Staphylococcus aureus Osteomyelitis: Bone, Bugs, and Surgery

OSTEOMYELITIS: HISTORICAL PERSPECTIVES, CLINICAL PRESENTATION, AND DIAGNOSIS

Osteomyelitis, technically defined, is an inflammatory state of bone most commonly caused by infection. Three clinical mechanisms lead to bone infection: osteomyelitis resulting from the spread of a contiguous source (trauma or surgical contamination), osteomyelitis occurring secondary to vascular insufficiency or neuropathy (e.g., diabetic foot ulcers), and acute hematogenous osteomyelitis, which is more common in pediatric patients (1, 2). Each clinical pattern has its own unique characteristics, risk factors, presentation, and etiology. Across all mechanisms leading to osteomyelitis, Staphylococcus aureus is the most common organism (1, 3, 4). This includes methicillin-resistant S. aureus (MRSA), which is responsible for a large proportion of antimicrobial-resistant infections in the United States. Accordingly, this review will focus on the role of S. aureus in osteomyelitis. The review will also highlight a basic approach to the diagnosis and treatment of osteomyelitis to help direct future research on bacterial pathogenesis.

Osteomyelitis causes substantial morbidity and mortality and is a major challenge in orthopedic surgery. In trauma, one of the most devastating complications from open fractures is infection, which is a primary mechanism for fracture nonunion (5). In total joint arthroplasty, one of the most common major surgical procedures, the primary cause for failure is infection (6–9). Diabetic foot ulcers are the most common cause for below-knee amputations in the developed world. In the pediatric population, a septic joint is one of the few emergent conditions in orthopedics that demand immediate operative intervention in most cases (2). All of these disease processes highlight the difficulty in eradicating osteomyelitis and the frequent failure of treatment modalities. As such, mortality can be as high as 8% (10).

Osteomyelitis is a historically significant disease. Although initially described by Hippocrates, it was first recognized as a type of infective abscess in medulla in the 18th century (11). Before the appreciation of infectious etiologies, the inflammatory component in bone had long been observed, leading to the term of osteomyelitis—an inflammation of bone. Prior to the advent of penicillin, the only relevant treatment approach was surgery, where the classic tenants of debridement were developed (12). The surgical author of this article respectfully notes the important contribution of penicillin, a nonoperative approach, that dramatically altered treatment when it was initially reported (13) but also observes later reports that gentle debridement can be used as a monotherapy alone in specific cases (14). Surgical versus conservative approaches to the treatment of osteomyelitis echo broader debates between surgery and medicine, and it is reasonable to assume this was one of the first respectful and collegial battlegrounds between the fields (15).

A standard approach to diagnosing osteomyelitis is based on clinical suspicion. Symptomatic presentation includes pain in an extremity and, in pediatrics, frequently a swollen extremity with possible redness. Initial evaluation includes measurement of serum C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR), radiographs, and a blood culture. CRP levels typically are elevated before the ESR, but ESR decreases significantly more slowly than CRP (16). This can allow CRP to monitor acute response to treatment (17). It is important to note that elevated ESR and CRP levels are nonspecific for osteomyelitis. Many different inflammatory states will have an elevated ESR and CRP, but it is atypical for a patient with acute osteomyelitis to have a normal ESR and CRP level. Osteolytic lesions around the involved bone, described as “rat bite” in nature, can help aid in the diagnosis. Radiographs cannot not rule out osteomyelitis, as it takes up to 2 weeks in children and longer in adults for imaging findings to be present. However, radiographs are important to rule out other diagnoses, including fracture and malignancy. Malignancy, in particular, can masquerade as infection and infection can appear as a malignancy, and so careful evaluation is always warranted. Definitive imaging of osteomyelitis should be completed with magnetic resonance imaging (MRI), which is considered the gold standard for imaging of osteomyelitis, as the associated inflammation can be observed much earlier in the disease process than with plain films (18). Computed tomography (CT), scintigraphy, and ultrasound have also been used to diagnose osteomyelitis, especially in those cases where patients have contraindications to MRI. For CT, contrast should be used when possible to delineate possible abscess formation. Following imaging, cultures should be obtained to guide therapy. This is typically accomplished through percutaneous biopsy or by direct sampling of bone during open debridement. It is important to appreciate that cultures may not always be positive, and this is termed culture-negative infection. Here, the bone is infected but an organism is unable to be cultured (19).

