Staphylococcus aureus Genomic Analysis and Outcomes in Patients with Bone and Joint Infections: A Systematic Review

Many studies have been published assessing the association between the presence of S. aureus genes and outcomes in patients with bone and joint infections (BJI), but it is not known if they have had similar findings. A systematic literature review was performed. All available data on studies in Pubmed between January 2000 to October 2022 reporting the genetic characteristics of S. aureus and the outcomes of BJIs were analyzed. BJI included prosthetic joint infection (PJI), osteomyelitis (OM), diabetic foot infection (DFI), and septic arthritis. Because of the heterogeneity of studies and outcomes, no meta-analysis was performed. With the search strategy, 34 articles were included: 15 articles on children and 19 articles on adults. In children, most BJI studied were OM (n = 13) and septic arthritis (n = 9). Panton Valentine leucocidin (PVL) genes were associated with higher biological inflammatory markers at presentation (n = 4 studies), more febrile days (n = 3), and more complicated/severe infection (n = 4). Other genes were reported anecdotally associated with poor outcomes. In adults, six studies reported outcomes in patients with PJI, 2 with DFI, 3 with OM, and 3 with various BJI. Several genes were associated with a variety of poor outcomes in adults, but studies found contradictory results. Whereas PVL genes were associated with poor outcomes in children, no specific genes were reported similarly in adults. Additional studies with homogenous BJI and larger sample sizes are needed.


Introduction
Staphylococcus aureus (S. aureus) is the most common pathogen in almost all types of bone and joint infections (BJI), mainly due to its ability to adhere and to invade host tissues, evade immune defenses, and form biofilm in cases of prosthetic joint infection (PJI) or foreign-body associated osteomyelitis (OM) [1]. Outcomes of S. aureus BJI are closely related to host criteria (body mass index, tobacco, comorbidities, immunosuppression), medical/surgical management (antimicrobial therapy, removal of foreign bodies, debridement and irrigation), and phenotypic characteristics of the bacteria (resistance to methicillin, rifampicin or fluroquinolones) [1][2][3]. However, the role of the genetic background of the bacteria on the outcomes remains unknown. Researchers have associated both clinical characteristics and outcomes of the S. aureus bloodstream infection with different S. aureus clonal complexes (CCs) or production of toxins (Panton Valentine leucocidin, PVL) [4][5][6]. However, it is not known if, in cases of BJI, there are similar associations. In a recent literature review aiming to address the distribution of S. aureus CCs in outcomes of BJI, all major S. aureus clones appear to be capable of causing BJI, and no specific clone was related to BJI [7]. However, no additional genetic analysis (such as presence of virulence factors) and/or outcome was analyzed in the review. The aim of this systematic review Figure 1 shows the flow chart of the screening process of our literature review. With the search strategy, 503 articles were identified in Pubmed. After abstract screening, 111 (22%) were assessed for eligibility with full text analysis. One article appeared to meet the inclusion criteria, but the number of genes analyzed in each group and the statistical analysis were not reported [8]. Ultimately, 34 articles were included in the review, 15 articles in children, and 19 articles in adult patients (see Supplementary Materials). Among the 34 articles, six case reports were included, all in adult patients.

