orthopedic hardware for human use, and multi-step revision procedures, although not impossible, remain challenging. Uniquely, the rat is one of few ideal species for modeling of mandibular osteomyelitis (91, 95) because of their size, anatomy, and general hardiness. Therefore, the strength of rat models lies within the ability to investigate pathogenesis and pursue initial investigations into therapeutic strategies to further understand in vitro data and gain in vivo knowledge prior to utilizing a larger animal model.

Rabbits provide many useful and reliable models of bacterial osteomyelitis. The systematic review by Reizner et al. (78) details significant historical developments and the review by Bottagisio et al. provides a thorough overview of model development and utility (114). Historically and currently, the most utilized models are long bone models, including tibial (115–121), femoral (49, 122–124), and radial (125–129). Alternative models such as joint prostheses (130, 131), mandibular defects (132), vertebral models (133–136) and implant infection via hematogenous seeding (137) exist. Induction of osteomyelitis among these various models can be accomplished via mechanical trauma, either defect (117) or fracture (49) creation and bacterial contamination with or without application of a sclerosing agent or foreign body placement (115), or through placement of contaminated implants (49, 129). Bone wax may be used to seal defect areas and prevent bacterial leakage and concomitant soft tissue infection (116). A benefit of rabbits compared to smaller models such as mice and rats is the improved ability to model chronic osteomyelitis (138) and perform revision procedures such as debridement (23, 116), which improves the capabilities of modeling human osteomyelitis and therapy. Rabbits offer a distinct advantage in studying bone disease because full segmental defects of the radius can be created without the need to stabilize the bone using orthopedic implants.

Majority of reports into pathogenesis utilize well-characterized and reproducible rabbit models and are related to the capabilities of various bacterial species and strains (117, 121, 128) to induce osteomyelitis, as opposed to mechanistic work that more often is performed in murine and rat models. For example, Gahukamble et al. (117) describe an investigation into the abilities of Staphylococcus lugdunensis (S. lugdunensis) and Propionibacterium acnes (P. acnes) to establish osteomyelitis in a model that was previously characterized with a strain of S. aureus isolated from an infected human hip prosthesis (139). Results indicated that both organisms could induce osteomyelitis and described varying severity and clinical presentation. This work again emphasizes the importance of considering model development and bacterial strain selection during experimental design.

Similar to murine and rodent models, there are studies aimed to improve the longitudinal monitoring of experimental osteomyelitis in rabbit models (21, 115) to improve utility of animal modeling and reduce required animal numbers. Odekerken et al., demonstrated that ¹⁸F-FDG micro-PET is a sensitive diagnostic tool for detecting early bone pathology, including early osteomyelitis (21), even in the presence of titanium implants (118). This method of imaging could differentiate between aseptic and infected bone as early as three weeks post-operatively and post-infection. Authors suggest that ¹⁸F-FDG PET carries potential as an early detector of clinical osteomyelitis cases, which is further confirmed by a retrospective analysis of clinical osteomyelitis cases performed by Wenter et al. (140). An important investigation geared toward improving available diagnostics was performed in a rabbit model of chronic osteomyelitis. In this study, the capability of PCR to return positive results was compared with traditional osteomyelitis diagnosis via radiographs and bacterial cultures of bone biopsies taken via different methods. Results indicated that PCR was a sensitive diagnostic tool and described techniques to determine species identification (23). It deserves recognition that while PCR is a strong tool to detect low bacterial burdens or metabolically inactive bacteria that may not yield positive bacterial culture, PCR results will not provide antibiotic susceptibility data. The described PCR techniques are useful for experimental models and also offer utility for clinical cases.