OSTEOMYELITIS PATHOPHYSIOLOGY

All osteomyelitis cases are not the same. The pathophysiology and etiology are dependent on the mechanism of disease development, the anatomic location in bone, and structural differences between adults and children. Acute hematogenous osteomyelitis is hypothesized to be associated with the venous architecture of certain anatomic locations within the skeleton, such as around the lumber spine or physes (growth plates). Blood vessels in these areas are thought to have relatively turbulent blood flow, to have limited phagocytic lining cells, and to be more susceptible to injury, creating a potential nidus for escape of bacteria from the bloodstream. Osteomyelitis is also associated with trauma, surgery, soft tissue spread, or periprosthetic implants, where contiguous infection or traumatic inoculation of microorganisms leads to establishment of bone infection. A final type of osteomyelitis is associated with vascular insufficiency, such as in chronic wounds of patients with peripheral vascular disease or diabetes, where limited blood flow or repetitive trauma create a nidus for infection and dysregulate immune responses (20, 21). Other host factors such as edema, tobacco use, immune status, metabolic disease, and malnutrition also contribute to the incidence and severity of osteomyelitis (22). Diabetes plays an especially important role, as it both perturbs the immune response and compromises the macrovascular and microvascular blood supply.

As mentioned above, acute hematogenous osteomyelitis is typically associated with venous pooling that is either physiologic or occurs after subacute trauma, and transient bacteremia subsequently results in deposition of microbes near the metaphysis (23). Acute hematogenous osteomyelitis is significantly more common in the pediatric population given the vascular architecture of the physes. Prior to ∼12 to 18 months of age, transphyseal vessels allow bacteria in the metaphysis to migrate to the epiphysis. Bacteria can also migrate through local vasculature and the Haversian and Volkmann canal systems to spread throughout an entire long bone. Inflammation from the initial immune response results in increased intramedullary pressure and decreased blood flow. S. aureus also contributes to bone thrombosis through the production of coagulases, which in conjunction with local inflammation, leads to bone necrosis (23, 24). A subperiosteal abscess will develop if the infection violates the metaphyseal cortex. This is the result of the increased inflammation and pressure from edema and purulence. The periosteum provides a major portion of the local blood supply to bone. Periosteal elevation from inflammation and purulence therefore limits local perfusion, leading to further necrosis. Necrotic bone segments become a nidus for chronic infection and are known as sequestra. In response to the formation of sequestra, the periosteum provides progenitor cells that contribute to local bone growth, and this walls off the infection with new mineralized bone, forming a pathological lesion known as an involucrum (20). Sequestrum and involucrum formation are not typically visualized until after ∼6 weeks of infection and are more indicative of chronic infection. A final bone lesion that can form in subacute or chronic osteomyelitis is a Brodie’s abscess, which occurs when sclerotic bone surrounds a purulent cavity (Fig. 1).

• Open in new tab
• Download powerpoint

FIG 1

Typical features of chronic osteomyelitis. Initial inflammation and infection in the metaphysis lead to necrotic bone becoming a nidus for chronic infection, known as a sequestrum. In an attempt to control the infection, new bone mineralizes around the sequestrum and is termed the involucrum. Active infection results in the formation of a sinus to the outer periosteum where a periosteal abscess can develop. (Copyright Kenneth L. Urish.)

There are important differences in the pathogeneses between pediatric and adult osteomyelitis and their sequalae. The most important difference involves the role of the physis in pediatric osteomyelitis and the connected risk of developing septic arthritis. The physis, or growth plate, is a specific well-defined area in bone between the epiphysis and metaphysis where the majority of longitudinal growth occurs until the child reaches their final height. It is composed completely of cartilage. The physis is a natural barrier to the spread of infection, as blood vessels and the interconnections of osteons do not cross this cartilage surface. In the hip, shoulder, elbow, and ankle, the joint capsule extends beyond the physis to the metaphysis. Once subacute osteomyelitis breaks through the cortex, it allows rapid spread into the joint. This is the clinical reason why septic arthritis is more common in the hip, elbow, and ankle in children than in the knee. An exception includes infants below 1 year of age, where the physis is well vascularized and allows bacteria to spread directly into the epiphysis and joint (24, 25) (Fig. 2).

• Open in new tab
• Download powerpoint

FIG 2

Pathogenesis of osteomyelitis-associated septic arthritis. Thrombosis of the venous and arterial vascular loops in the metaphysis leads to decreased blood flow, bacterial attachment, and acute infection. (A) The physis forms a physical barrier preventing spread of the infection into the epiphysis. (B) As the infection spreads, it reaches the metaphyseal periosteum and develops a periosteal abscess. In the hip, shoulder, elbow, and ankle, the joint capsule attaches below the physis. (C) This allows the periosteal abscess to circumvent the vascular barrier of the physis and invade the joint, resulting in a septic joint (25). (Copyright Kenneth L. Urish.)