Studies Included
We included 15 studies that reported BJI in children. Most BJI studied were either acute hematogenous OM (n = 13) or septic arthritis (n = 9). Three studies included only methicillin susceptible S. aureus (MSSA) isolates [9][10][11], and one study only methicillin resistant S. aureus (MRSA) isolates [12]. In other studies, the prevalence of MRSA varied from 2.7% (Morocco) to 88% (USA). Only two studies performed whole genome sequencing (WGS) analysis [12,13]. The 13 other studies used specific PCR assays to detect genes, specifically PVL (13/13), mecA (3/13), the agr system (4/13), or other genes (3/13) (Table 1).  1 Modified severity of illness scoring system >5. 2 Chronic OM (based on imaging studies and/or histopathology) and/or deep venous thrombosis and/or fracture at the site of the infection. 3 Orthopedic complications included chronic OM, pathologic fracture, growth arrest/limb length discrepancy, avascular necrosis, and/or chronic dislocation. 4 Chronic OM was defined as the presence of any of the following after receiving ≥ 4 weeks of effective antimicrobial therapy: (a) A sequestrum or permeative lucency in bone visible on plain radiographs, (b) New or worsening physical exam signs or symptoms referable to the infected bone/joint (e.g., pain, swelling, drainage, etc.), or (c) Readmission for the treatment of OM.
Four studies evaluated the relationship between PVL-positive strains and complications in BJI in long term follow-up. One study with a mean time of follow-up of 25 months found that 12/14 children with PVL-positive strains had complications (including 2 late fractures, 2 leg-length discrepancies, and 4 cases of radiographic bone abnormalities) versus none of the 17 children with PVL-negative strains [18]. Belthur  However, PVL strains were not associated with progression to chronic OM [11] 3.2.3. Sequence Type/Clonal Complex One study found that the USA300-0114 clone was associated with bone fracture [14]. Another study found patients infected with the USA300 clone had a more severe inflammatory syndrome than with other clones [10]. In another study, the USA300 clone and PVL-positive strains were associated with ICU admission in univariate analysis but not in multivariate analysis [9].

Other Genes
The fnbB gene was evaluated in three studies (one using specific PCR, two using WGS), but no significant difference associated with outcomes was reported [15,16,20].
Two studies with PCR detection of clfA, cna, fib, clfB, bbp, eno, and ebpS did not find any differences between outcomes and the presence of these genes in children with BJI [16,20].
In two studies, agr type III was associated with orthopedic complications and chronic OM in multivariate analysis [11,21].
In the two studies with WGS analysis, one found 40 genes associated with increased severity of illness, including the presence of lukF-PV genes [13]. The second study did not show any association between virulence genes and severity [12].

Quality of Studies
Of 15 included studies, four were prospective, but only one was multicenter. The inclusion criteria were homogenous; however, some studies included only AHO, septic arthritis, or both. Sample size was small in most studies, and multivariate analysis was possible in only 3 of them. The outcomes were well defined but differed among studies: two studies used the Modified Severity of Illness scoring system, whereas other studies used a composite score (ICU and/or death) or ICU admission alone. Complications were also defined differently in these studies. Only two of the studies used WGS analysis, but one study had only 12 patients which limited power to compare groups. In the other study, no multivariate analysis was performed, probably due to the high number of variables with significant differences on univariate analysis. No studies reported an analysis for an outcome of relapse or recurrent infection.

Studies Included
We included 14 studies that compared the presence of genes and outcomes in BJI. Six studies reported outcomes in patients with PJI, two in patients with DFI, three with OM, and three with various BJI. MSSA isolates were the only bacteria included in three studies and MRSA in two studies. In other studies, the prevalence of MRSA varied from 1% (Sweden) to 70% (India). Detection of genes was performed by a multiplex PCR in one study, DNA microarrays in three studies, and an analysis of WGS data in three studies. The other seven studies used PCR targeting PVL (n = 3), clonal complex (n = 3), and a number of different genes (PVL, mecA, tsst, eta, etb) (n = 1) ( Table 2).

PVL
Among 14 studies included, 11 reported analyses of PVL genes and their outcomes; three studies included patients with OM, six with PJI, one with DFI, and one with multiple types of BJI. Only one study identified an association between PVL-positive strains and chronic OM with a longer duration of evolution of OM (mean ± SD, 192.1 ± 150 months versus 50.0 ± 150 months [p = 0.00060]) [24].

