A Journey into Animal Models of Human Osteomyelitis: A Review

Osteomyelitis is an infection of the bone characterized by progressive inflammatory destruction and apposition of new bone that can spread via the hematogenous route (hematogenous osteomyelitis (HO)), contiguous spread (contiguous osteomyelitis (CO)), and direct inoculation (osteomyelitis associated with peripheral vascular insufficiency (PVI)). Given the significant financial burden posed by osteomyelitis patient management, the development of new preventive and treatment methods is warranted. To achieve this objective, implementing animal models (AMs) of infection such as rats, mice, rabbits, avians, dogs, sheep, goats, and pigs might be of the essence. This review provides a literature analysis of the AMs developed and used to study osteomyelitis. Historical relevance and clinical applicability were taken into account to choose the best AMs, and some study methods are briefly described. Furthermore, the most significant strengths and limitations of each species as AM are discussed, as no single model incorporates all features of osteomyelitis. HO’s clinical manifestation results in extreme variability between patients due to multiple variables (e.g., age, sex, route of infection, anatomical location, and concomitant diseases) that could alter clinical studies. However, these variables can be controlled and tested through different animal models.


Background
The use of animal models in medicine is an ancient story, with the first written records dated to 2000 BC as stated in the book "Laboratory Animal Medicine" (2nd edition, chapter 30, page 1206) "The earliest written records of animal experimentation date to 2000 BC when Babylonians and Assyrians documented surgery and medications for humans and animals" [1]. From this written milestone, animals as models to mimic human diseases were implemented mainly to resemble, more precisely, human pathology. The use of models is nowadays strictly regulated to avoid unnecessary suffering of the animals (ethical concerns), ensuring respect for animal welfare and the development of alternative methods in compliance with current European legislation (DIRECTIVE 2010/63/EU on the protection of animals used for scientific purposes) [2]. The need for animal models in orthopedics allowed the development of new therapeutic approaches and the refinement of traditional surgical methods to a new concept of personalized medicine [3][4][5]. This review describes the history of using animal models in orthopedics to mimic human osteomyelitis with a critical description of the disease's microbiological basis and immunological characteristics. PubMed was used as a reference point to search for articles over more than 120 years (1900 to December 2021) on large and small animal models.
nutrients. As the biofilm matrix envelops bacteria and protects them from external stimuli, there is a progressive decrease in nutrient and oxygen availability [18]. To overcome this problem, bacteria act at two levels, rearrangement of cells in a metabolic-dependent way (e.g., anaerobes in the inner core and aerobes at the distal surface) and transcriptomic shift (e.g., quiescent cells at the interface with the surface and active bacteria in the outer part of architecture) [28,36]. The last phase of biofilm formation is characterized by the rapid enzymatic remodeling of the biofilm's external layer, leading to the dispersion of bacterial cells, which rapidly convert from sessile into motile form. The production of specific enzymes is dependent on the specific EPS produced (e.g., Escherichia coli produces N-acetyl-heparosan lyase, Pseudomonas aeruginosa, and Pseudomonas fluorescents produce alginate lyase, and Streptococcus equi produces hyaluronidase) [7].

Device-Related and Non-Device-Related Biofilm Infections
The ability of bacteria to colonize almost all surfaces gives them the possibility to cause severe infections resulting in therapy difficulties and economic loss. Device-related infections (e.g., osteomyelitis) primarily originate from bacteria on patients' skin during implant insertion and then reach the device surface from the incision site or by hematogenous seeding, starting the biofilm formation [37]. Contact lenses, central venous catheters, mechanical heart valves, peritoneal dialysis catheters, prosthetic joints, pacemakers, urine catheters, and vocal prostheses are suitable bacteria targets. [7,20,38]. Non-device-related bacterial biofilm infections include dental plaque, cystic fibrosis, osteomyelitis, periodontitis, and chronic inflammatory diseases [8]. The mouth can be colonized by up to 500 species of bacteria [39]. The following bacterial species were seen in osteomyelitis of the jawbones Streptococcus spp., Eikenella spp., Staphylococcus spp., Actinomyces spp., Klebsiella spp., Lactobacillus spp., Haemophilus spp. [40], and anaerobic bacteria, such as Bacteroides spp., Peptostreptococcus spp., and Fusobacterium spp. [39]. The colonization of dental surface follows specific steps; Gram-positive cocci are the primary colonizers (Streptococcus mutans, Streptococcus mitis, Streptococcus sanguinis, Streptococcus oralis), followed by Gram-positive rods, Actinobacteria (A. israelii and A. viscosus), and several Gram-negative cocci [41]. Grampositive facultative anaerobic bacteria are the most abundant biofilm producers in healthy adults, while, during chronic gingivitis, the number of Gram-negative anaerobic bacteria increases about ten times that of commensal species found in healthy individuals [8]. L-form bacterial pathogens and biofilms can cause chronic inflammatory disorders [42]. The mode of action to induce chronic disorders resides in the production of molecules that attach and inactivate the vitamin D receptors; these proteins modulate the activity of the innate immune cells [8].

Microbiology of Bacterial Osteomyelitis
From a microbiological point of view, most HO is generally monomicrobial [8,43,44], but few clinical cases are caused by more than one microorganism [45]. The most isolated pathogen (80% of culture-positive cases) from clinically relevant HO is S. aureus, followed by group A Streptococci (GAS) [8,43,44]. However, the bacterial etiology of osteomyelitis varies with the individual age [44]. The incidence is approximately 8 cases per 100,000 children (<5 years) per year [46]. In infants, S. aureus, S. agalactiae, and E. coli represent the most frequently isolated microorganisms, while in children, S. aureus, S. pyogenes, and Haemophilus influenzae [43]. In a retrospective study on 167 children in Malawi, Beckles (2010) reported the prevalence of causative agents found in microbiological examinations as follows "61% episodes were caused by S. aureus, 4% by E. coli, 2% by Streptococcus, 2% by Pseudomonas, 1% by Bacillus subtilis, 1% by Proteus and in 29% no microorganism could be detected" [47,48]. In addition to S. aureus, Salmonella spp. has to be noted as a potential causative agent of the disease; however, with a low prevalence (0.45% of osteomyelitis infections) [49,50]. Beyond the "common" bacteria, the identification of Kingella kingae as a new potential aetiological agent is rapidly increasing, especially in young children (excluding the neonatal period), with a prevalence of up to 82% [51,52]. Historically, one of