INFECTIOUS ETIOLOGIES OF OSTEOMYELITIS

As noted above, the term “osteomyelitis” technically refers to inflammation of bone, which can result from both infectious and noninfectious etiologies. One particularly important noninfectious etiology of osteomyelitis that can mimic infection is chronic nonbacterial osteomyelitis (CNO), which is also known as chronic recurrent multifocal osteomyelitis (CRMO). CNO is an autoinflammatory disease that triggers sterile bone lesions that may affect one or more locations in the skeleton, and the clinical presentation can be strikingly similar to that of infectious osteomyelitis (26). Although the cause of CNO has not been fully elucidated, recent studies have implicated specific genetic loci (27, 28). Readers are directed to several excellent reviews on this topic for more information (26, 27, 29).

The vast majority of cases of osteomyelitis are caused by infection. Although fungal species such as Blastomyces, Coccidioides, Candida, and Aspergillus cause osteomyelitis in certain clinical scenarios, bacteria are the preeminent microbial agents of human osteomyelitis. The likelihood of isolating a given bacterial species from infected bone is heavily influenced by the mechanism of disease development (i.e., hematogenous, posttraumatic, or contiguous spread), host comorbidities, and environmental exposures.

Broadly speaking, acute hematogenous osteomyelitis is typically a monomicrobial disease, whereas osteomyelitis stemming from trauma or contiguous infection is often polymicrobial. The Gram-positive bacterium Staphylococcus aureus is the most frequently isolated etiologic agent of osteomyelitis across all disease mechanisms and is discussed in greater detail below. Although S. aureus causes the vast majority of acute hematogenous osteomyelitis cases, Streptococcus pyogenes, Streptococcus pneumoniae, and Kingella kingae are also important causes (30, 31). Unlike staphylococcal and streptococcal spp., Kingella kingae, a fastidious Gram-negative coccobacillus, often presents indolently and occurs most frequently in children of less than 3 to 5 years of age (32). In neonates, two additional bacterial causes of acute hematogenous osteomyelitis include group B Streptococcus and enteric Gram-negative bacteria, reflecting their known prevalence as agents of sepsis in this age group (33).

The microbiology of osteomyelitis following trauma is dependent on the mechanism and severity of injury, the presence of environmental contamination, and whether or not indwelling devices are required to stabilize damaged bone. The incidence of infection following skeletal trauma ranges from as low as 2% for low-grade open fractures to up to 50% for the most severe injuries (34–37). This clinical observation forms the basis for administration of short courses of prophylactic antibiotics in the setting of open fracture, a practice confirmed to reduce infection incidence in clinical trials and by a Cochrane Database systemic review (38, 39). In open fractures without gross environmental contamination, osteomyelitis typically is caused by skin flora, with a preponderance of cases caused by S. aureus and coagulase-negative staphylococci. However, in open fractures with gross contamination, an array of environmental organisms, including Gram negatives (e.g., Pseudomonas aeruginosa, Enterobacter cloacae, and Escherichia coli), other Gram positives (e.g., Bacillus and Enterococcus spp.), anaerobes (e.g., Clostridium spp), and nontuberculous mycobacteria, can also lead to subsequent osteomyelitis (20, 38, 40–42). Moreover, patients with high-grade open fractures and extensive soft tissue damage are at enhanced risk of nosocomial infection with pathogens such as methicillin-susceptible S. aureus (MSSA)/MRSA and multidrug-resistant Gram negatives, possibly due to the propensity for delayed wound closure and operative fixation (41, 43). When these injuries are sustained during combat, the microbiology of Gram-negative infection expands to include particularly drug-resistant organisms such as Acinetobacter baumannii (44, 45). Unfortunately, cultures of open fractures at initial operative intervention are poorly predictive of subsequent osteomyelitis etiologies; therefore, apart from injury severity, it is difficult to predict which patients will progress to infection (46–48).

Osteomyelitis following contiguous infection most commonly occurs in patients with neuropathic or vascular insufficiency wounds, decubitus ulcers, and severe or penetrating trauma to soft tissues. This mode of osteomyelitis is particularly common in individuals suffering from diabetes, where up to one-quarter of all patients will develop extremity ulcers, and of these, a substantial proportion will involve bone (49, 50). These infections are almost always polymicrobial, with S. aureus, coagulase-negative staphylococci, group B streptococci, enterococci, and Gram-negative bacilli frequently isolated from bone biopsy specimens (51, 52). Yet, traditional cultures almost certainly underestimate the microbial diversity of chronic wound infections involving bone. 16S sequencing studies have revealed a complex chronic wound microbiome that is dynamic and compositionally different from the surrounding skin (53–57).