Sequence Type/Clonal Complex
Two studies identified differences comparing CC398 strains vs. other strains with a clinical outcome, but there were conflicting results in the studies. In one study, patients with S. aureus CC398 DFI were compared with patients infected with other S. aureus clones. Patients with S. aureus CC398 DFI had a more severe infection (International Working Group of the Diabetic Foot-Infectious Diseases Society of America classification grade 4) compared to other clones [25]. In the other study, the authors compared patients with S. aureus CC398 with BJI to patients with BJI caused by other S. aureus clones. Patients with S. aureus CC398 had a lower treatment failure rate than with other clones (0% vs. 37.3% [p = 0.032]) [26]. In a study of patients with acute PJI, the clone ST239 was associated with a higher number of surgical debridements (ST239 n = 10/11 vs. ST8 n = 4/10, p < 0.05), a higher pre-operative CRP level (mean ± SD, 279.16 ± 109.53 mg/L vs. 66.8 ± 57.56 mg/L, p < 0.05) and having fevers more often than other clones [27]. In three other studies, no difference was found comparing CC or ST and their outcomes [28][29][30].

Other Virulence Genes
In 2 studies that evaluated the presence of specific genes and the outcomes of chronic OM, results were discordant. In one study, bbp and sei genes were associated with chronic OM, whereas no tested gene (bbp, cna, fnbB, sdrD, sdrE) was associated with this outcome in the other study [24,31].
In one study, the genes sak, scn, and chp had a higher frequency in patients with the failure of PJI treatment, but this was not statistically significant (sak 80.0% vs. 37 [32]. In one study with WGS analysis, the serine protease gene splE and resistance gene blaZ were significantly less frequent in patients with resolved S. aureus PJI than unresolved infections (splE 40% vs. 70% [p ≤ 0.05]; blaZ 50% vs. 90% [p ≤ 0.01]) [29].
In the second WGS study, cap5H, cap5J, and cap5K genes were associated with treatment failure in patients with PJI, whereas cap8H, cap8I, cap8J, cap8K, and sspP were associated with treatment success, and cap8H and cap8K with the eradication of S. aureus [30].
The third study, using the analysis of WGS data did not show any differences between genome analysis and outcomes in patients with PJI [33].
In one study with PJI, infection by strains that were agr type II showed a trend toward association with treatment failure (46.2% vs. 24.1%, p = 0.099) [28], whereas agr type II was associated with resolution of infection (n = 8/10, 80% vs. n = 2/10, 20% [p = 0.0256]) in another study [29].    such as supplementary debridement >30 days after the first surgery, prosthesis removal, or supplementary course of antibiotics; if prosthesis removal (i.e., chronic infections), failure also included the absence of a new prosthesis replacement. 5 Death from any cause within 90 days after surgery, persistent or relapsing signs of staphylococcal infection, and/or the need for salvage therapy due to S. aureus, including antimicrobial suppressive therapy and unplanned surgeries (except for extra debridements in the first 30 days after the initial therapeutic surgery). 6 No data for two strains. 7 Clinical Success: (a) patients with no clinical or laboratory evidence of infection with all antibiotics discontinued; (b) patients with no clinical or laboratory evidence of infection but continuing oral antibiotics because of a high risk for relapse. Clinical Failure: (a) patients experiencing clinical and laboratory deterioration during antibiotic therapy that required surgical intervention; (b) patients experiencing clinical and laboratory deterioration after discontinuation of antibiotic therapy. 8 Absence of infection with clinical or microbiological criteria and no further treatment (surgery or antibiotics). 9 CRP level >10 mg/L at the time of diagnosis. 10 Treatment failure: (a) persistence of septic symptoms despite appropriate surgical and medical treatment; (b) relapse owing to isolation of the same MSSA strain after cessation of treatment; or (c) the need for a new surgery for sepsis more than 5 days after the initial surgery. 11 Infection lasting < 4 weeks. 12 At least one of the following: (a) need for prosthesis removal (one-or two-stage exchange, resection arthroplasty, or amputation); (b) persistent clinical and laboratory signs of infection; (c) need for suppressive antibiotic treatment of any pathogen including S. aureus; or (d) death during antibiotic treatment when no other evident explanation was apparent. 13 All of the following: (a) no signs of infection and no antibiotic treatment directed at S. aureus (regardless of whether the primary intervention required prosthesis removal, i.e., one-stage, two-stage, or resection arthroplasty); (b) no S. aureus-associated relapse after completed primary intervention; and (c) the patient did not die during PJI treatment. 14 When adjusting for age and sex, the associations for cap5H, 8H, 8I, 8J, 8K, and treatment outcome remained significant. cap8H was still significantly associated with eradication.
In studies of OM, surgical therapy differed among studies depending on the type of BJI included. In three studies with device-related OM [26,34,36], data on rifampicin used was reported in only one [34]. In three OM studies, analysis of acute/chronic OM were reported with different definitions of acute infection (i.e., <1 month, <2 months and <24 months) and chronic infection (i.e., ≥1 month, >12 months, and ≥24 months) [24,31,36].
In studies of DFI, one study analyzed severity of infection [25], and another defined failure as lower limb amputation [37]. However, in this last mentioned study, patients with OM and with non-complicated SSTIs were analyzed together [37]. Results of the specific group of DFI with OM were only reported for the presence or absence of mecA.