Rabbits are widely utilized to test therapeutic strategies for the clearance of bacterial osteomyelitis. Rabbits are hindgut fermenters, which means that they may process oral antibiotics differently than humans (64). Nonetheless, rabbits have been widely utilized for evaluation of systemic and locally delivered antibiotic therapies (114, 123, 141). Rabbits also are a useful modeling system for evaluation of antibacterial coatings upon implants and local drug delivery systems (114, 129), as demonstrated by the use of silver ion doped calcium phosphate beads (120). There have also been investigations into alternative therapies for osteomyelitis, including the work performed by Kishor et al., investigating the use of bacteriophages to clear chronic osteomyelitis (142). In this study, S. aureus specific phages were purified, characterized, and utilized as a therapeutic in a model of acute and chronic femoral osteomyelitis. High doses of phage cocktail were found to be effective to clear S. aureus infection. This work presents an intriguing consideration for specific therapy of bacterial osteomyelitis. Another interesting study investigated the use of locally applied ozonated oxygen in a rabbit femoral model. While this treatment did not eliminate osteomyelitis, it did seem to lessen the clinical and radiographic markers of disease (122).

Rabbits fill a unique niche in in vivo osteomyelitis research. They are often utilized when the research goal involves assessment of orthopedic hardware or locally applied therapeutics and a small animal is needed, whether that need is dictated by animal housing limitations or by stage of research development. Rabbits provide a more relevant size to evaluate some human orthopedic implants, as well as an appropriate size to be maintained long-term so that revision procedures can be performed. Rabbits also provide a more similar immune system and long bone density to humans than mice and rats provide (143, 144). Despite these benefits, rabbit models are accompanied by more complex challenges including respiratory depression under anesthesia, hindgut fermentation, which impacts the ability to assess oral antibiotic therapies, and variation in bone healing response of young rabbits compared to humans. Most rabbit modeling should be performed in mature rabbits to maximize translation of results to clinical patients.

Pigs are not as widely utilized to model bacterial osteomyelitis, but the models that are available are effective, well-characterized, and have seen logical progression. Studies may utilize either mini-pigs or commercial pigs. Perhaps the most widely utilized model of porcine osteomyelitis is a hematogenous model (56, 145–149). Alternative models include mandibular osteomyelitis (46, 150), tibial implant-related osteomyelitis (151–154), and traumatic tibial osteomyelitis (155). When the hematogenous model of osteomyelitis was initially introduced, an inoculum of S. aureus (S54F9) was administered IV through a lateral ear vein without any additional trauma. This IV inoculation resulted in acute, suppurative pneumonic and osteomyelitic lesions. Lesions of osteomyelitis were found primarily in the long bones, but also in the costochondral junctions of ribs (147). This model has been modified and is most frequently used by administering bacterial inoculums into the femoral artery (145, 146, 148, 149). Femoral artery inoculation is reliable in inducing osteomyelitis localized to the injected limb. This technique may produce concurrent soft tissue infections, injection site abscesses, and the degree of disease during the study may be variable (148, 156). However, this remains a strong technique for modeling acute hematogenous (juvenile) osteomyelitis.

Pigs have not been utilized as widely as mice and rats to investigate pathogenesis of osteomyelitis, but there are a few interesting reports. One such study was carried out in a hematogenous model of osteomyelitis to determine the infection potential and disease characterization of three different strains of S. aureus (56). This work compared the typically utilized strain of porcine S. aureus (SF549) with two human strains of S. aureus (UAMS-1 and NCTC-8325-4). Results indicated that UAMS-1 and NCTC-8325-4 were less successful in establishing osteomyelitis than the porcine specific strain. Authors hypothesize that this may be due to increased host specificity, in contrast to rodent models, and that inoculation dose may play a role, which again brings attention to the importance of model and bacterial strain selection during experimental design. Additionally, an interesting discovery of biofilm within bone lesions shortly after infection was made and raises the concern that biofilms may form quite early on in disease. Jødal et al. investigated blood perfusion using [¹⁵O]water PET, and confirmed their hypothesis that blood perfusion would be increased in osteomyelitis-diseased bone as compared to healthy bone. While blood perfusion was increased in diseased bone as compared to healthy bone, blood perfusion was four-fold greater in areas of soft tissue infection than diseased bone (156).