S. aureus deserves special consideration as an etiologic agent of osteomyelitis given its prevalence across all mechanisms of osteomyelitis development and based on the considerable focus on staphylococcal pathogenesis within the osteomyelitis literature. The preeminence of S. aureus as an etiologic agent of osteomyelitis is likely driven by two factors. First, approximately 25% to 30% of the global population is thought to be colonized by S. aureus, with estimates reaching 50% to 70% when considering health care workers and transient colonization (58). Second, staphylococci produce many virulence factors, including adhesins, cytolytic toxins, immunoevasion factors, superantigens, and antioxidant systems (59–61). The involvement of specific S. aureus virulence factors in the pathogenesis of osteomyelitis has been inferred from animal models and observation studies in humans and is discussed in greater detail below.

Although a wide variety of microorganisms are recognized as etiologic agents of osteomyelitis, it is important to note that cultures from affected bone during osteomyelitis may be negative. Depending on the patient population under study, osteomyelitis is culture negative in up to 30% to 40% of cases, with chronic infections being the least likely to yield a pathogen (62–66). In some cases of culture-negative osteomyelitis, an offending microbe is not isolated due to preceding antibiotic therapy. However, in other cases, the modality of microbiology sampling (i.e., open debridement versus image-guided biopsy) and the presence of low bacterial load or fastidious organisms (e.g., Kingella and Propionibacterium acnes) impact culture positivity. The potential for false-negative culture results has led many infectious disease clinicians to advocate for withholding antibiotics prior to microbiologic sampling in select cases of osteomyelitis. For those patients who require operative debridement, this is likely an optimal practice. However, data are conflicting regarding the extent to which preculture antibiotics impact subsequent culture results. In pediatric acute hematogenous osteomyelitis, for example, short courses of empirical antibiotics do not appear to significantly impact culture positivity (64–68). Yet, these observations must be balanced against the clear improvement in patient outcomes, including antibiotic deescalation and exclusion of alternative diagnoses (e.g., CNO), which have been shown to stem from clinical practice guidelines that encourage deferral of antibiotics before source culture (69).

TREATMENT OF OSTEOMYELITIS

The mainstays of osteomyelitis treatment are targeted antimicrobial therapy, source control, and correction of medical comorbidities. Targeted antimicrobial therapy, by definition, requires identification of the offending microorganism by source or blood culture. Bone culture may not be feasible in all patients prior to administration of empirical antibiotics. Source control entails debridement of necrotic bone, drainage and irrigation of abscessed tissue, and removal of infected hardware if feasible. Correction of hyperglycemia and amelioration of peripheral vascular disease are critical components of therapy in patients with osteomyelitis related to chronic wounds.

Treatment of osteomyelitis is improved by evidenced-based clinical guidelines that focus on various aspects of care, including coordination of multidisciplinary care, risk stratification, and antimicrobial therapy. Copley and colleagues established clinical care guidelines (CCG) in the context of a Multi-Disciplinary Musculoskeletal Infection Program at Children’s Hospital of Dallas (70). The CCG involved daily rounding with orthopedics, pediatrics, infectious disease, nursing, and social work services. Patients with suspected musculoskeletal infection were enrolled in a clinical algorithm with interventions based on initial MRI findings and lab work. A key to the success of this clinical pathway was the creation of a dedicated MRI slot for children with suspected musculoskeletal infection, enabling immediate transition to the operating room if necessary. Patients enrolled in the CCG were significantly more likely to have a blood culture drawn before antibiotic administration, to have a pathogen identified, to have fewer antibiotic changes, and to receive a longer duration of oral therapy. The same research group went on to create guidelines for culture of musculoskeletal infection aimed at minimizing low-yield cultures in those children without suggestive clinical histories (65). Spruiell et al. similarly reported the implementation of a multidisciplinary CCG for pediatric musculoskeletal infections at Children’s Hospital Colorado (69). Key elements of this CCG were an emphasis on obtaining at least 2 blood cultures prior to antibiotics, early transition to oral antibiotics, avoidance of long-term intravenous access when feasible, and coordination of care between services. After implementation of the CCG, the authors noted significant reductions in length of stay, duration of intravenous (i.v.) therapy, fever duration, and time to culture, with a trend toward increased frequency of pathogen identification. In addition to CCGs, several groups have created classification and scoring systems to predict disease severity and adverse outcomes. Copley et al. created a scoring system for assessment of severity in acute hematogenous osteomyelitis using clinical (respiratory rate, intensive care unit [ICU] admission, febrile days), laboratory (CRP), and radiographic (evidence of disseminated infection) findings (71). This scoring system performed well on both initial and subsequent validation cohorts, and correlated with length of stay (71, 72). Mignemi et al. also established a classification system for musculoskeletal infection severity with significant correlation to hospital outcomes and inflammatory markers (73). Collectively, these tools allow clinicians to stratify patients according to severity and make personalized treatment decisions.