Immune Response
The innate immune response is one of the first weapons against pathogens, and it acts through the identification of a broad range of microbial components (e.g., nucleic acids, cell wall constituents) that can be recognized by specific receptors on the cells surface (e.g., polymorphonuclear neutrophils (PMNs), dendritic cells (DCs), macrophages) [59,60].
The first line of defense at the site of infection are PMNs that, moving from the bloodstream, can detect and fight both planktonic bacteria and bacterial biofilms and recruit monocytes/macrophages and modulate their activity [61][62][63]. PMNs chemotaxis is activated by different chemo-attractant molecules (cytokines and chemokines), and their primary function is to phagocytize and generate reactive oxygen species (ROS) against opsonized bacteria. However, when talking about biofilm, PMNs efficiency is highly dependent on its maturation state [64,65]. For example, Stroh et al. demonstrated that biofilm opsonization with neither immunoglobulin G nor complement did improve the degree of cellular adherence of PMNs to the biofilm surface, facilitated degranulation, or inducted phagocytosis in vitro. On the other hand, the generation of ROS was critically dependent on the biofilm opsonization with immunoglobulin G [66].
Monocytes and macrophages are also known to phagocyte planktonic bacteria. However, their impact on bacterial biofilms is still unclear. There is evidence that S. aureus biofilms can change macrophage phagocytosis efficiency and bactericidal action in different ways, helping, for example, to shield S. aureus cells from Toll-Like Receptors (TLRs) recognition and consequent macrophages activation. In addition, thanks to their great size, bacteria are difficult to gulp when clustered in biofilms, even for professional phagocytic cells such as macrophages [67].
Specific immune responses to biofilm bacteria have yet to be discovered when looking for the key distinctions in immune responses to planktonic or biofilm bacteria; the only obvious difference is the difficulty of removing the bacteria when biofilm is present, causing the infection to become chronic [68]. However, as far as bacterial biofilms are concerned, the presence of lactoferrin has been revealed to be of particular importance, as it can suppress the formation of biofilm for some staphylococcal species [69,70].
Whenever there is a bacterial infection with the production of a biofilm, it also leads to the activation of T-cells and monocytes, which is traduced in a local increase of proinflammatory cytokines (e.g., Tumour Necrosis Factor-α (TNF-α), interleukin (IL)-1, interleukin (IL)-6) [64,66,[71][72][73]. Unfortunately, sometimes, and for reasons still not completely understood, this continuous release of inflammatory mediators does not help control the infection and spreads into osteolytic and tissue-damaging processes [71].
The presence of bacteria activates immune cells and directly impacts bone tissue, in which cellular components increase the expression of Toll-like receptor 2 (TLR2). For example, when in contact with S. aureus, not only the TLR2 expression but also the activity level of Jun N-terminal kinases (JNK) seem to directly impact osteoblast apoptosis and osteogenic differentiation after the bacterial invasion [74,75]. Furthermore, it has recently been shown that bacterial endotoxins, especially lipopolysaccharide (LPS), can also promote cell apoptosis and inhibit the differentiation of osteoblasts by JNK pathway activation [76]. These findings were confirmed in an in vitro study, suggesting that methicillin-resistant S. aureus biofilms liberate soluble molecules capable of decreasing osteoblasts viability and osteogenic potential, indirectly promoting osteoclast activity [77].
In terms of cytokines, those from the TNF superfamily promote apoptotic cell signaling in a variety of cell types (including osteoblasts) via a direct interaction between the soluble TNF-related apoptosis-inducing ligand (TRAIL) and specific cell receptors, such as osteoprotegerin (OPG), which acts as a decoy soluble receptor for TRAIL and the receptor activator of nuclear factor-kappa-β ligand (RANKL), promoting osteoclastogenesis [78][79][80].
As part of the immune system, osteoblasts may create antimicrobial peptides in response to bacterial invasion, with human defensins being one of these antimicrobial peptides that can be extremely effective in combatting a wide range of pathogens. [81,82]. The expression of β-defensins in bone tissues has already been demonstrated in both healthy and infected bone, concluding that antimicrobial peptides might also play a role in osteomyelitis [83,84]. Furthermore, a decrease in antimicrobial peptide expression was found after administering immunosuppressive drugs, which was expected to modulate the susceptibility to osteomyelitis [85].
On the other hand, MyD88 is an essential component of the innate immune system by communicating with IL-1R and TLRs. In part, thanks to the capability of IL-1 to intermediate the neutrophil enrollment and to endorse the formation of a fibrous capsule (abscess) capable of containing the bacteria (e.g., S. aureus) [91,92]. Moreover, especially in the bone marrow, IL-1 is also responsible for promoting granulopoiesis [93,94] and has an already known role in anti-staphylococcal immunity and compelling evidence that IL-1 signaling is capable of affecting bone cells in vitro. Putman et al., 2019, demonstrated that MyD88 and IL-1R signaling are necessary for efficient antibacterial immune responses during osteomyelitis but, on the other hand, may also promote osteoclastogenesis and host-mediated bone loss during osteomyelitis [95].

Elements of Bone Anatomy
Bone is a composite material and specialized connective tissue which consists of organic and inorganic components [96]. From a functional point of view, bone is the primary reservoir of calcium and phosphate in the body. On top of that, bone contributes to hematopoiesis and plays a protective and mechanical role. It has cellular components whose active constituents are devoted to the production and rearrangement of the osteoid matrix. There are seven cell types in the bone: (1) progenitor bone cells, (2) osteoblasts, (3) osteocytes, (4) osteoclasts, (5) hematopoietic cells, (6) vasculature, and (7) endosteal cells. Concerning bone formation steps, vascular buds first invade the mesenchymal model. Osteoprogenitor cells then migrate through vascular buds and differentiate into osteoblasts giving rise to primary ossification centers and, as a result, cartilage model forms. Subsequently, growth happens appositionally and interstitially. Of note, bone marrow forms by resorption of a central portion of the cartilage, and secondary ossification centers develop at bone ends, thus leading to growth plates. Osteocytes derive from osteoblasts that remain imprisoned in the calcified matrix after synthesis. [97,98]. The matrix has an organic component consisting of collagen fibers, chondroitin sulfate, amorphous substance (e.g., proteoglycans and glycoproteins), and an inorganic component consisting mainly of calcium phosphate and carbonate.
The mature bone tissue has a lamellar structure. The lamellae are aggregated into parallel layers, and each lamella is formed by cells and the extracellular matrix. The osteon is the morphofunctional unit of the bone. It is made of concentric lamellae around the Haversian canal containing arteries, veins, and nerves. Instead, the canaliculi and the canals of Volkmann (arranged transversely or obliquely) allow communication between osteocytes and osteons, respectively.