Afzelius et al. (149, 157) have made multiple investigations involving ideal tracing agents for diagnosing osteomyelitis. They investigated the use of more specific radiotracers, including: ⁶⁸Ga-labeled DOTA-K-A9, DOTA-GSGK-A11, [¹⁸F]NaF, [⁶⁸Ga]Ga Ubiquicidin, and [⁶⁸Ga]Ga-DOTA-Siglec-9, and compared them to the use of [¹⁸F]FDG. This study demonstrated no accumulation of the more specific radiotracers, but positive accumulation of [¹⁸F]FDG (149). Investigators also compared [^99mTc]Interleukin-8 (IL-8) scintigraphy with [¹⁸F]FDG PET/CT in a hematogenous porcine model of osteomyelitis and found that [^99mTc]IL-8 was simple to prepare and use, and that it was capable of detecting 70% of lesions compared with 100% sensitivity of [¹⁸F]FDG PET/CT. This makes [^99mTc]IL-8 scintigraphy a promising candidate for further investigation for use in children, to decrease the radiation exposure, as compared to utilizing [¹⁸F]FDG PET/CT (157). Another interesting study was performed by Lüthje et al., who investigated the regulation of various acute phase proteins during osteomyelitis and found a significant pro-inflammatory local response to osteomyelitis, with limited systemic response. These findings confirm that osteomyelitis remains challenging to diagnose based on systemic findings and adds to the understanding that local investigation is necessary (153).

Most porcine studies thus far have been accomplishing model development, pharmacokinetic work (152) and diagnostic methods. There is even one investigation into bone regeneration techniques in the face of osteomyelitis (46). Hill et al. (155) completed a study utilizing tibial implant-associated osteomyelitis and found that they could prevent osteomyelitis by administering combination antibiotic therapy every 6 h for 7 days. Jensen et al. (154) comment that pigs provide an ideal model for investigation into implant surface coatings, medical and surgical treatment regimes, and vaccination against S. aureus.

Pigs, particularly mini-pigs, offer many benefits, including size that is appropriate for complex or multi-stage procedures and for assessments of orthopedic hardware for human use. Porcine bone possesses similar fracture stress to human bone (158), hematogenous modeling creates a very similar situation to juvenile hematogenous osteomyelitis, and the gastrointestinal system of pigs is appropriate to receive oral antibiotics. There are many challenges when using pig models, including rapid growth and excessive mature body weight when utilizing commercial pigs (159), shorter long bones than found in people (154), the greater expense associated with a large animal model, variation in degree of disease manifestation, as well as a generally fractious demeanor. Porcine models are not currently as widely utilized as small animal models of osteomyelitis but provide an ideal model for the study of hematogenous osteomyelitis, offer great capabilities into investigation of imaging techniques, and are an area of interest for further development in the modeling of osteomyelitis. In general, commercial pigs are suitable for proof of concept and model development work, as they are less expensive than mini-pigs, but for longer-term studies and more appropriate translational work, mini-pigs should be utilized.

Kaarsemaker et al. (160) initiated development of ovine models of osteomyelitis via creation of a tibial defect and subsequent bacterial inoculum injection into the medullary cavity of adult sheep. This study provided valuable information, including the ability to establish osteomyelitis in sheep and also the requirement for peri-operative systemic antibiotics to lessen the risk of fatal sepsis. Since then, a variety of long bone models have been developed, focused on the tibia (161, 162) or the femur (163), and often involving hardware infected with biofilm or planktonic bacteria (164) with or without revision procedures (162, 165). There remain a variety of techniques of creating bone injury, from unicortical defects and medullary canal inoculation (165) to osteotomies stabilized with experimental hardware (166). Recently, Moriarty et al. (162) established a model to replicate a failed two-stage revision procedure utilizing a MRSA infected intramedullary nail. This will likely be a valuable model to evaluate therapeutic strategies moving forward.