Before reviewing evidenced-based guidelines for the provision of antibiotic therapy for osteomyelitis, it is important to consider antibiotic penetration into bone. Two recent systematic reviews compiled pharmacokinetic data on antibiotic penetration into bone (74, 75). Most of the data on antibiotic penetration into human bone is obtained by administering doses to patients who subsequently undergo total hip or knee arthroplasties. In these studies, most antibiotic classes had substantial penetration into bone, with median serum/bone concentrations ranging from approximately 0.1 to 0.9 (75). However, pharmacology studies in healthy tissue failed to model elements of infection such as bacterial biofilm, abscess formation, and necrotic bone, which could dramatically impact antibiotic penetration in the setting of osteomyelitis. This highlights the important role of surgery in the debridement of necrotic and nonviable tissue to improve antibiotic delivery, as discussed below (7). Furthermore, there has been a greater appreciation for the increased antibiotic tolerance of biofilms (9, 76–80). Finally, the characteristic structure of staphylococcal communities within abscesses is such that the bacteria are shielded from innate immune cells, and potentially antibiotics, by a rim of fibrin and other host factors (81).

There are several evidenced-based clinical guidelines related to antimicrobial treatment of osteomyelitis. A central focus of each guideline is tailoring of therapy based on source culture results. However, empirical therapy is a mainstay of osteomyelitis care for patients who are suspected to have severe or disseminated disease, for patients who have pending source cultures, and for patients in which source culturing was not feasible. Empirical therapy decisions have been heavily influenced by the emergence of community-acquired MRSA and are also shaped by local antibiograms, patient history, and disease severity. Clindamycin is a particularly common choice for empirical antibiotic coverage of suspected S. aureus osteomyelitis, but utility can be limited by resistance patterns (>20% at some institutions) and the fact that it is a bacteriostatic antibiotic and therefore not ideal for patients who are critically ill. Additionally, recent antibiotic susceptibility trends in pediatric S. aureus isolates indicate an overall increase in the incidence of MSSA relative to that of MRSA and a trend toward increased clindamycin resistance in MSSA isolates (82). Collectively, these observations underlie the common practice of targeting both MSSA and MRSA in empirical therapy for osteomyelitis.

For MRSA implant-associated osteomyelitis, the IDSA (Infectious Disease Society of America) recommends empirical therapy with an i.v. agent (vancomycin or daptomycin) plus rifampin for the initial 2 weeks, followed by rifampin plus another oral agent (fluoroquinolone, trimethoprim-sulfamethoxazole, a tetracycline, or clindamycin) to complete 3 to 6 months of therapy in early-onset infections (83). Long-term suppressive oral antibiotic therapy is recommended for those patients in whom debridement and hardware removal are not feasible. For pediatric acute hematogenous osteomyelitis, the IDSA guidelines recommend empirical vancomycin therapy, with the consideration of i.v. clindamycin for patients who do not have ongoing bacteremia and in hospitals where clindamycin resistance is historically low (83). A minimum duration of 4 to 6 weeks is recommended. For native vertebral osteomyelitis, IDSA guidelines recommend empirical coverage of staphylococci, streptococci, and Gram-negative bacilli with a combination of vancomycin and a third- or fourth-generation cephalosporin (84). Empirical therapy should be tailored based on culture results and then targeted parenteral or highly bioavailable oral antibiotics given for 6 weeks. Finally, IDSA recommendations for diabetic foot osteomyelitis include an initial parenteral regimen followed by prolonged oral therapy, but there are insufficient data to support a particular empirical regimen or duration of therapy (85). In addition to IDSA guidelines, the European Society for Pediatric Infectious Disease (ESPID) has published clinical practice guidelines for pediatric osteomyelitis (86). ESPID guidelines recommend prompt initiation of empirical therapy after cultures are obtained, with regimens targeting both MSSA and MRSA in regions where MRSA prevalence is higher than 10% to 15% of all S. aureus infections. A typical regimen would therefore include clindamycin with or without an antistaphylococcal beta-lactam. For children with severe infection or at high risk of clindamycin-resistant MRSA, vancomycin is a preferred empirical agent, with or without the inclusion of clindamycin or an antistaphylococcal beta-lactam. Furthermore, coverage for Kingella kingae with a first-generation cephalosporin such as cefazolin should be included for young children (<5 years) with osteomyelitis. Duration of therapy is a minimum of 3 to 4 weeks, with an early switch to oral antibiotics guided by susceptibility patterns and clinical improvement of the patient (86). Although initially a controversial aspect of therapy for osteomyelitis, there is now compelling data that prolonged intravenous therapy is not superior to oral therapy for acute hematogenous osteomyelitis and may in fact lead to more complications (69, 87).