Histopathological Generalities on Osteomyelitis
When bone tissue becomes infected, bacteria induce an acute inflammatory reaction, with infiltration of neutrophils and systemic symptoms. Bacteria and inflammation affect the periosteum, the blood supply decreases, and the bacteria spread inside the bone, causing necrosis [56,99].
In children, the periosteum is not entirely attached to the cortex, which causes a subperiosteal abscess to form on the bone surface. In addition, the lifting of the periosteum further compromises blood flow to the bone, causing segmental bone necrosis, known as "sequestrum" [100].
In chronic osteomyelitis, numerous inflammatory cells and the released cytokines stimulate bone resorption by osteoclasts, fibrous tissue growth, and deposition of new bone tissue in the periphery. When new bone tissue deposits, a sleeve (called "involucrum") around the devitalized bone is formed, and the rupture of the subperiosteal abscess can lead to a soft tissue abscess and possibly form a "draining sinus" [100,101]. Clinically, the gold standard method for diagnosing osteomyelitis remains the bone biopsy coupled with histopathologic examination and tissue culture [58].

Periprosthetic Joint Infection (PJI)
Most PJIs are caused by intra-operative contamination, which can result in early or delayed infection, whereas hematogenous seeding is less common, with late infections occurring years later. Both early postoperative and hematogenous infections have an acute onset, despite their different pathogenesis. On the other hand, chronic late infections can be caused by less virulent microorganisms, and while they are thought to be caused by intraoperative contamination, symptoms appear slowly. Although there is no consistent definition in the literature, most authors agree on classification: early = 3-6 weeks, chronic = 6 weeks, and beyond. This 3-6 week cut-off aims to discriminate against patients requiring DAIR vs. 2-stage revision arthroplasty [101,102]. Early infections can result from perioperative bacterial inoculation, while delayed infections may result from less virulent bacteria inoculation or hematogenous sources. Late infections are commonly caused by an infection carried by the blood to the prosthesis site [101][102][103]. New major criteria for diagnosing PJI were drawn out by Parvizi (2018) as two positive cultures or the presence of a sinus tract. Giving score values for specific clinical parameters (e.g., elevated serum CRP (>1 mg/dL), D-dimer (>860 ng/mL), erythrocyte sedimentation rate (>30 mm/h), elevated synovial fluid white blood cell count (>3000 cells/µL), α-defensin (signal-to-cutoff ratio > 1), polymorphonuclear percentage (>80%), and synovial CRP (>6.9 mg/L)), a final aggregate score ≥6 indicates an infected patient, while between 2 and 5 indicates the inclusion of intraoperative findings for confirming or refuting the diagnosis [104].
The specific infection of the knee prosthesis begins with the adhesion of bacteria and biofilms' formation [105,106]. The implant is quickly covered with adhesins (e.g., fibronectin, fibrin, fibrinogen) in the extracellular fluid. Fibronectin is important in the adhesion of S. aureus as it can bind proteins that promote bacterial adhesion [107]. Besides the well-known role of fibronectin-binding proteins (FnBPA and FnBPB), staphylococci harbor a specific locus called "intercellular adhesin locus (ica)" that expresses the polysaccharide intercellular adhesion (PIA), one of the fundamental constituents of biofilm [108]. After the initial adhesion, a thin layer of slime produced by the host induces an inflammatory reaction in the host itself [109]. Biofilm formation is the defense mechanism by which these organisms can escape the immune system [110]. As a result, S. aureus can invade and colonize immune cells, such as macrophages and neutrophils [110]. Furthermore, bacteria can naturally mutate some essential metabolic genes, giving rise to different subpopulations called small colony variants (SCVs). These mutants are not particularly virulent but can persist viable inside host cells by attenuating the virulence and antibiotic resistance profile [111,112].

Post-Traumatic Osteomyelitis
"Post-traumatic osteomyelitis" means bone infection following an open fracture or following the treatment of a closed fracture with intramedullary nailing or plating to stabilize the fracture [101]. Transcutaneous bacterial contaminations in open fractures are frequent. When bone tissue is involved, bacteria induce an acute inflammatory reaction, as we have seen previously. Bacteria and inflammation spread within the bone and percolate through the Havers systems and periosteum, compromising callus formation [100,113,114].

The History of Animals as Models of Osteomyelitis
The scientific literature offers numerous reports about AMs to investigate the pathogenesis, the diagnosis, and the treatment of osteomyelitis. In this context, AM's use is fundamental for developing preventive and effective therapies [115]. In some cases, HO's clinical manifestation results in extreme variability between patients due to multiple variables (e.g., age, sex, route of infection, anatomical location, and concomitant diseases) that could alter clinical studies. On the other hand, animal models may be used to control and test these factors [115].
Between the second half of 1800 to 1970, the availability of scientific reports targeting animals as models for studying osteomyelitis was minimal. However, at the end of the twentieth century, a progressive increase in the number of reports focused on this topic was noticed, corroborating the hypothesis that the development of antibiotics positively influenced osteomyelitis studies involving AM [116].
The two most pioneering scientific reports about AM (rabbit) of osteomyelitis were done by Rodet (1885) and Lexer (1894), which administered S. aureus intravenously to induce the formation of abscesses [117,118]. After several years, three independent research papers: Starr (1922), Haldeman (1934), and Thompson and Dubos (1938), observed that after intravenous and/or intra-bones injection of S. aureus in rabbits, typical osteomyelitis lesions microscopically resemble those found in humans [119][120][121]. In all these studies, a common factor was found; the direct injection of living or attenuated bacteria caused the death of animals within a few days after treatment by disseminating abscesses in different organs and with occasional bone lesions [115]. In 1941, Scheman proposed a new rabbit model of osteomyelitis based on S. aureus and morrhuate sodium injection directly into the tibial metaphysis. In this study, experimental subjects did not die of sepsis, and chronic osteomyelitis lesions arose within the injection site over several weeks [122].
From these fundamental studies, historically considered the backbone of AM research on osteomyelitis, researchers have developed new models and protocols to formulate preventive approaches to understanding specific molecular features of this clinical condition [115,116]. From the beginning of the twentieth century, the rabbit was not considered the unique model anymore to characterize osteomyelitis, and several other species were further investigated: Hamblen (1968) and Zak (1982) introduced rat models [123,124], Ueno (1974) used mouse models [125], Deysine (1976) and Fitzgerald (1983) focused on dog models [126,127], Passl (1979) worked on acute osteomyelitis in guinea pigs [128],  studied the first avian (chickens) model of acute HO [129,130], Patterson (1993) developed a mini swine model of chronic mandibular osteomyelitis [131], and Curtis (1995) and Kaarsemaker (1997) worked on an open fracture goat model and a new model of chronic osteomyelitis in sheep, respectively [132,133].

