Animal models of orthopaedic infections . A review of rabbit models used to induce long bone bacterial infections

Received 27 September 2018; Accepted 11 February 2019; Published 15 March 2019 Author affiliations: Laboratory of Clinical Chemistry and Microbiology, IRCCS Istituto Ortopedico Galeazzi, Milano, Italy; ‘Cantacuzino’ National MedicoMilitary Institute for Research and Development, Bucharest, Romania; Cell and Tissue Engineering Laboratory, IRCCS Istituto Ortopedico Galeazzi, Milano, Italy. *Correspondence: Arianna B. Lovati, arianna. lovati@ grupposandonato. it


IntRoductIon
In orthopaedics, implant procedures to re-establish the function of affected joints (primary and revision arthroplasties) and complex fractures are annually increasing worldwide.Infections following these orthopaedic procedures still represent a major complication in this field, with high morbidity for patients and increased hospitalization costs.To control intraoperative bacterial infections or chronic osteomyelitis related to the capability of bacteria to form biofilm, research is focused on the identification of novel prophylactic and therapeutic strategies.The multifaceted interactions among pathogens, host and treatments limit the responses obtained by in vitro studies, meaning that animal models must be used to translate findings to clinics.Of particular concern is the lack of a unique animal model to completely resemble the human condition.However, among several models of osteomyelitis and implant-related infections described in small rodents and large animals, the rabbit has been widely used as a reliable and reproducible model of orthopaedic infections.This review examines the relevance of rabbits to develop clinically representative models by analysing the pros and cons of the different approaches published in the literature.This analysis will aid in increasing our knowledge concerning orthopaedic infections by using this species.More importantly, the present analysis aims to be a tool for researchers who need to approach pre-clinical studies in the field of bone infection and have to identify the most appropriate animal model to verify their scientific hypothesis.

MEthods
The literature search was performed in the PubMed database by considering articles published in English from 2008 until February 2018.The search strategy was conducted by searching for 'rabbit' combined with the keywords 'osteomyelitis' , 'orthopaedic infection(s)' , 'bone infection(s)' , 'prosthetic infection(s)' , peri-prosthetic infection(s)' and 'implant-related

Abstract
The development of infections is one of the main complications in orthopaedics, especially in the presence of implants for the osteosynthesis of compound fractures and joint prosthesis.Indeed, foreign materials and implants act as substrates for the adhesion and proliferation of bacterial strains able to produce biofilm, causing peri-implant osteomyelitis.The eradication of biofilm remains a great challenge for the host immune system, as well as for medical and surgical approaches, thus imposing the need for new prophylactic and/or therapeutic strategies in which animal models have an essential role.In vivo orthopaedic models have mainly been used to study the pathogenesis of infections, biofilm behaviour and the efficacy of antimicrobial strategies, to select diagnostic techniques and test the efficacy of novel materials or surface modifications to impede both the establishment of bone infections and the associated septic loosening of implants.Among several models of osteomyelitis and implant-related infections described in small rodents and large animals, the rabbit has been widely used as a reliable and reproducible model of orthopaedic infections.This review examines the relevance of rabbits for the development of clinically representative models by analysing the pros and cons of the different approaches published in the literature.This analysis will aid in increasing our knowledge concerning orthopaedic infections by using this species.This review will be a tool for researchers who need to approach pre-clinical studies in the field of bone infection and have to identify the most appropriate animal model to verify their scientific hypothesis.infection(s)' .This search was focused on the rabbit models used in the field of orthopaedics to study new prophylactic or therapeutic strategies to fight against bacterial colonization and biofilm development, or to diagnose bone infections.

REsults
In this research, 708 studies were found, of which 298 were excluded because of duplicates.Only 382 studies were considered after the removal of non-English articles.One hundred and forty-two records were excluded due to their content being unrelated to orthopaedic infections.Of the 240 articles eligible for this analysis, 70 articles were not considered because they described infections unrelated to bone or orthopaedics.Moreover, 8 review articles, 7 in vitro studies and 28 spine infection articles were excluded because they did not fit the eligibility criteria for inclusion.Of the remaining records, another 12 articles were excluded.In particular, 10 in vivo studies described orthopaedic infections in animals different from rabbits and 2 articles were not found on the internet, or the authors never responded to our request to receive their studies.In conclusion, a total of 115 studies describing rabbit models were included in this review and analysed separately according to 4 different categories: acute or chronic osteomyelitis (n=60); foreign body or implant-related bone infections (n=36); peri-prosthetic joint infections with a partial joint replacement (n=10); and infected fractures (n=9) (Fig. 1).