The role of surgery in osteomyelitis follows a standard surgical principle for infection: debridement (12). General indications for operative intervention include the presence of a subperiosteal or soft tissue abscess, osseous sequestrum, or signs of chronic infection such as in the sinus tract. Other indications include when the infection fails to respond to initial antibiotic therapy or concurrent septic joint arthritis (24, 88, 89). Surgery serves an important role in improving the local environment during osteomyelitis through debridement of the devitalized tissue, decompression of the bacterial burden, and enhancement of antibiotic delivery. During debridement of necrotic bone, viable tissue is identified by punctate bleeding on the bone and wound edges. Tissue that has no bleeding is not viable and should be removed as a central tenet of surgical management.

Surgical indications for osteomyelitis are dependent on etiology. In acute hematogenous osteomyelitis, surgical debridement is common but not absolutely necessary depending on the infectious lesion and severity of disease. Yet, there is still an important role for the surgeon for diagnostic biopsy of infected bone to identify the causative organism, although in some institutions, this procedure is often performed by interventional radiology. Osteomyelitis typically requires debridement when surrounding soft tissue is involved. Classic examples include adult diabetic foot ulcers, open fractures and trauma, septic joints, and periprosthetic joint infection. For diabetic ulcers, debridement is typically reserved for wounds that have deeply progressed to involve bone. In this scenario, surgical debridement is essential to create a viable wound bed and remove necrotic tissue. In orthopedic trauma, a main principle in the management of open fractures, which have a high risk of developing infection, is urgent irrigation and debridement to prevent infection and promote healing (5, 90–92). Early surgical intervention, antibiotics, and stabilization of the fracture are tenets of what has been termed “damage control orthopedics” in a trauma setting (93, 94). In osteomyelitis associated with implant infections such as periprosthetic joint infection, surgery serves an equally important role, where both surgery and antibiotics are critical to successful treatment of the infection (9, 19, 77, 79, 80, 95).

An additional advantage of surgery is the ability to provide local delivery of antibiotics at high doses. A common technique is to use antibiotic-laden poly(methyl methacrylate) (PMMA), otherwise known as bone cement. Here, antibiotics are added to the PMMA powder prior to the start of polymerization. A typical mix includes 1 g of vancomycin and 1 g of either gentamicin or tobramycin per bag of PMMA. High viscosity Palacos cement should be used, as it has been shown to have superior elution properties. It is important to appreciate that antibiotic elution is only for a limited time period of a few days (96, 97). The advantages include the ability to locally deliver high doses of antibiotic, fill dead space, and maintain soft tissue tension. The disadvantage is that it requires an additional surgery to remove. There are a variety of other new local antibiotic delivery methods being developed that are beyond the scope of this review (98).

BACTERIAL PATHOGENESIS IN THE CONTEXT OF BONE INFECTION

Much of our current understanding of bacterial pathogenesis and host-pathogen interactions during osteomyelitis is derived from animal models, the vast majority of which have used S. aureus as a model pathogen. Most animal models of osteomyelitis involve traumatic inoculation of a bacterial pathogen, with or without implant placement (99). Modeling hematogenous osteomyelitis has proven more difficult, although several groups have reported success using a preceding implant surgery (100–102), and at least one group has had success studying hematogenous S. aureus osteomyelitis without a preceding bone injury (103). Animal studies of osteomyelitis are conducted in both rodents and larger animals, an appropriate strategy given the wealth of genetic tools available for mice, balanced with the comparative difficulty of stabilizing murine fractures.

S. aureus has a wide array of virulence factors that promote invasion of, and survival within, host tissues (59–61, 104). This includes adhesins, cytolytic toxins and exoenzymes, immunoevasion factors, and superantigens. Of particular importance to the initiation of osteomyelitis are the microbial surface components recognizing adhesive matrix molecules (MSCRAMM) group of adhesins. MSCRAMMs are proteins with IgG-like subdomains that are capable of binding components of the extracellular matrix, such as collagen, fibrinogen, bone sialoprotein, and fibronectin (59). The staphylococcal adhesin Cna enables binding to collagen and contributes to the pathogenesis of osteomyelitis (105–107), although strains that lack collagen binding capability in vitro have been isolated from human cases of osteomyelitis (108). Several S. aureus adhesins are capable of binding fibrinogen, including bone sialoprotein binding protein (Bbp), clumping factors A and B (ClfA and ClfB), fibronectin binding proteins A and B (FnBPA and FnBPB), extracellular matrix binding protein (Emp), the extracellular adherence protein (Eap), and the staphylococcal coagulases (Coa and vwBP). The direct role of these proteins in human osteomyelitis has not been fully delineated, although the coagulases and fibrinogen binding are critical for staphylococcal abscess formation and survival in other murine models (81). Interestingly, some adhesins appear to have a more pronounced role in osteomyelitis associated with medical comorbidity. Farnsworth et al. found that S. aureus isolated from obese/type 2 diabetic mice displayed upregulation of four adhesin genes (clfA, clfB, bbp, and sdrC), and inactivation of clfA reduced infection severity in obese but not wild-type mice (109). Finally, the ability to bind bone matrix sialoprotein was found to be enhanced among S. aureus isolates from patients with osteomyelitis compared to that among endocarditis isolates (110, 111). The surface proteins responsible for this binding were later identified (112, 113), and antibodies to one of these proteins, Bbp, might have diagnostic value for osteomyelitis (114).