What Should Be Considered When Designing Animal Models of Osteomyelitis?
The complex dynamics that surround osteomyelitis should become simpler when using animal models. However, choosing the best model is one of the most critical decisions. Multiple variables simultaneously act during an experiment involving AM, and each of them should be responsible for misleading and biased during procedures. The most significant variables related to experimental designs are animal species, age of the experimental animal, route of inoculation, bacterial species, infection promoters, and evaluation [115,116]. Table 1 reports the advantages and disadvantages of animal models used to mimic human osteomyelitis. The main advantages and disadvantages of animal models of human osteomyelitis need to be carefully checked when planning an in vivo experiment. Each model can be used to mimic and elicit specific characteristics of the human disease. For example, large animals (dog, porcine, sheep, and goat) present some anatomical features that can resemble humans, allowing to test new surgical approaches and new therapeutical strategies. On the other hand, the presence of very characterized mouse strains and the development of specific antibodies can be very useful in studying the pathophysiology of the disease using targeted immunohistochemistry or in vivo imaging techniques.

Which Is the Most Appropriate Animal Model for the Experimental Procedure?
It is utopic to talk about the perfect animal species that should mimic osteomyelitis; however, it is possible to define an ideal animal species to understand this disease better. This hypothetical model should have molecular, cellular, structural, and mechanical features found adequately in human bone, together with easy housing and handling, low cost, high tolerance to pharmacological treatment (e.g., antibiotic administration), and a size that allows surgeries that resemble clinical practice [115]. It is clear that each animal model has specific features that make it better for a particular purpose; it is also evident that an in-depth analysis should help establish the best model for each experimental procedure. For example, comparing femoral cortical and lumbar trabecular bone tissues between animals (e.g., dogs, pigs, chickens, and rats) and humans was shown that canine and porcine models have mineral proportions similar to those of humans [140]. Furthermore, pigs and dogs show advantages in refining surgical procedures (e.g., excision of abscesses and implant of prostheses/medical devices) due to the long bone size [141].
On the contrary, the small size of mice, rats, rabbits, and chickens makes them inadequate for long bone surgery procedures [115]. Beyond this, mice, rats, and rabbits' bones are unique for quantitative microbiology tests (e.g., cell count). In this case, the different gastrointestinal physiologies of these animals can affect the results. For example, rabbits have a pseudoruminant gastrointestinal system, which is unsuitable for testing antibiotics absorption and deposition in long bones [142]. In contrast, mice and rats tolerated broad-spectrum antibiotics [143].

Age of the Animals and Route of Inoculation
The second most crucial parameter about the eligibility of the best animal model for HO studies in long bones is age since this disease arises during childhood [46]. Unfortunately, the importance of this parameter is often underestimated, and the majority of reports do not mention in detail the age of the used models, often referred to as "adult" or "young" animals [115]. In 1943, Weaver and Tayler studied the difference in HO development between young (6-8 weeks old) and adult rabbits [144]. They found that young models were best suited for HO development while, in adult animals, the procedure underwent failure. Another problem relates to sex and sexual development; most studies involving mice used females of 8-to 10-weeks-old [115]. At this stage, mice are sexually active, and this physiological condition was thought to influence the inability to establish large macroscopic lesions in most of these models [145][146][147]. Pigs and avian HO models were developed in growing animals, demonstrating typical lesions commonly found in infants [116].
Three inoculation routes have been commonly used for experimental HO: inoculation into the bone cavity, local inoculation through nutrient arteries (intra-arterial inoculation), and systemic inoculation through intravenous injection [115,116].
To mimic the HO infection development and microenvironmental alterations in bone tissue is essential to correctly deliver the pathogens in a defined anatomical compartment without creating sepsis or metastatic infection. Intravenous injection results are the least likely to overcome this challenge [148]. Since HO often occurs in a single bone without disseminating systemically, the intra-arterial route is considered the best inoculation to generate a predictable and anatomically confined infection. Moreover, from a surgical perspective, this type of inoculation made in the femoral artery is more manageable than other arteries because of its large diameter and easy anatomical access [149,150] (Figure 1). generate a predictable and anatomically confined infection. Moreover, from a surgical perspective, this type of inoculation made in the femoral artery is more manageable than other arteries because of its large diameter and easy anatomical access [149,150] (Figure 1). The implant-related method (top) induces osteomyelitis using an infected implant to study and develop coating or implants with antibacterial properties. This method is convenient during studies focusing on anti-biofilm molecules. Creating a fracture in the bone followed by its infection with bacteria is the post-traumatic method (middle). In this case, the method helps understand how bacteria can spread along with the bone tissue and through the bloodstream. Moreover, external fixation wires impregnated or covered with antibacterial molecules can be done to characterize the shelf-life of new implants better. Finally, the intravenous injection of bacteria can achieve hematogenous osteomyelitis (lower), spontaneously reaching bones. The implant-related method (top) induces osteomyelitis using an infected implant to study and develop coating or implants with antibacterial properties. This method is convenient during studies focusing on anti-biofilm molecules. Creating a fracture in the bone followed by its infection with bacteria is the post-traumatic method (middle). In this case, the method helps understand how bacteria can spread along with the bone tissue and through the bloodstream. Moreover, external fixation wires impregnated or covered with antibacterial molecules can be done to characterize the shelf-life of new implants better. Finally, the intravenous injection of bacteria can achieve hematogenous osteomyelitis (lower), spontaneously reaching bones.