Studies Included
Six studies comparing initial and relapse S. aureus strains in BJI after treatment failure were included: 3 case reports (with one patient) [38][39][40], and 3 case series with 3, 4 and 14 patients, respectively [33,41,42] (Table 3). Three studies analyzed only patients with PJI [33,40,41] and one study included OM with foreign bodies in one patient and PJI in two patients [42]. Two other studies included OM and native septic arthritis [38,39]. First and persistent strains were analyzed in two studies [38,39], relapse in two studies [40,42], and recurrent and relapse strains in two studies [33,41]. Two studies analyzed specifically small colony variant (SCV) strains [38,40]. All six studies used WGS to compare strains.

Analysis of Genetic Differences between First and Recurrent/Relapse Isolates
The first study assessed the mutations responsible for genetic characteristics of a vancomycin-intermediate S. aureus (VISA) SCV in a patient with septic arthritis during long term treatment. Between the first and last VISA SCV isolates, 13 genetic differences in SCCmec, mprF, cls2, clpX, and fabF were found. The mutation of fabF (encodes a fatty acid synthase) was probably responsible for the SCV phenotype of recurrent strains [38].
In the second study, the authors characterized two different ST398 MSSA isolates causing septic knee arthritis and lack of response to antimicrobial therapy. No difference in virulence or resistance genes was found between the first and recurrent strains. In core genome SNPome analysis, only one SNP differed between the two strains [39].
The third study assessed genetic differences between a first MSSA isolate and an SCV isolate in a case of relapsing PJI. The loss of a 42.5 kb prophage in the genome of the relapse-SCV strain (ϕSa2) and three deletions, including a non-truncating deletion within the rpoB gene, were associated with the SCV isolate [40]. A striking difference between the index and relapse-SCV isolates was the loss of a ∼42.5 kb prophage in the genome of the stable relapse-SCV strain (ϕSa2) Only 5 SNPs identified, of which 2 were intergenic and 3 were intragenic, with only 1 introducing a nonsynonymous mutation Small INDELs (1 bp) inducing truncated proteins, due to a frameshift change in sasA (encodes SasA protein, associated with binding of platelets) and in glpD (encodes aerobic glycerol-3-phosphate dehydrogenase protein) Deletions leading to the emergence of truncated protein forms for an alpha-beta hydrolase (9 bp INDEL) and a putative serine protease (6 bp INDEL) A 9 bp (3 aa, LysGlyPro488−490) non-truncating deletion within the rpoB gene The fourth study evaluated mutations differentiating both the first and a relapse S. aureus strain in one patient with PJI and both the first and a recurrent strain in two patients with PJI. Mutations in genes coding for proteins involved in fibronectin binding (ebh, fnbA, clfA, and clfB), which distinguished later PJI isolates from the first PJI isolate were found in all three patients. Additionally, in the patient with the relapse strain, mutations in lysostaphin, multidrug resistance, pheromone binding, and epimerase were associated with the relapse strain [41].
The fifth study assessed genetic differences between the first and persisting or relapsing MSSA strains in one patient with foreign body OM and two patients with PJI. Authors reported six, eight, and nine mutations between different paired strains but with no effect on the major regulatory systems that are known to control the expression of virulence in S. aureus [42].
In the sixth study, the authors evaluated genetic differences between first and relapse isolates in three patients with PJI and first and recurrent isolates in other three patients with PJI. In five of the six patients, specific mutations were identified: in bbp, sdrD, clfA, fnbA, and fnbB for two patients, in fmt for one patient, and a deletion in DNA-3-methyladenine glycosylase in one patient [33].