Adhesion to host tissues is a critical first step in the pathogenesis of osteomyelitis. However, after establishment of infection, S. aureus faces a hostile host environment due to innate immune defenses and nutrient limitation. These host factors are counteracted by staphylococcal toxins and degradative exoenzymes that incapacitate immune cells and liberate nutrients (60). Some of these secreted factors have been linked to the pathogenesis of osteomyelitis. For example, phenol-soluble modulins (PSMs), amphipathic peptides with broad cytolytic and immunomodulatory activity, trigger murine and human osteoblast cell death and contribute to bone destruction during osteomyelitis (115–117). The alpha-type PSMs, in particular, are both necessary and sufficient for the cytolytic activity of S. aureus supernatants toward primary osteoblasts in cell culture (116). Two additional staphylococcal toxins, Panton-Valentine leukocidin (PVL) and alpha-hemolysin (Hla), contribute to osteomyelitis pathogenesis and extraosseous complications in a rabbit model (118). Furthermore, passive immunotherapy with antibodies against both Hla and ClfA improved outcomes in a murine hematogenous implant model (100). Understanding the full contribution of staphylococcal toxins to osteomyelitis pathogenesis will likely require human studies in addition to animal models, however, as some toxins have species specificity (119). To this end, PVL has been associated with enhanced disease in pediatric patients with acute hematogenous osteomyelitis (120). Moreover, the host immune response to S. aureus leukocidins can be used as a culture-independent diagnostic in children with musculoskeletal infection (121, 122).

In addition to cytolytic toxins that induce death of innate immune cells, S. aureus also expresses both surface-bound and secreted immunoevasion factors (61). S. aureus protein A (Spa) promotes immune evasion through the ability to bind Fc regions of antibodies, interfere with complement activation, and serve as a B-cell superantigen (123–125). In addition to these functions, protein A may contribute to the pathogenesis of staphylococcal osteomyelitis by altering both osteoblast and osteoclast biology. Spa binds directly to osteoblasts via tumor necrosis factor receptor-1 (TNFR-1), leading to increased receptor activator of NFκB-ligand (RANK-L) expression and activation of apoptosis (126–128). In differentiating osteoclasts, Spa interacts with TNFR-1 and the epidermal growth factor (EGF) receptor to promote osteoclastogenesis and bone destruction (129). Another immune evasion protein, the major histocompatibility complex (MHC) class II analog protein Map, has also been found to contribute to the pathogenesis of osteomyelitis by interfering with T-cell function and promoting chronic infection (130).

Given the important roles of multiple staphylococcal virulence factors in invasive infection, it is not surprising that there has been a considerable research focus on S. aureus regulatory proteins that control virulence factor production. The accessory gene regulator (agr) locus encodes a peptide quorum-sensing system that controls expression of multiple toxins, proteases, and protein A. Agr was one of the first regulatory loci linked to the pathogenesis of staphylococcal osteomyelitis. Research by Gillaspy et al. in a rabbit model of acute osteomyelitis clearly indicated a role for Agr in disease severity, although inactivation of Agr did not eliminate the ability of staphylococci to colonize bone (131). Subsequent studies in mice revealed that genetic inactivation or pharmacologic inhibition of Agr decreases bone destruction during acute osteomyelitis (116, 132). These observations partly reflect the direct regulation of PSMs by AgrA (116, 133, 134). However, although the Agr system clearly contributes to the pathogenesis of acute osteomyelitis, its expression may not be favorable during chronic infection, as many staphylococcal isolates from human osteomyelitis cases have agr mutations (135, 136). A second S. aureus regulator strongly linked to the pathogenesis of osteomyelitis is the staphylococcal accessory regulator SarA. Inactivation of SarA reduces virulence in murine models of both acute hematogenous and posttraumatic osteomyelitis (115, 137, 138). The major mechanism by which SarA promotes invasive infection is through repression of proteases, and protease null strains of S. aureus are hypervirulent in multiple models of invasive infection (115, 139–141). A similar dysregulation of protease activity partially underlies the significant attenuation of S. aureus sae locus mutants, which have reduced bacterial burdens and cortical bone destruction during experimental osteomyelitis, although direct regulation of staphylococcal virulence factors by Sae is also an important determinant of pathogenesis (116, 137). Finally, the staphylococcal respiratory response (Srr) two-component system was discovered to be essential for S. aureus survival during osteomyelitis using transposon sequencing (Tn-Seq) (133). The srr locus was shown to contribute to osteomyelitis, in part, through coordinating bacterial responses to the hypoxic environment of infected bone. This study also identified a total of 213 genes that contributed to bacterial survival in bone, a large number of which encode hypothetical proteins that will require further research to elucidate their connection to osteomyelitis pathogenesis.