Features of the rabbit as a model of bone infection: pros and cons
The rabbit is one of the most used animal species in musculoskeletal research, featuring in 35 % of all reported studies in this field [1].Numerous studies utilize rabbits, mainly thanks to the advantages of the species.First of all, rabbits are docile and easy to handle, and the costs associated with their management and surgical facilities are cheaper than those for large animals.More importantly, the size of rabbits, together with their availability and minimal phylogenetic development, satisfy the requirements of orthopaedic implant dimensions and ethical acceptability [2], thus bridging the gap between rodents and expensive large animal models for translational research.Indeed, the size of rabbits allows the implant of orthopaedic devices resembling the human situation in the case of implant-related infections (pins, screws, K-wires), partial joint arthroplasty (silicon elastomer devices) and fracture synthesis (plates and screws).For these purposes, the rabbit tibia allows an easy surgical approach and the femoral medullary canal is wide enough to facilitate the placement of internal nails [3].Indeed, the tibia (60.9%) and femur (27%) are the most used surgical sites in rabbit models of bone infections, whether related or unrelated to implant devices.Only a few studies have described rabbit infection models in different long bones, such as the radius (7.8%), humerus (3.5%) and ulna (0.8%) (Fig. 2).Despite their suitable dimensions, multiple and bulky implants are clearly limited, thus requiring rabbits heavier than 3 kg or larger species, with the exception of the femoral condyles, which can receive multiple small-sized implants or wide cylindrical implants [4].Among several advantages over other species, rabbits reach both sexual and skeletal maturity at around 5-6 months of age, thus permitting easy and quick breeding and providing a mature bone structure for orthopaedic purposes.About 68 % of the analysed records declared the gender of animal used to study bone infections; 51 % employed male rabbits, 42 % employed female rabbits and only 7 % used both genders.Males are used more often because they grow more quickly than females and reach greater dimensions.In fact, males gain 1 and 0.7 mm of femoral and tibial length, respectively, for every 2 weeks of age, compared to 0.2 and 0.4 mm for females [5].Again, both genetic standardization and the large number of individuals available for research purposes make the rabbit useful for multiple investigations.To satisfy the aforementioned requisites, the New Zealand rabbit is the most suitable strain in orthopaedic research, and has been reported as being used in 99 % of the studies.In this review, indeed, just one study used a different strain, specifically the Japanese White rabbit [6].
In general, rabbits are characterized by bone fragility due to having a small and light bone structure corresponding to 6-8 % of their body weight.Indeed, one of the major causes of failure for orthopaedic approaches is the high incidence of post-surgical femoral or spine fractures [7,8].Despite gross differences in the bone shape, mechanical load and microstructure of rabbit bones compared to those of humans, great similarities have been reported in terms of the bone mineral density of the long bones [9].Differences in bone microstructure and the faster bone remodelling could represent limitations for studying the osseointegration of implants, but they do not affect results in the case of bone infections, which mainly depend on the immune system response of the host, except for fracture healing associated with bacterial contaminations.The plantigrade stance of rabbit hindlimbs limits the examination of bone changes in the case of partial knee joint replacement.Regarding the host response, the rabbit immune system is genetically more similar to that of humans than are rodents' genes, with only small differences in tissue and cellular organization [10].For these reasons, rabbits have served as an experimental model for infectious diseases, with their infection susceptibility and pathogenesis being similar to those of humans [11].For example, sensitivity to Panton-Valentine leucocidin (PVL), a cytotoxin present in the majority of community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA), differs between species.Indeed, studies concerning the role of PVL are conducted in rabbits because, like humans, they are highly susceptible to this leukotoxin, while neutrophils from rodents are largely resistant [12][13][14].Moreover, the inflammatory response (periosteal reaction), the osteolytic lesions, the bone sequestra and the compromised range of motion in infected long bones in rabbits are comparable to those in humans [15].Unlike in the rodent response, rabbits do not experience the spontaneous eradication of bacterial contaminations, especially those mediated by S. aureus, and advance towards a chronic onset of infection.However, depending on the virulence of the bacterial strains used and on the number of colony-forming units (c.f.u.) inoculated to induce osteomyelitis; infections can spread from bone to other sites or determine a systemic bacteraemia.
Many studies have analysed the effectiveness of systemic and local antimicrobial agents on rabbits because they have similar pharmacokinetics to humans, despite this species being sensitive to specific antibiotics (penicillin, vancomycin, etc.), with severe adverse effects (diarrhoea, dehydration and mortality) being experienced, mainly in the case of longterm, high-dose antibiotic therapy [15,16].Furthermore, rabbits are employed to test antibiotic concentration in vivo and to study pharmacodynamics; thus this is a well-known species in the field of pharmacology [17].According to these premises, rabbit experimental models of bone infections allow us to control the type of bacteria and their load (clinical or subclinical infection), as well as the progression of disease.
With the purpose of establishing a detectable bone infection, Mariani et al. reproduced a rabbit model of osteomyelitis by injecting increasing bacterial loads (10 2 , 10 3 , 10 4 and 10 6 c.f.u.) of S. aureus ATCC 25923 [18].In particular, they monitored the progression of osteomyelitis by means of a PCR-based molecular diagnostic technique to overcome the limitation of standard microbiology and radiology to establish the presence of bacteria, especially in the case of low-grade inocula.
Finally, some of the other disadvantages of using rabbits as a model in orthopaedics are their greater inclination to post-operative distress and higher sensitivity to anaesthetic, analgesic and antibiotic drugs compared to small rodents.Indeed, extra post-operative care by expert personnel and veterinarians are fundamental to control the pain, discomfort and abnormal behaviour of these animals, which might lead to a higher rate of adverse events and mortality (gastrointestinal stasis and ileus, diarrhoea, bone fractures, bedsores, etc.) [7].Another limitation in the use of rabbits for the development of orthopaedic infections is the strict control of specific pathogens related to this species that could jeopardize the study results, thus requiring specific pathogen-free (SPF) facilities for rabbit housing.