Quality of Studies
The studies were heterogeneous, and, thus, no meta-analysis was possible. In one study, clinical metadata for compared isolates were not clearly reported between relapse or persistent strains. Moreover, antimicrobial and surgical therapy were not reported [41].

Discussion
This review examines published studies assessing the presence or absence of S. aureus genomic content and clinical outcomes in patients with BJI. Whereas the carriage of PVL genes was correlated with increased disease severity in children with BJI, adult studies were usually limited by small sample size.
3.5.1. Role of S. aureus Lineage Some S. aureus lineages have been associated with poor outcomes in patients with bacteremia [6]. The best example is the USA300 lineage, belonging to ST8/CC8-MRSA-IV, which rapidly emerged in the late 1990s to become the dominant community-acquired MRSA strain in North America by 2004 [43]. Selection and diffusion of this clone is likely related to the acquisition of various mobile genetic elements, specifically SCCmec type IVa carrying the mecA gene, phage φSa2 carrying the PVL genes, the arginine catabolic mobile element (ACME) type I, and the acquisition of mutations causing resistance to fluoroquinolone [43,44]. USA300 has been reported to be independently associated with metastatic complications in patients with bacteremia but not in solid organ transplant recipients [5,45]. However, USA300 in these studies was not defined by the distinctive PFGE pattern but by the following criteria: S. aureus belonging to spa-CC008, SCCmec type IVa, and the presence of both PVL genes and ACME. One study analyzed the association of USA300 strains and outcome in 50 adult patients with OM. USA300 lineage was defined by PFGE and only 44% had PVL genes. Neither PVL genes nor USA300 strain type were associated with treatment failure [35]. In children, in cases of OM, the specific clone USA300-0114 was associated with bone fractures and the USA300 lineage was associated with both intensive care unit hospitalization and a more severe biological inflammatory syndrome [9,10,14].
In Europe, MSSA CC398 was associated with mortality in patients with bacteremia in a single center study [4]. In two studies, while CC398 was associated with severe infection in DFI, failure of treatment seems to be less common in patients with BJI [25,26]. However, the lack of CC typing of other S. aureus in the first study [25] and the small number of patients and different BJI included in the second study [26] make the results difficult to interpret. ST239-MRSA is the dominant MRSA strain in some countries and has been reported to be multidrug resistant and able to produce several toxins, mainly, staphylococcal enterotoxins (SEA, SEK, and SEQ), exfoliative toxins A and B (ETA and ETB), and PVL [46]. In a study in Taiwan from 2016 to 2019, ST239-MRSA has been reported to be associated with elevated inflammatory serum markers and also with stronger biofilm formation than other types (mainly ST8 and ST59) of MRSA in PJI, suggesting that this strain type may be more virulent than others [27].