Bacterial Species Used to Induce Experimental Osteomyelitis
The most common bacterial species used to induce HO in animal models is S. aureus. Many scientists used strains from a private bacterial collection (mainly derived from patients with HO), while others used ATCC strains (e.g., S. aureus ATCC 49230, originally isolated from a patient with chronic osteomyelitis). However, HO can also be induced by other bacterial species, including E. coli, P. aeruginosa, and S. epidermidis. From a microbiological point of view, HO is traditionally monomicrobial, and the infusion of a single bacterial species can easily mimic the microbiological environment found in humans. On the other hand, CO is usually polymicrobial [151]. To mimic this complex bacterial environment, the type of strain used and the number of bacteria to be inoculated remain the two pivotal factors to be considered [116]. Hidaka (1985) used a combination of S. aureus (1 × 10 4 CFU/mL) and P. aeruginosa (1 × 10 5 CFU/mL) to experimentally induce osteomyelitis in mice, using silk thread inserted into the metaphysis of the tibia [152]. In 1988, Sakaeda elaborated a combined (aerobes-anaerobes) reliable model of polymicrobial osteomyelitis using clinical isolates of S. epidermidis and E. faecalis as aerobes (1 × 10 5 CFU/mL) and B. fragilis and B. bivius as anaerobes (1 × 10 6 CFU/mL). The microorganisms were loaded in silk threads and inserted into the bone marrow of a rat. [153]. The literature reports a wide range of bacterial inoculum to experimentally induce osteomyelitis, ranging from 10 2 to 10 10 CFU/mL [154]. However, some authors demonstrated that a less than 10 4 CFU/mL concentration resulted in an unsuccessful infection [114,135]. Helbig (2015) developed a rat model (female, six-month-old Sprague-Dawley) of delayed osseous union achieved by intramedullary inoculation of S. aureus (10 3 CFU/mL) [155]. Some authors used higher bacteria loads. Schaer (2012) developed a sheep model to study the antibiofilm properties of a hydrophobic polycationic coating implanted in the tibia using S. aureus ATCC 25923 suspensions (10 6 , 10 8 , and 10 10 CFU/mL) [156]. Bilgili (2015) used a rat model (male, four-month-old Sprague-Dawley) of osseous union inoculating S. aureus (10 8 CFU/mL) [157]. In conclusion, there is no consensus in the literature on a standardized bacterial inoculum to achieve a single and predictable bone infection [154].
In most studies, bacteria were not administered alone but combined with sclerosing agents (e.g., morrhuate sodium solution and arachidonic acid) to facilitate bone infection. Sclerosing agents act on small blood vessels in the bone's medullary canal, causing sclerosis of vascular tissue and subsequent necrosis. This action causes a decrease in the local host defense system and positively influences the proliferation of bacteria, leading to osteomyelitis [116]. Standard protocols inject 0.1 mL 5% sodium morrhuate solution for tibia or femur models in rabbits [122,158] and 25 µL in rats [159]. Other ways can be utilized to enhance the development of bone infection. Foreign bodies (e.g., bone cement, silicone catheter, metal wires, intramedullary nail, or rods) are preferred to sclerosing agents not only to physically insert a non-self body directly in medullar or bone marrow cavities but also to study the impact of bacterial biofilm on bone tissues and mimic device-related biofilm infections [160].
Macroscopic and microscopic evaluation of experimentally induced osteomyelitis uses fundamental methods, including clinical observation, microbiology, radiography, and histology [116]. After the bacterial inoculation or device-implant, the continuous record of physiological parameters or clinical symptoms (e.g., temperature, lethargy, and weight loss) can easily indicate the progression of localized infection, soft tissue degeneration, or generalized dissemination of pathogen (sepsis). In addition, typical clinical signs (e.g., erythema, abscess, or sinus formation) should help in the final diagnosis that has to be supported by other microscopic methods. Culture-dependent techniques are the most diffused and used, but recent advances in genomics and the rapid progression in cultureindependent protocols open a new era in microbiology. Nowadays, the 16S metagenomic profile can easily detect polymicrobial osteomyelitis without the prolonged time and high costs due to classical culture-based microbiology [136,161,162]. Pineda (2009) suggested some radiographic criteria for evaluating the severity of the infection and the use of other diagnostic techniques to help the final diagnosis of osteomyelitis [163].

Specific Animal Models
In a recent review of animal models for the studies of implant-related infection, Lovati (2016) stated that among all the available and validated animal models, "rats represent the most used (53%), followed by mice (25%), rabbits (20%) and large animals (2%)" [114].

Rat Models
Historically, the introduction of the rat as a model to mimic osteomyelitis (in particular internal fixation models of osteomyelitis) has to be acknowledged by the pioneering studies by Scheman (1941), Hamblen (1968), and Zak (1982), which injected in tibial metaphysis a 5% morrhuate sodium solution followed by a suspension of S. aureus [122][123][124]. These types of rat tibial models were recognized as the first to assess chronic osteomyelitis and, at the same time, helped scientists to study the administration and delivery of multiple bacteria [164,165]. Further studies demonstrated that foreign bodies (e.g., fibrin glue, bone cement, fibrin foam) could improve the infection rate up to 85% [116]. In 1985, Rissing drilled a cortical canal in the medullary portion of the tibia; the canal was filled with foreign bodies (calcium phosphate granules) [159]. Other models used the intramedullary insertion of titanium Kirschner wire instead of a canal drilled in the tibial metaphysis. Lucke (2003) studied some antibiotic delivery mechanisms after the fixation of a contaminated Kirschner wire (S. aureus) in the tibia. Bisland (2006) studied the effect of photodynamic therapy using a bioluminescent strain of biofilm-producer S. aureus grown onto Kirschner wires implanted in the tibial medullar cavity of Sprague-Dawley rats [166,167]. Akiyama (2013) investigated the antibacterial abilities of silver ions (Ag + ) within hydroxyapatite against an MRSA infection in the medullary canal of rat tibia, demonstrating a time-dependent Ag + release [168].
Beyond the use of rats as the tibial osteomyelitis model, some researchers used this rodent model to recreate femoral implant-associated chronic osteomyelitis. For example, Robinson (2011) developed a model of induced implant-associated osteomyelitis after fracture repair using intramedullary pins followed by inoculation with S. aureus (10 4 CFU/mL) [137]. In the same year, Ozturan studied the efficacy of moxifloxacin compared to teicoplanin in the treatment of a femoral model of osteomyelitis; the results demonstrated that moxifloxacin therapy was an effective alternative to teicoplanin [169].
Two old studies reported the design of a novel rat model for mandibular osteomyelitis (in one case HO); the first one was made in Russia by Solov'ev (1992), followed by the Swedish report of Hienz (1995). Solov'ev drilled the rat mandible and filled the empty cavity with bacteria, following the disease [170]. Hienz injected a 5% sodium morrhuate solution directly into the mandible and tibia, followed by an intravenous injection of S. aureus into the femoral vein to mimic hematogenous spread, effectively generating both mandibular and hematogenous osteomyelitis [171]. However, currently, the literature has few reports about rats' use as a model for mandibular osteomyelitis.
Model of total joint arthroplasty cannot be realized using rats mainly due to the small size of this rodent [165]. Kalteis (2006) contaminated the femoral cavity of rats with S. aureus ATCC 29213 and implanted a metal device together with two antibiotics (moxifloxacin or vancomycin). Results indicated that moxifloxacin was more effective than vancomycin [172]. Harrasser (2016) evaluated the effect of a titanium screw coated with hydroxyapatite alone or associated with a low concentration of silver against low and high doses of S. aureus ATCC 25923 introduced in the tibial metaphysis of rats. Infections resulted independently using both the bacterial concentrations even in hydroxyapatite alone or associated with a low concentration of silver [173]. In 2013, Haenle generated a tibial defect implanted with a titanium screw and contaminated with low and high inocula of S. aureus ATCC 25923 to verify a relationship between bacterial dose and severity of implant-associated infection. Results indicated a significantly high viable bacterial count in bone infected with a high bacterial dose [174]. Fan describes a joint replacement model in the rat with ultra-high molecular weight polyethylene and titanium components. A 3 mm deep hole was drilled in the tibia and immediately injected with S. aureus ATCC 12600, and a titanium screw was implanted. The results confirmed the development of a local infection for four weeks [175].
Beyond the well-recognized role of S. aureus in inducing experimental osteomyelitis, some researchers used other bacterial strains. [114]. In 1990, Nelson studied an antibioticresistant experimental rat model of Pseudomonas osteomyelitis without the injection of scle-rosing agents, obtaining a reliable model of chronic osteomyelitis [176]. In 2001, Hendricks developed a rat model of polymicrobial osteomyelitis using S. aureus and P. aeruginosa injected at different concentrations (10 2 , 10 3 , and 10 4 CFU/mL) into the spinous process lumbar vertebra of Sprague-Dawley rats. Results indicated the synergy between the two bacterial species when low levels of each organism were present in the wound [177]. In 2006, Mecikoglu developed a rat model of periprosthetic infection using S. epidermidis ATCC 35984 and Kirschner wire fixed with bone cement into the medullary femoral canal to investigate the eradication of a biofilm-related infection using a proteolytic enzyme (serratiopeptidase). The authors concluded that the enzyme helped bacterial eradication [178]. In 2016, Lovati developed a rat model of S. epidermidis-induced non-union of the femoral fracture using different concentrations (10 3 , 10 5 , and 10 8 CFU/mL) of a biofilm-producing methicillin-resistant S. epidermidis injected within fractures, following synthesis with stainless steel plates and screws. A dose-dependent effect between the bacterial inoculum and the non-union rate was demonstrated. Other studies involving rats as models of S. aureus bone infection after osteotomy, periosteal reaction, thickening of the cortex, myeloid hyperplasia, and polymorphonuclear cells in granulation tissue have been observed [114,179] ( Table 2).