Rabbit models of acute or chronic osteomyelitis
In Table 1, studies using a rabbit model of acute and chronic osteomyelitis are listed.The very first experimental rabbit models of acute and chronic osteomyelitis were those by Scheman in 1941 [19] and Norden in 1970 [20], respectively.Norden used a sclerosing agent within the tibial metaphysis injected with a S. aureus suspension to obtain a progressive chronic osteomyelitis.The rate of infection was 70 % or 100 %, depending on the bacterial load, low grade (10 3 c.f.u.) and high grade of infection (10 4 to 10 6 c.f.u.), respectively.Only a few records reported the rate of infection obtained by using Norden's procedure, with a mean of 89.6 % [21][22][23][24][25].Moreover, the grade of infection varied from mild to severe local bone changes in the analysed studies, especially when the time set for the development of osteomyelitis ranged from 2 to 4 weeks (3.4 mean), and according to the differences in the employed bacterial strain.In Table 1, 33 works used sclerosing agents, sodium morrhuate (29), sodium tetradecyl sulfate (3) and arachidonic acid (1), to induce bone aseptic necrosis and enhance bacterial attachment and proliferation after the injection of the micro-organism.This approach was mainly used to assess novel antibiotic treatments or systems to deliver antimicrobial agents [14,23,[26][27][28][29][30][31][32][33][34], among which bioactive borate bioglass was the most widely employed substance for this purpose over the last 10 years [21,24,25,[35][36][37][38][39].
Other studies used the model to assess the activity of silver or magnesium ions or of platelet derivates against infections [22,43], or against the activity of a vaccine to clear infections [13,40,50].
The models proposed by Scheman [19] and Norden [20] resembled a haematogenous osteomyelitis that commonly occurs in paediatric patients, and which becomes chronic when left untreated and diffuses through the tibial metaphysis.However, this type of infection is rare in adult patients, who are most frequently subjected to nosocomial infections or bacterial penetration in the case of traumatic injuries.Because of the wide panorama of bone contaminants in humans, most works have used S. aureus or MRSA strains to determine chronic osteomyelitis, except two studies that used Escherichia coli [39] or Propionibacterium acnes [44].The study of Yin et al. was unique in that it developed a rabbit model of chronic osteomyelitis by simulating multibacterial infections (MRSA, Pseudomonas aeruginosa, Klebsiella pneumoniae and Acinetobacter baumannii) in the presence or absence of sodium morrhuate, yielding a 100 % osteomyelitis induction rate and a well-tolerated severe tibial infection in all treated rabbits 8 weeks post-infection [42].Moreover, they also obtained the same rate of infection by injecting P. aeruginosa (10 7 c.f.u.ml −1 ) alone without the use of a sclerosing agent [42].
To better resemble human occurrence and to assess the activity of locally delivered treatments (antibiotic enriched cements or hydrogel, bioglass, etc.), several authors preferred to inject the bacteria directly in the absence of sclerosing agents [24, 52, 54, 55, 58-64, 66, 78].Others employed bone wax to confine bacteria and impair the ability of bone to clear the infection [79].
Two studies reported that the injection of bacteria without sclerosing agents resulted in a chronic osteomyelitis rate of 67.5 % [51,57].
To assess the early efficacy of antibiotic treatments in bone infections (3-4 days), some authors described an acute model of osteomyelitis without sclerosing agents by using MRSA with higher bacterial loads (10 8 to 10 9 c.f.u.) [67][68][69][70].This approach impeded the spontaneous elimination of bacteria by the host immune system [80].Two studies evaluated the effectiveness of decalcified bone matrix chips [71] or a coating hydrogel for titanium foils [72] as delivery vehicles for vancomycin to prevent the establishment of bone infection in femoral or tibial metaphysis, respectively.In these cases, bacteria were injected intraoperatively and the preventative strategies were applied immediately after, without waiting for the development of an acute or chronic osteomyelitis.Four studies described rabbit models of osteomyelitis developed by determining 1 cm critical-sized radial defects [74][75][76][77], which were to assess the efficacy of antibiotic-loaded scaffolds to prevent or treat osteomyelitis [74][75][76] or to diagnose infections during the acute phase of development (2 weeks) by means of 18 F-FDG-PET [77].In contrast to the segmental bone defect models, Brooks et al. performed a non-critical cortical defect in the radius to test resorbable antibioticeluting polymer composite bone void filler for the prevention of perioperative infections [73].