Role of the agr System
The accessory gene regulator (agr) locus of S. aureus is a quorum-sensing virulence regulator that coordinates the expression of various virulence factors [47]. agr system activation leads to decreased production of cell-wall-associated factors, causing the dispersion of biofilm, the spread of an infection, and a simultaneous increase in exoprotein gene expression [48]. Loss of agr activity can result in abundant biofilm formation and strains deficient in autolysis, which can contribute to the persistence of infection and poor outcomes [36,49]. agr dysfunction is detected phenotypically by an absence of δ-hemolysin activity or absence of δ-toxin production and not by genetic background [50,51]. The effect of agr dysfunction in patients with BJI was also not reported in this review. However, agr type III was associated with orthopedic complications and progression to chronic OM in children, whereas agr type II was associated with resolved infection in adults with PJI [21,29]. This finding could be an effect of clonal origin or genetic differences in toxin prevalence, but no effect of PVL gene carriage or specific clonal complex were associated with outcomes in these studies. Notably in children, agr type III was associated in multivariate analysis with orthopedic complications, whereas USA300 strains and strains carrying PVL genes were not statistically different from other strain types in clinical outcomes [21].

Role of PVL Genes
PVL is a toxin composed of two components, LukS-PV and LukF-PV, that, after assembling, forms pores in leukocyte membranes and leads to neutrophil lysis [52]. S. aureus carrying PVL genes are more frequent in skin and soft tissue infections than in other more invasive infections [53]. Moreover, PVL is associated with increased infection severity in children and adults with pneumonia and skin and soft tissue infections [54][55][56]. However, the role of PVL in other invasive infections is not clear. In patients with bacteremia, results of studies are contradictory. Studies suggested an association between PVL and prolonged duration of bacteremia, prolonged duration of fever, and higher risk of development of sites of metastatic infection but not with sepsis/septic shock or mortality [57][58][59][60]. In a recent multicenter study in Australia and New Zealand in children with S. aureus bacteremia, PVL was the only virulence factor associated with poor outcomes with a composite score comprising 90-day all-cause mortality,  [61]. Given the association of PVL with some severe infections, it would seem possible that the presence of the toxin may impact the clinical outcomes of BJI.
Our review suggests that children with PVL-positive S. aureus OM had higher serum inflammatory markers at presentation, a longer course of illness, more days of fever, and more complicated infection (muscle abscess, pyomyositis, subperiosteal abscess, visceral abscess, and deep venous thrombosis). Moreover, blood cultures were more likely to be positive, and children were more likely to require ICU admission [15]. Such a clinical picture should prompt the clinician to consider the possibility of infection by a PVL-positive S. aureus strain and to be aware of associated complications such as deep venous thrombosis and multifocal infection.
However, in adults, PVL-positive S. aureus was not associated with severity or poor outcomes in BJI. Only one study found an association with risk of chronic OM [24]. This difference may be explained by the fact that in children BJI is often caused by hematoge-nous spread of bacteria (hematogenous OM), whereas BJI in adult studies are usually more heterogenous. Indeed, the prevalence of PVL is more common in S. aureus invasive disease such as severe sepsis and in community-acquired infection (especially for MRSA strains) [43]. Another explanation could be the higher prevalence of PVL-positive S. aureus in children with OM compared to adults (median 40% [4.8-100] vs. 7% [0-57]). Differences in the prevalence of PVL-positive S. aureus have already been described in a French multicenter study in patients with S. aureus pneumonia hospitalized in ICU between 2011 and 2016. In this study, 95% (n = 19/20) of patients less than 3 years of age, and 86% (n = 6/7) between 3 and 18 years of age had PVL-positive S. aureus, compared with only 44% (n = 60/136) in older patients [56].