Mouse Models
In 1974, Ueno reported the first mouse tibial model. Experimental osteomyelitis was induced by creating a canal into the tibial metaphysis of mice, which was successively filled with silk thread loaded with suspensions of S. aureus or P. aeruginosa [125]. Some years later, Kobayakawa (1979) used S. aureus to create a HO mouse model by inserting a silk thread into the proximal tibial metaphysis. Results demonstrated that when a foreign body was not used, S. aureus started to diffuse into most bones and later colonized the femur and tibia, while the foreign body promoted a localized infection [138]. In 1998, Pesanti and Lorenzo worked on Ueno's same model, using a silk thread with a low dose of S. aureus ATCC 49230 directly introduced within the tibial canal of animals. The study aimed to evaluate the activity of IL-4 in inhibiting the function of the osteoclasts during the bone repair process in chronic osteomyelitis [139]. An acute HO mouse model was developed in 1999 by Chandha by an epiphyseal fracture in the proximal tibia, followed by intravenous inoculation of S. aureus. The infected mice increased circulating B lymphocytes and CD4 + T lymphocytes. Histopathological findings confirmed that polymorphonuclear leukocytes infiltrated the proximal tibia [145]. In 2014, Lovati developed an implant-related infection in type I diabetic mice. A stainless-steel needle was covered with 10 3 CFU of S. aureus ATCC 25923 and inserted within the femoral canal of animals. Prostaglandin E1 (PGE 1 ) was systemically injected as a preventive osteomyelitis agent. Results indicated that the simultaneous administration of a PGE 1 enhanced the local blood flow and improved antibiotic therapy [180]. In mice models, bioluminescent bacterial strains are used to perform realtime tracking of pathogens through bone tissue of host immune cells. In 2008, Li described an implant-associated osteomyelitis murine model using a bioluminescent S. aureus strain (Xen29). A stainless steel pin was coated with the pathogen and implanted transcortically through the tibial metaphysis of mice. Serology found a progressive increase in IgM (protective humoral response) after one week, which converts to a specific IgG2b response by day 11 post-infection [181]. In 2010, Bernthal used bioluminescent S. aureus and genetically engineered mice expressing fluorescent neutrophils to develop a dose-related model of post-arthroplasty infection. After the infection of the femoral canal with bioluminescent S. aureus (ALC2906 strain), both bacteria and host neutrophils (engulfed by phagocyted pathogens) were monitored in real-time using an in situ vivo imaging technique. The result demonstrated the establishment of chronic osteomyelitis with a low grade of inoculum (5 × 10 2 CFU/mL) [182]. The osteomyelitis mouse model proposed in 2012 by Funao was the first that used in vivo bioluminescence imaging (BLI) without sacrificing mice enrolled. After the femur's surgical exposition, S. aureus Xen29 was inoculated into the medullary cavity. The authors concluded that a high proportion of granulocytes was detected in the infected group's peripheral blood after seven days. Moreover, serological analyses showed high levels of IL-1β, IL-6, and C-reactive protein [183]. Pribaz (2012) compared the biofilm formation of two bioluminescent S. aureus strains (Xen36 and Xen40) on both stainless steel and titanium K-wires implanted within the femoral canal of mice [184]. In 2002, Yoshi developed a mouse model of tibial osteomyelitis and determined, over time concentration of cytokine levels (IL-1β, IL-4, IL-6, and TNF-α) in bone. After the surgical exposition of the proximal tibia, a hole was drilled, and a silk suture (seeded S. aureus) was inserted. Results indicated an immediate release (early phase) of IL-1 β and IL-6 due to bone damage (which dramatically decreases after a few days), followed by a rapid production (latent phase) of IL-4 and TNF-α associated with histopathological changes (bone resorption and formation) [185]. In two studies focused on the periprosthetic joint infection through titanium K-wires in mice, Heim (2014Heim ( , 2015 studied the role of myeloid-derived suppressor cells (MDSC) and their pro-inflammatory activity associated with a bioluminescent and biofilm producer S. aureus. These two publications clarified the role of IL-12 in recruiting MDSC and impairing the phagocyte activity, reducing the pro-inflammatory events [186,187] ( Table 3).  Compared to the control group, huNSG mice have increased weight loss, osteolysis, and bacterial spread to internal organs. Moreover, through flow cytometry and immunohistochemistry, more human T cells are present in infected huNGS mice than in uninfected ones. [189] Abbreviations: na = not available.
Microorganisms 2022, 10, 1135 20 of 37 One of the essential advantages of using mice as models to mimic osteomyelitis is using bioimaging to deeply understand the intimate connection between immune response and pathogen dissemination without euthanizing the experimental group. Beyond this, the small size of this model could be positive to the reduced cost and easy handling, but on the other hand, it makes two-stage surgical revisions and multiple procedures in a single mouse more challenging.