Rabbit models of foreign body or implant-related bone infections
Site-specific models that insert foreign bodies (cements, needles, pins, etc.) into the same sites used in clinics have been widely described to assess preventative or therapeutic strategies to impede or reduce biofilm formation [81].The implantation of a foreign body does indeed increase the possibility of developing biofilm-related infections in vivo compared to the previously described osteomyelitis models.These models are mainly used to study the influence of metallic or polymeric materials on biofilm behaviour when growing on different surfaces [81].Moreover, they might be used to test the efficacy of therapeutic strategies after the removal of contaminated implants and bone debridement, mimicking the procedures of the revision surgeries.In 1973, Andriole et al. used for the first time a stainless steel pin to synthesize an infected tibial fracture [82].In  [84,85].Indeed, daily exercise, daylight exposures that maintain good levels of vitamin D, and a mature skeletal for the animals lower the incidence of spontaneous fractures after pin insertion.
The remaining 26 records evaluated the implant-related infections through the direct insertion of devices into the bone structure or within the medullary canal of long bones.In the first study by Odekerken et al., a post-operative implantrelated infection simulating an acute trauma was generated and infection parameters (haematological, clinical and bone integrity parameters) during the course of the osteomyelitis development up to 6 weeks were provided [87].This interesting intravital approach model lowered the number of involved animals and contributed to a dynamic analysis of the pathology.Moreover, in this study, an imaging-based scoring system and a calcium-binding fluorophore-based histological evaluation have been developed to discriminate the grade of infection.Odekerken et al. also proposed a successful approach for the evaluation of a chronic model of implantrelated osteomyelitis [88,89].These authors performed 18 F-fluorodeoxyglucose micro-positron emission tomography ( 18 F-FDG micro-PET), post-mortem micro-computed tomography (micro-CT) and calcium binding fluorophorebased histology; however these investigations only provide optional information on the infection and are mainly related to bone mineralization.A different opinion was proposed by Chatziioannou et al. who found that 18 F-FDG PET/CT scan is a sensitive and specific tool to monitor therapeutic strategies in experimental foreign-body osteomyelitis and that 18 F-FDG PET/CT scans accurately distinguished between a bone infection with a metal implant from a bone with a metal implant but without infection [100].
Two authors employed a rabbit model of titanium implant-related infection to assess the osteoprotective and osteoinductive activities of tested coatings to ameliorate osseointegration in the presence of a bacterial infection, even if they did not investigate their antimicrobial property [90,96].The polymer/lipid combination has potential as a de novo drug delivery system to treat bacterial bone infections, including osteomyelitis [116] Table 2. Continued Continued [91][92][93][94][95][97][98][99] or treatment options to fight bone infections [56,[101][102][103][104].In particular, those studies that investigated preventative strategies used a rabbit infection prophylaxis model based on the study of Alt et al. [118].They created the infection by injecting bacteria directly within the medullary canal followed by the introduction of the implants to be tested [92][93][94][95][97][98][99].However, none of these studies reflected the clinical situation because they delivered antibiotics locally rather than using a systemic antibiotic prophylaxis.Indeed, conventionally in clinics, the local delivery of antimicrobial agents (i.e.antibiotic-impregnated cement) can be used in the treatment of chronic osteomyelitis as an adjuvant method of systemic antimicrobial treatment [119].