Role of Other Virulence Genes
S. aureus displays a vast array of virulence factors, many of which exhibit broad functionality that allow bacteria to adapt to different hosts and environments. S. aureus expresses many proteins on its surface. The most well-studied are cell wall proteins belonging to the family of microbial surface components recognizing adhesive matrix molecules (MSCRAMM), which play a key role in S. aureus BJI by initiating staphylococcal attachment. MSCRAMM includes clumping factor proteins (ClfA and ClfB) and surfaceanchored proteins (SdrC, SdrD, SdrE) [62].
S. aureus biofilms are complex structures produced by bacteria that attach to surfaces and contain communities of S. aureus. Biofilms harbor an extensive exoproteome, including proteins with functions related to immune evasion and pathogenesis such as hemolysins, nucleases, lipases, proteases, and collagenases. Among these factors, capsular polysaccharides of serotypes 5 (cap5) and 8 (cap8) are recognized as important virulence factors [6,7]. Despite the main role of these proteins in biofilm formation, few studies in our review reported a higher prevalence of genes encoding these proteins in patients with BJI having poor outcomes. One study reported more than 40 genes significantly associated with severity of illness in children with OM, including enterotoxin genes, capsule gene (capE), PVL genes, and MSCRAMMs such as cna and bbp [13]. One study in adults with PJI found that cap5 was associated with treatment failure and cap8 with treatment success and the eradication of S. aureus. Some studies have shown that the loss of capsular polysaccharide production was associated with the persistence of S. aureus in humans with chronic OM, whether the cap5 or cap8 genes were initially present [63,64]. Additionally, it is important to note that USA300, often thought to be a virulent strain in BJI, does not form a capsule, suggesting that capsule production is not necessary for virulence [65]. These findings highlight the difficulty of analyzing only the presence of genes without correlation using phenotyping studies.

Strengths and Weaknesses of the Literature and Perspectives for the Future
The current review provides the most comprehensive overview to date of the role of S. aureus genotyping associated with the outcomes of S. aureus BJI. Studies included in this review were identified from a search of the Pubmed database using a systematic search strategy. However, despite having applied stringent inclusion criteria, it is possible that some relevant articles were excluded. It is important to note that treatment as a covariate affecting reported outcomes was not typically reported in this review, mainly due to lack of data. Indeed, treatment could significantly impact the outcomes of BJI regardless of the genetic background of S. aureus.
Current literature has certain weaknesses, which could be improved on in future studies. First, few studies recorded and analyzed important independent factors associated with outcomes. Indeed, several factors play an important role in the failure of treatment in BJI and, specifically, in PJI or in device-associated OM. Patients' comorbidities, antimicrobial therapy, and surgical treatment are essential to the success of therapy. Future studies should take into account these clinical factors to assess outcomes in patients with BJI, especially in PJI and OM with a foreign body.
Second, the definition of the outcomes varied by study and types of BJI examined, making it difficult to compare results across studies. A homogenous cohort of patients with BJI (e.g., hematogenous PJI or acute PJI, OM with or without device infection) and validated criteria of outcomes, with a minimum duration of follow-up of 1 year, would provide results that could be compared across studies with a clinically relevant outcome [66].
Third, the phenotypic characteristics of S. aureus strains, including biofilm and SCV formation are important factors in the poor outcomes of BJI. Studies in this review included a wide range of S. aureus genes and virulence factors, but it is important to note that the presence of a given gene does not necessarily imply a specific protein product or cell function. Moreover, the large number of genes analyzed in studies using WGS raises the possibility that some of the associations observed resulted from chance alone. Studies assessing the phenotype and genotype of S. aureus are scarce and, in our review, we included mostly studies with a small number of participants, which limited the generalizability of the results. Future studies should include more patients and have a multicenter design to produce reliable results that are generalizable to other populations. Moreover, phenotypic studies are necessary to better understand the role of these genes in the outcomes of BJI. For example, the ability of strains to produce biofilm and its relation to the outcomes of BJI should be addressed along with genetic analyses.

Conclusions
Whereas PVL genes were associated with poor outcomes in children, no specific genes were reported similarly in adults. Further phenotypic and transcriptomic studies are needed to achieve a better understanding of the influence of virulence factors of S. aureus on the evolution and outcomes of BJI.

Conflicts of Interest:
The authors declare no conflict of interest.