Rabbit Models
In 1885, the first attempt to create an animal model to study osteomyelitis was performed with rabbits [118]. However, only several tries later, the original tibial model with chronic, progressive osteomyelitis in rabbits was described by Scheman in 1941 and then later enhanced by Norden and Kennedy in 1970. In this model, a sclerosing agent (5% sodium morrhuate) was inoculated in the tibia, followed by S. aureus to produce chronic osteomyelitis [122,158]. In 1973, Andriole created a model that maintained chronic staphylococcal osteomyelitis for a long period and introduced the possibility of using a foreign body (e.g., stainless steel pin). They fractured the rabbit's tibia using a simple three-pronged clamp and a stainless steel pin for fixation and were able to follow the infection for up to 18 months, and 88% of the rabbits with bacterial inoculation pinning presented evidence of chronic osteomyelitis [190]. These models relied on the virulence of Staphylococcus to create a model of the human disease and demonstrated that an extraneous substance or trauma was necessary to create a chronic, progressive infection. Later on, other authors used the same techniques without fracture [191,192].
When it comes to open fracture rabbit models, the first was published in 1982 by Ashhurst, primarily using the tibia [193]. Worlock then modified this model to generate a model with fixation of an intramedullary rod [194], and since then has been used by other studies [195]. Unfortunately, the intramedullary rod described by Worlock lacks rotational stability.
In 1991, Johansson developed the first model with anaerobic organisms. In this study, Ringer solution was inoculated in each animal's left femur, while in the right femur, after treating the proximal metaphysis with 5% sodium morrhuate, Bacteroides fragilis was inoculated. As a result, all animals developed osteomyelitis on both sides, but with more prominent symptoms on the right one [196].
Rabbit models created to study osteomyelitis show the importance of preparing the bone before inoculation for chronic, progressive infection. Nowadays, several investigators, like Schulz, use these resources to develop methods of applying local therapies to avoid complications of systemic antibiotics and improve efficacy [197].
Rabbits are the smallest animals that can be used as models for prosthesis-related osteomyelitis. In addition, rabbit models of knee arthroplasty components can be found as a model for joint replacements for humans [198], and in 2005, Craig developed a new arthroplasty model in which a metal screw and ultra-high molecular weight polyethylene washer were held to the non-articulating surface of the lateral femoral condyle [199]. The first hematogenous model involving trauma to the proximal epiphysis of the tibia was produced using a three-point bending force over the proximal part of the tibia, creating a reproducible shearing injury to the physis. This was followed by an intravenous injection of a high bacterial load and provided good histologic results [200,201]. In the 1990s, this model was revised by Johansson. Metallic hardware was implemented in the distal tibia without any fracture created before the fixation. After the incision had, S. aureus was inoculated into the auricular vein. Unfortunately, this model had a relatively poor infection and high mortality rates [202]. This could be due to the injury that creates a susceptible area to the host immune system, theoretically allowing greater and different infection rates (Table 4).
Nowadays, rabbits are used in many studies involving osteomyelitis because they are relatively inexpensive and because their size makes them versatile, being relatively easy to handle, manipulate, and maintain. To find a rabbit model to perform more local therapeutic strategies on the infected bone The new technique did not influence the motion of the hind limb and mimicked well the intramedullary pinning of long fractured bones, but it did pose some risks for postoperative infections [197] New Zealand White rabbits M ±4.2 kg MRSA. 0 CFU in one knee and 10 4 CFU in the contralateral knee (Group A) 10 2 CFU in one knee and 10 3 CFU in the contralateral knee (Group B). The radiographic signs of healing were completely absent in all the rabbits on 30th postoperative day, indicating inert bone ends. [203] Abbreviations: na = not available.

Poultry Models
The poultry model for osteomyelitis was developed to mimic the natural course of acute hematogenous osteomyelitis, often due to traumatic events in the bones and surrounding tissues. The model proposed by Emslie and Nade used S. aureus inoculated in chickens. The animals were sacrificed at different time points (6, 12, 24, 48, 96, 192 h). Lungs, liver, and kidneys showed no bacterial infections. The growth rate of animals in the S. aureus group compared to the control group significantly decreased. The average weight of the animals after 8 h post-injection decreased in the treated group, while it increased in the control group. The lesions were appreciated even after 12 h post-inoculation. Normal vascularization and supporting mesenchymal cells were missing around the bacterial foci. Inflammatory cells were present in the extravascular spaces and the lacunae of chondrocytes surrounding an abscess with necrotic areas. This model confirmed that the bacteria initially reach the metaphyseal vessels, and the occasional involvement of the epiphysis was analogous to what occurs in human infants. In addition, while the cartilage matrix prevents one further spread of bacteria, it creates a protective barrier that prevents access to inflammatory cells [129]. In 1990, Daum and colleagues developed a chicken model of staphylococcal bacteremia, septic arthritis, and osteomyelitis. Chickens were intravenously injected with three different S. aureus loads, developing bacteremia in 80%, 90%, and 100% of the treated groups of animals, respectively. Osteomyelitis was developed 1 up to 23 h after inoculation [204].
Further, researchers designed experimental avian models to develop osteomyelitis after intravenous injection of appropriate amount and strains of Staphylococcus spp. to induce bacteremia without triggering septicemia [205][206][207]. Researchers studied the role of double-stranded RNA (dsRNA) in the infected bone of a Staphylococcus-induced chicken osteomyelitis model at a molecular level. They showed that dsRNA accumulation in bone tissues activated NACHT, LRR, and PYD domain-containing protein (NLRP)3 inflammasome and increased IL18 in vivo and in vitro, thus identifying dsRNA as a new target for the treatment of osteomyelitis [208] (Table 5). In orthopedic research, dogs are recognized as one of the most suitable models due to their high similarity (compared with all other non-human species) with human bones in density and mineral composition [140].
In 1976, Deysine et al. proposed a canine model of acute hematogenous osteomyelitis, which involved the injection of barium sulfate with 10 5 CFU/mL of S. aureus into the tibial nutrient artery. Unfortunately, the induced lesions differed from the human ones [127,164]. Consequently, in 1983, Fitzgerald proposed a canine model of chronic osteomyelitis due to open fracture. In this case, S. aureus load was inserted into the dog's tibia within artificial support consisting of acrylic bone cement. The resulting infection was compatible with subacute osteomyelitis or infection following total knee arthroplasty. This model is useful for studying surgical procedures, evaluating the effectiveness of therapies, and investigating reactions from foreign bodies [126]. In 1985, Petty et al., using 187 dogs, evaluated the effects of different implant materials on the infection rate after bacterial inoculation. At the femur level, a suspension of bacteria was inserted (S. epidermidis, S. aureus, and E. coli), and then the different implants were placed. They saw that any material used increased the risk of S. aureus infection, while E. coli and S. epidermidis infections were successfully induced using polymethylmethacrylate (PMMA) polymerized in vivo [209]. In 1994, Garvin et al. used the model proposed by Fitzgerald to study gentamicin's effect on two different implants. After six weeks of treatment, the animals were sacrificed. The PMMA implant was removed intact, while only fragments of the polylactide/polyglycolide were extracted (this was due to reabsorption). Local therapy had better effects than systemic therapy. The reabsorption of the biodegradable implant suggested a good strategy to avoid removing the implant after therapy [164,210]. To study the occurrence, magnitude, and kinetics of bacteremia and the resultant osteomyelitis and septic arthritis From 1 to 23 h after inoculation, osteomyelitis remained uniform in continuously bacteremic animals [204] Abbreviations: na = not available. However, despite the positive anatomical attributes shown by these models, nowadays, few canine models are still developed and used, mostly due to ethical concerns and economic reasons.