Most investigations into therapeutics in ovine models have been centered upon experimental implants, systemic or local antibiotic therapies, and the ability to replicate the multi-stage revision procedures utilized in human medicine. There have been multiple investigations into local drug delivery devices to clear osteomyelitis. Boot et al. performed a multi-stage revision procedure and compared an injectable hydrogel impregnated with gentamicin and vancomycin to an antibiotic-loaded bone cement impregnated with gentamicin and vancomycin. Investigators were able to clear significantly more cases of osteomyelitis in the experimental hydrogel group, compared to the bone cement group, thereby presenting this material as a promising candidate for further exploration (165). Stewart et al. investigated another concept in local drug delivery by creating a vancomycin-modified titanium plate that demonstrated decreased clinical signs of infection, prevented biofilm formation and promoted bone healing in an infected tibial osteotomy model (166).

Currently, sheep are most often utilized for investigations into therapeutics utilizing long bone models. As such, sections regarding pathogenesis and diagnostic innovations were not included. Regardless, sheep are a valuable animal resource for the modeling of bacterial osteomyelitis, particularly focused on long bones. Sheep provide an ideal long bone size to perform complex procedures, replicate the treatment strategies utilized in clinical cases such as multiple revision procedures, and assess orthopedic hardware and devices for human use. Many characteristics of ovine bone are similar to that of humans, including torsional stiffness and osteogenesis (64), which adds to the strength of ovine modeling. Challenges associated with ovine modeling include the risk of sepsis, which may require peri-operative antibiotics, as well as the cost of housing and maintaining a large animal.

Most caprine models of osteomyelitis utilize the tibia, although models have variable approaches. Salgado et al. described a unicortical tibial defect with concurrent application of a sclerosing agent. Staphylococcus aureus was inoculated into the medullary canal and the defect was sealed with bone wax. In this model, goats received a perioperative dose of IV antibiotics. Induction of osteomyelitis was successful and no goats suffered from fatal sepsis (167). In an adaptation of this model, the sclerosing agent and perioperative antibiotics were omitted, and osteomyelitis was successfully induced, again with no reported sepsis (168). Other tibial models include the internal fixation of a tibial osteotomy (169) and percutaneous pin placement throughout the tibia (170). Through these investigations, researchers have also proposed histology scoring systems, to aid in the evaluation of model development (169).

Similar to sheep, goats serve as viable translational models for investigations into therapeutic strategies. Wenke et al. utilized a similar model to that of Salgado et al. to investigate the efficacy of tobramycin-loaded calcium sulfate pellets compared to the efficacy of tobramycin-loaded antibiotic beads to treat bacterial osteomyelitis. Calcium sulfate and bone cement formulations loaded with tobramycin performed well, raising interest into the use of calcium sulfates for local drug delivery, as they do not require an additional procedure for removal (168). Tran et al. (169) investigated a silver-based antibacterial coating on intramedullary nails. In an experiment utilizing two goats, the goat that received the experimental implant displayed less severe signs of osteomyelitis than the control goat. An interesting experiment was performed to investigate the utility of a directly applied electric current to eliminate osteomyelitis over the course of 3 weeks. Authors found that electric currents were able to prevent signs of infection and suggest that this would be effective in clinical situations (170). Salgado et al. also reported an investigation of muscle vs. non-muscle flaps for reconstruction of defects and effective clearance of osteomyelitis. This study was designed as a result of discrepancies in the literature, with some reports of muscle flaps being superior and vice versa. This study found no difference between muscle and non-muscle flaps and re-emphasized that the most critical factor in treatment of bacterial osteomyelitis is thorough debridement (38).