Most of the studies reported in
Most of the studies evaluated the efficacy of preventative strategies in long-term follow-up (4-10 weeks).By contrast, Giavaresi et al. and Neut et al. focused on the activity of an antibiotic-loaded hydrogel or hydroxyapatite (HA) coatings during the very early stage of a bacterial contamination (7 days), thus overcoming the potential masking of animal cell-mediated immunity [98,99].
Other studies used implant-related osteomyelitis models to test novel therapeutic strategies to eradicate an established chronic infection [56,[101][102][103][104][105].For this purpose, a modified Norden's model was developed by implanting a metal device within the tibial trabecular bone or femoral medullary canal to develop chronic osteomyelitis in 3 weeks [20].
Interestingly, Gahukamble et al. developed a model of tibial infection by using a stainless steel implant without any interlocking bolts to avoid any wear-induced particles or corrosion that could interfere with the development of osteomyelitis [106].They demonstrated that P. acnes can cause osteomyelitis in the absence of implant material wear debris without clinical signs of infection.By contrast, S. lugdunensis, an unusually virulent coagulase-negative staphylococcal (CoNS) species, confirmed its potentially pathogenic behaviour, underlining the fact that the biofilm formation is not a prerequisite for infection.More importantly, this model showed that the progression of implant-related osteomyelitis follows a bacterial species-specific course and highlights the potential of P. acnes and S. lugdunensis to cause implant-related osteomyelitis.
Among the authors who have described the use of metal implants to develop infections, two employed a bone defect with implants to test preventative strategies and implant osseointegration in the presence of bacteria [108,109].In addition to metal devices, other implantable materials can favour biofilm formation on their surfaces because of their porosity.According to a recent study, the higher porosity and the greater the surface, the greater the biofilm formation [107].Surdu-Bob et al. investigated various foreign bodies with different surface areas to increase the chance of biofilm formation and localization of infection [107].In particular, a S. aureus suspension was injected within three bone defects of the rabbit tibial metaphysis before the implantation of K-wires, stainless steel spheres and cotton meshes.In particular, the cotton mesh-based model provided a higher surface area c.f.u., colony-forming unit;MRSA, methicillin-resistant S. aureus;MRSE, methicillin-resistant S. epidermidis;nd, not defined;PMMA, polymethyl methacrylate.Table 2. Continued with the shortest disease onset time interval, the highest survival rate (90%) and the highest disease reproduction rate (90%).Moreover, this approach assured minimal invasion and reduced suffering for animals, and the possibility of easy removal of the foreign body before treatment administration.
To increase the surface area for the bacterial colonization, others implanted contaminated porous tantalum cylinders loaded with antibiotic-impregnated microspheres in radial segmental defects with the aim of investigating a preventative strategy against bone infection [110].However, the rate of success was low because only 64 % of the animals receiving control untreated implants developed an infection.
Regarding the use of foreign bodies, bone cements are widely used in the literature [111][112][113][114][115][116][117].In particular, Moojen et al. tested the efficacy of silver-containing PMMA bone cement to prevent MRSA infections [111].However, the silver-enriched cement had no effect on the prevention of infection, probably due to the presence of bacteria both at the implant surface and in the surrounding tissues.
Lankinen et al. used two different models of bone infection in rabbits in order to compare the 18 F-FDG-PET/CT characteristics in acute/subacute osteomyelitis with S. aureus and in foreign body-associated subacute/chronic osteomyelitis with S. epidermidis [114].In the case of S. aureus (10 5 c.f.u.ml −1 ), osteomyelitis was induced without the use of a sclerosing agent.Otherwise, a sodium morrhuate treatment together with bone cement left in place was employed to determine a bone infection with a lowvirulence S. epidermidis (10 9 c.f.u.ml −1 ).
Two studies evaluated the prophylactic efficacy [116] and the osteoinductive properties of calcium phosphate cements in contaminated rabbit bones [117].These models are highly relevant and common in the study of trauma patients with open contaminated fractures.