Small Ruminants' Models (Sheep and Goat)
One of the most desirable characteristics of sheep that made them eligible for use as models is the size of their bones, which allow for mimicking long bone osteomyelitis; moreover, the rate of osteogenesis between humans and sheep remains similar. However, two microscopic differences make sheep not eligible for microenvironment studies; the sheep bones are denser and present fewer Harversian canals [140,211].
In 1997, Kaarsemaker et al. developed a sheep chronic osteomyelitis model for toxicological and therapeutic studies. A sclerotic agent and S. aureus were injected into the medullary cavity of the tibia. Chronic inflammation, osteolysis, new bone tissue, bacteria, and granulation tissue developed. Unlike canine models, using the sclerosing agent instead of an implant made this an excellent model for studying the factors influencing surgical therapies and drug release systems [132,164]. More recently, an intramedullary nailing contaminated fracture sheep model was developed. After the exposition of the tibia, an osteotomy was created. An inoculum of S. aureus was loaded on bovine type I collagen.
Researchers concluded that nail fixation might not be appropriate in all models because it could increase bacterial virulence, causing less efficacy of antibiotic treatment [212].
Unlike the anatomical similarities with sheep, goats' models have not been widely used and developed. However, like sheep, these animals can easily mimic human long bones osteomyelitis.
In 2005, Salgado et al. developed a model of chronic osteomyelitis. A hole was drilled, and sodium morrhuate and S. aureus were injected. Ultimately, 96% of the animals highlighted the typical pathology findings of osteomyelitis, both concerning radiographic and histological analysis [164,213]. Similar to the study by Salgado, Beardmore created a similar defect model by drilling a hole in the proximal tibial metaphysis of goats and filling it with S. aureus. Without using a sclerosing agent, osteomyelitis was induced [214] ( Table 6). To develop a surface modification of titanium fracture hardware with vancomycin to prevent bacterial colonization in a large animal model The modified titanium plates treated with antibiotic-derived compounds inhibited the colonization of the implant. Moreover, treated groups showed bone-healing. [217] Abbreviations: na = not available.

Porcine Models
The pulmonary intravascular macrophages, which decrease bacteremia and enhance survival following hematogenous injection of bacteria, are an essential feature that makes pigs the model of choice for HO procedures [218].
Jensen and colleagues used a stainless steel implant to mimic implant-associated tibial osteomyelitis. A K-wire was drilled in the medullary cavity of the tibia of SPF pigs (specific pathogen-free), and S. aureus was loaded in the artificial cavity with a small steel implant. Animals were monitored for five days, and all groups noted signs of localized, acute osteomyelitis to varying degrees on computed tomography (CT) scan and implant cavity cultures [219].
A porcine model of post-traumatic osteomyelitis was described in 2001 by Hill. A steel fragment was fired on the right tibia of pigs to mimic a ballistic injury; the wound was contaminated with S. aureus ATCC 29213 loaded on a strip of sterile bovine collagen placed into the bone defect. All animals developed osteomyelitis and infection of surrounding soft tissues. After 14 days, a radiographic examination showed a central radiolucent area and surrounding sclerosis. Histology examination confirmed the diagnosis of osteomyelitis after finding purulent material in bone and associated osteonecrosis [220].
In 2010, Jensen and colleagues developed a hematogenous porcine-based model. A catheter was placed in the left ear vein to inoculate S. aureus strain at different time points. In animals euthanized at 12, 24, and 48 h, long bones and lungs showed clear signs of infection. However, after 48 h, the pulmonary bacterial load diminished, and with no evidence of bacteremia, this was attributed to the pulmonary intravascular macrophages that can phagocyte S aureus [142,218]. In 2017, Jensen explored a recent modification of this model. An intra-arterial catheter was placed into the right femoral artery, followed by an S. aureus infusion. The pigs were monitored after inoculation and recovery from anesthesia the following days and noted signs of disease [221] (Table 7).
As stated above, the pharmacological approach to treating osteomyelitis resides in systemic antibiotic administration. Since an adequate antibiotic concentration can reach the infection site, the physiology of animal nutrition is essential in choosing a particular model. Pigs, like humans, are omnivores; they respond to antibiotics administration and diffusion similarly, making them eligible for oral and/or systemic antibiotic treatment evaluation [140,219]. Two of the main disadvantages of porcine are their rate of bone growth (faster than humans) and the length of both tibia and fibula, which limit their use in implant evaluation procedures [219].  To compare the infection potential of the porcine strain (S54F9) with two S. aureus strains of human origin (UAMS-1 and NCTC-8325-4) in this model and to examine the development of HO with a focus on pathology and the localization and microenvironment of S. aureus In three, one, and none of the recipients of porcine and human strains, respectively, bone lesions were present. On the CT scans, the vascularized bone tissue was seen as foci of increased opacity. [225] Yorkshire-Landrace crossbred pigs To describe a novel porcine implant associated osteomyelitis model.
A significantly higher volume of bone lesion, number of neutrophils, concentration of acute-phase proteins in serum and enlargement of regional lymph nodes were induced by a high inoculum. Therefore, a threshold of 40 neutrophils for 10 high power fields was considered for the histopathological diagnosis of high-grade IAO. [219] Landrace SPF F 35 kg S. aureus S45F9. 10 4 CFU Implant-associated osteomyelitis Tibia 2-4-6 days To elucidate how deep implant-associated osteomyelitis can go into the peri-implanted bone tissue within a week On implants and from 25 µm to 6 mm into pathological bone area S. aureus bacteria were identified. [228] Abbreviations: na = not available.

Conclusions
Using animal models to mimic and characterize human osteomyelitis remains a fundamental challenge and depends on the physiology and specific anatomical features of the animals used. The selection of the most appropriate model has to be done considering multiple characteristics: the disease under investigation, pros and cons of the specific animal model, type of experimental design, experimental time-points, reliability, and repeatability of the predicted model. Nowadays, used animals have specific advantages and disadvantages mainly derived from handling, costs, housing, dimension, and anatomical similarity to human lesions. For example, rats and mice have been overused since they were introduced as experimental animals. Rabbits and other large animals have adequate dimensions to ensure a more precise surgical intervention, but some are too hard to handle, and results are not repeatable between laboratories. The refinement of procedures, including animal models, the definition of new guidelines for a scientifically acceptable animal-driven experimentation, and the use of complementary technologies could help the scientists choose the most appropriate model and drive the research in finding new strategies to better understand also at molecular levels what happens during the crosstalk between pathogen multiplication, immune system, and bones.