Similar to sheep, goats possess great utility in modeling bacterial osteomyelitis, and this utility lies primarily within the size and composition of the caprine long bones, specifically the tibia. Long bone size and composition makes goats ideal for complex procedures and multi-stage surgeries. Goats provide an excellent model for assessment of orthopedic hardware intended for human use, as well as examination of local drug delivery devices and experimental coatings. Goats have not suffered from the reported sepsis that affected sheep when receiving intramedullary bacterial inoculation, which may aid researchers when selecting either sheep or goats as a model. Similar to any large animal model, goats are accompanied by greater costs than small animal models. As the majority of caprine modeling has been performed to either establish a reliable model or assess treatment options, the sections for pathogenesis and diagnostic investigations were omitted.

Canine models have been used in the past to model osteomyelitis, although today they are not widely utilized. Similar to caprine and ovine models, canine models have primarily utilized long bones, specifically the tibia (53, 171, 172) and the femur (47, 173, 174), although a vertebral model has also been described (175). Models vary in approach. Deysine et al. described an injection of bacterial inoculum into the tibial nutrient artery without any additional trauma. This approach was effective in establishing osteomyelitis, but also resulted in the loss of three dogs from septicemia (53). Most other models report bone trauma and bacterial inoculation of the medullary canal, whether that is by direct inoculation or placement of an infected implant (47, 171, 174). Khodaparast et al. had success in establishing osteomyelitis via application of a penetrating captive bolt device to the tibia of dogs to create an open fracture. This approach was selected in order to mimic traumatic osteomyelitis. This model involved the placement of microdialysis probes for sample collection, which is a valuable tool (172) when investigating the dynamics of local environments, whether that is physiologic dynamics or drug delivery profiles.

As described above, Khodaparast et al. (172) established a tibial fracture model of canine osteomyelitis and placed microdialysis probes with the goal of exploring the role of vascular endothelial growth factor (VEGF) as a rate-limiting step in wound healing. This was investigated by measuring VEGF mRNA levels in response to S. aureus osteomyelitis and S. aureus osteomyelitis treated with a rotational gastrocnemius muscle flap. The muscle flap was investigated because wound healing is accelerated in the presence of well-vascularized tissue. VEGF mRNA levels were found to be greater in the animals with osteomyelitis that received the rotational muscle flap as compared to those who did not. This finding suggests that type of surgical closure impacts specific biological signals and cellular pathways, and may add strength to the recommendation for utilizing muscle flaps for improved wound healing in reconstructive surgeries. Another investigation into pathogenesis was performed by Chen et al. (175) who aimed to investigate the presence, type, and origin of bacteria adjacent to metal implants utilized in the surgical management of pyogenic vertebral osteomyelitis. Investigators found that bacteria were retrieved not only from metal implants, but also from surrounding bone, despite the lack of radiographic signs of infection. These findings suggest that metallic implants are not necessarily the source of persistent or recurrent bacterial infection in vertebral osteomyelitis.

Despite there being few reports, there are canine models of osteomyelitis that investigate treatment strategies. Two models focused on the prevention of osteomyelitis, and found that the placement of gentamicin impregnated bone cement could prevent the development of osteomyelitis in the experimental models (171, 173). Similarly, Huneault et al. (174) investigated the ability of cross-linked high amylose starch (CLHAS) implants loaded with ciprofloxacin to prevent and cure chronic femoral osteomyelitis. This study demonstrated strong preventative efficacy of the ciprofloxacin loaded implants, and also showed that ciprofloxacin loaded implants and oral ciprofloxacin had similar efficacy in clearing bacterial osteomyelitis.

Dogs provide strong models for long bone osteomyelitis. Benefits include appropriate size to perform complex and multi-stage procedures, bone composition and density that is most similar to humans out of the available species (158), temperament that is amenable to handling, as well as well-characterized anesthetic and imaging protocols. Despite these strengths, canine models are no longer frequently utilized for osteomyelitis research. Osteomyelitis research is terminal, and ethical concerns are raised when considering these companion animals as research models. Therefore, despite the provided benefits, it is unlikely that dogs will have a resurgence in popularity for osteomyelitis modeling.