Rabbit models of peri-prosthetic joint infections and partial joint prosthesis
Peri-prosthetic joint infections (PJI) are rare, but they still comprise a large burden in orthopaedic surgery, with an incidence of 1-2 % in patients in the first 2 years after surgery.To better mimic what occurs in clinics, the rabbit is the smallest experimental animal suitable to perform a partial knee replacement with a tibial component (Table 3).The model, consisting of a silicone-elastomer implant (silastic HP) as a tibial prosthetic component and intraoperatively injected bacteria, was first described in 1996 [120], and was then used by several authors [121][122][123][124].
Rabbit models simulating an acute PJI were also developed by using different metal devices (titanium, stainless steel) implanted within the knee joint at the level of femoral condyles by press fitting screw or cylindrical devices to replicate intra-articular implantation (Table 3) [6,[125][126][127][128][129].Bacteria were then locally injected intraoperatively [125][126][127][128] or coated on the metal devices [6,129].The latter approach better resembles the contamination that might occur during the implant in primary arthroplasty, and it offers a surface on which bacteria can form biofilm.However, in these cases, the bacterial load is unknown.In these scenarios, to investigate the effectiveness of antimicrobial treatments a few days after implant surgery (4 to 7 days), a revision surgery is commonly performed to remove the infected devices and to carry out debridement at the site of infection to mimic what happens in humans.The models of PJI cause a concomitant bone and peri-implant soft tissue infection that could affect both the tested treatments and the analyses regarding the systemic response to infection.

Rabbit models of infected fractures
Impaired fracture healing can be classified as delayed union or nonunion, segmental defects and fracture-related osteomyelitis.Compound fractures are commonly synthesized with internal fixators or external fixators.In addition to the type of fixation, the complexity of stabilizing the fracture firmly in small laboratory animals represents a hurdle to the development of these models.For this reason, just nine studies reporting rabbit long bone fractures have been found and included in this review (Table 4).Indeed, rabbits give us the opportunity to use human-scale implants, and local treatments may be delivered in more controllable and realistic volumes.Moreover, the bone size of rabbits is large enough to permit easier use of plates and fixation screws.In the case of early acute osteomyelitis, surgical debridement without the removal of the osteosynthesis device could be a potential approach.Otherwise, antimicrobial therapy associated with the removal of the infected device is the preferred treatment method for late-stage osteomyelitis.The latter was used by Hamel et al. to compare the use of external and internal fixation associated with antibiotic treatment after the removal of the primary intramedullary nail used to establish a MRSA-infected fracture in rabbits [130].Most of the other studies that performed a long bone osteotomy employed an internal fixation with titanium or stainless steel plates and screws [49,[131][132][133][134][135].However, high exclusion rates or clinically unrealistic implant systems have been described [133].A more realistic, standardized and replicable study was performed by Arens et al. who modelled a peri-operatively contaminated humeral open fracture fixed with either a human-scaled locked plate or a custom-designed interlocked intramedullary nail [132].The use of humerus, rather than femur or tibia, is meant to minimize the weight bearing on the operated limb, since rabbits predominantly rest on their hind limbs, and also to avoid a high failure rate due to the greater forces applied in the hind limbs [132,134].Implant stability is a major factor in the outcomes of experimental infections in animals.Thus, plate and screw fixation should be preferred to avoid the rotational movements that commonly occur when using intramedullary nails, which can lead to misleading results regarding the development of infections.Importantly, partial plate fixation before completion of the osteotomy is highly recommended to overcome the difficulty of anatomical reduction, and to avoid high surgical failure rates, as well as animal mortality.Using the fracture model, ter Boo et al. c.f.u., colony-forming unit;MRSA, methicillin-resistant S. aureus;nd, not defined.Table 4. Continued injected planktonic bacteria to study an antibiotic-enriched preventive treatment [134], while Zhang et al. evaluated a treatment option for infections by implanting plates already colonized with mature bacterial biofilm [49].Finally, two studies described models of long bone ostectomy by removing a 1 to 1.5 cm bone segment to study bone regeneration or a locally delivered antibiotic therapy in the presence of an infected bone gap, with or without fixators [136,137].
Unfortunately, both studies determined a critical-sized bone defect that can produce misleading results when associated with bone infection.Indeed, a critical bone defect is also unable to spontaneously heal over the lifespan of the animals in the absence of infection.Thus, models of bone defects should avoid the combination of critical defects and bacterial inoculation to better understand the pathogenesis or to assess the effectiveness of a therapy against infection [138,139].