Staphylococcal virulence factors are a major determinant of osteomyelitis pathogenesis, but the host responses to bacteria in bone are equally important and dictate tissue responses that contribute to adverse outcomes. Many of the innate immune responses that are necessary to control bacterial replication during invasive infection have important roles as modulators of bone homeostasis (142, 143). For example, interleukin-1 (IL-1) cytokine signaling is necessary for host defenses against S. aureus in multiple infection models, including osteomyelitis (142, 144, 145). Yet, IL-1 is also an important regulator of osteoclastogenesis, having initially been identified as “osteoclast activating factor” (146). In line with these observations, the IL-1 receptor (IL-1R) was demonstrated to contribute to infection-associated trabecular bone loss and osteoclastogenesis during osteomyelitis (144). Moreover, inhibition of inflammasomes, multiprotein enzymatic complexes that process IL-1R ligands, reduces bone destruction during osteomyelitis (147). Understanding if such immunomodulatory therapies could be deployed in human infection requires further research. Finally, S. aureus can invade and persist within both osteoblasts and osteoclasts, leading to substantial alterations in cell physiology (148–151). To what extent invasion of osteoblasts and osteoclasts by S. aureus contributes to human disease is a topic for future research, as this has been observed in vitro but not in vivo.

CONCLUSIONS AND OPPORTUNITIES FOR FUTURE RESEARCH

Osteomyelitis remains a clinical challenge and is best approached with evidenced-based clinical care guidelines and a multidisciplinary team. Although recent changes in antibiotic susceptibility patterns have dictated empirical antibiotic selection, the treatment of osteomyelitis is largely unchanged over the past decades and centers on source control, tailored antibiotic therapy, and correction of medical comorbidities. Studies using animal models have shed light on host and bacterial factors that contribute to the pathogenesis of osteomyelitis and offer hope for new therapeutic targets, drug delivery strategies, and immunomodulatory therapies to lessen the incidence and morbidity of osteomyelitis.

There are ample opportunities for future research on osteomyelitis. In the clinical arena, continued attention to interdisciplinary clinical care guidelines and severity classification will drive improved care at both the local and national levels. Ongoing changes in antimicrobial susceptibility patterns will require continued vigilance to assist in empirical antibiotic selection. Moreover, standard in vitro antibiotic susceptibility testing is unlikely to model the complexity of bacterial physiology in vivo, where factors such as biofilm, abscess formation, nutrient deprivation, and tissue hypoxia might render bacteria more tolerant of antibiotics (152). Whether or not new local and bone-targeted delivery agents can overcome these factors to enhance treatment of osteomyelitis is an important area for future study (153). Furthermore, compelling data from both animal models and human samples indicate that S. aureus can directly invade the osteocyte lacuno-canicular network, which may have important ramifications for treatment failure (154, 155). This research also speaks to the fact that S. aureus exists within architecturally distinct niches during infection, such as within bone marrow abscesses, sequestra, or the canalicular network or on the surface of intact cortical bone or implants. It is likely that heterogeneity in bacterial physiology exists within these distinct niches; therefore, animal models that homogenize entire infected bones, although necessary, may miss critical facets of host-pathogen interactions during osteomyelitis. Accordingly, new methodology should be developed to study host-pathogen interactions during osteomyelitis in a spatially resolved manner. On the host side, infection induces significant changes in bone remodeling that may contribute to a failure of antibiotic delivery or efficacy. Observations from the emerging scientific discipline of osteoimmunology indicate that many of the proinflammatory responses to bacterial pathogens in bone have important effects on osteoblast and osteoclast biology and thereby contribute to dysregulation of bone homeostasis during osteomyelitis (143). Whether or not these host responses can safely be targeted to mitigate bone destruction without significantly impacting critical innate immune responses remains to be determined. Recently, the International Consensus Meeting on Musculoskeletal Infection was convened, which outlined additional major challenges in the field and future research objectives (156–158). In summary, understanding both the bacterial and host factors that contribute to antibiotic tolerance or failure is therefore a key direction for future research.