Ethical concerns in animal models of bone infections
The development of animal models of bone infections presents a high risk of mortality due to the potential spread of a bacteraemia, but also occurrences of physiological responses leading to general status impairment, such as severe body temperature changes, inappetence and consequent weight loss, systemic inflammatory reactions and pain.In the analysed studies, the mortality rate in models using sclerosing agents was mainly found to range between 9 and 23 % (mean 13%) due to the bacteraemia developed by the high bacterial load (10 5 -10 8 c.f.u.) [14,32,38,39,47].However, mortality rate is not comparable among different studies because the virulence of the various bacterial strains differs.Only two studies using the same MRSA stain (ATCC 43300) obtained the highest (23%) [38] and the lowest (9%) mortality rate [47], which was related to high (10 8 c.f.u.) and low (10 5 c.f.u.) bacterial loads, respectively.To minimize the mortality rate, humane endpoints for animal models of infectious diseases should be established to impede spontaneous death and minimize suffering [140].Particular attention must be paid to the objective, quantitative parameters: body temperature changes (±4 °C), rapid or progressive weight loss (10 to 20 %), C-reactive protein (CRP) levels and white blood cells alterations [88,89].Moreover, the clinical signs of distress, such as dehydration, persistent recumbency, loss of reflex and neurological abnormalities, and self-induced trauma, must be evaluated over the time of the experiments.In cases where alarming changes occur, analgesia or specific antibiotic treatments could be administrated; however, these measures could mask or compromise the experimental integrity of the study on infections.Hence, animal suffering must be minimized while meeting scientific needs.conclusions Rabbit hind limbs are commonly used to create models of bone infections.In particular, the tibia is easily reachable due to the minimal presence of soft tissue, and the femur has adequate dimensions to permit the introduction of internal fixators or to fix plates in order to mimic implant-related infection models.Mature rabbit condyles can also carry 3-4 mm defects to implant metal or material foreign bodies.In general, rabbit bones and joints are large enough to accept partial joint prosthesis.Despite these advantages, there are significant differences between rabbit and human bone.The rate of bone remodelling is faster in young rabbits, which are often used in these studies, and this makes them less representative of the bone response to infections in adult humans.The use of mature animals is recommended to overcome this limitation.Indeed, rabbits have the advantage over other animal models of reaching skeletal maturity at a relatively early age, which limits research costs.

Fig. 1 .
Fig. 1.Research strategy.Flow diagram of the selection process.

Fig. 2 .
Fig. 2. Schematic representation of rabbit models to induce bone infections based on the results of the literature search.

Table 1 .
Continued Continued Bottagisio et al., Journal of Medical Microbiology 2019;68:506-537 To investigate the effect of Herba Epimedii on bone repair after antiinfection treatment in vivo.The bioactive-composition group of H. Epimedii (BCGE) contained four flavonoids with a total content of 43.34 %.

Table 2 .
Table 2 developed a rabbit model of implant-related infection to investigate preventative Bottagisio et al., Journal of Medical Microbiology 2019;68:506-537 Rabbit models of foreign body and implant-related orthopaedic infections (36 records)

Table 4 .
Rabbit models of infected fractures (nine records)