Osteomyelitis (OM) is a serious infectious disease in orthopedics, with an inflammatory and destructive process on bone and its surrounding tissues including marrow, cortex, and periosteum (Lew and Waldvogel, 2004). The high recurrence rate, long duration time, and serious bone destruction of OM bring huge physical and economic burdens to these patients (Lew and Waldvogel, 2004; Kalinka et al., 2014). Staphylococcus aureus, which is notorious for its strong adaptive capacity to adverse environments and the ability of elaborating diverse virulence factors (Otto, 2014), is the most common pathogenic bacteria of OM (Lew and Waldvogel, 2004). The diagnosis of S. aureus OM relies on a combination of clinical symptoms, microbiology, histopathology, laboratory data, and imaging examinations (Lew and Waldvogel, 2004). Although biopsy of bone tissue or marrow is not always appropriate or feasible for OM patients, the microbiological cultures of sinus-tract or wound specimens from the OM infection sites of these patients are also useful for pathogenic diagnosis, especially when a S. aureus strain is isolated (Mackowiak et al., 1978). To solve the problem of the difficult-to-treat S. aureus OM, most attention have been paid on therapeutic approaches including inadequate debridement and antimicrobial usage, few focused on bacteriological factors involved.

Several previous studies attempted to seek for virulence factors associated with prevalence of S. aureus OM. The bbp gene (encoding bone sialoprotein binding protein) was considered to be associated with OM (Tung et al., 2000), but different studies presented opposite conclusions (Tristan et al., 2003; Kalinka et al., 2014). One study suggested that the bbp gene was associated more with OM/arthritis than with endocarditis (Tristan et al., 2003), while another found that the bbp gene played no role in OM (Kalinka et al., 2014). The fnbB gene was also once reported to be more frequently carried in patients with S. aureus OM, but the sample size was small in that study (only 21 OM isolates, 10 sepsis isolates and nasal colonization isolates) (Kalinka et al., 2014). Panton-Valentine leucocidin (PVL) is a pore-forming toxin composed of two components, LukS-PV and LukF-PV, which constitute a heptamer, leading to lysis of neutrophil, monocytes, and macrophages (Bakthavatchalam et al., 2017; Takadama et al., 2017). PVL received great attention because it is thought to be an important toxin of S. aureus and an epidemiological biomarker of severe S. aureus infections in several studies (Vandenesch et al., 2003; Kaneko and Kamio, 2004; Sheikh et al., 2015). For example, PVL-positive S. aureus strains are more likely to cause severe necrotizing pneumonias and severe skin and soft tissue infections (Morgan, 2007; Bakthavatchalam et al., 2017). They were reported more frequently isolated from young patients (Muttaiyah et al., 2010). However, the significance of PVL is controversial because other studies challenged this opinion (Shallcross et al., 2013). For example, several clinical trials show that severe SSTIs caused by PVL-positive and PVL-negative strains do not result in different outcomes (Tong et al., 2015). A meta-analysis involved in 76 studies from 31 countries found that pvl genes were comparatively rare in invasive disease with poor prognosis (Shallcross et al., 2013). For OM infections, in a rabbit OM model, PVL was proposed to play a role in the persistence and rapid local extension of OM caused by community-acquired methicillin-resistant S. aureus (CA-MRSA) (Cremieux et al., 2009). In the United States, PVL-positive S. aureus accounted for up to two-thirds of S. aureus-related acute pediatric hematogenous OM (Bocchini et al., 2006; Sdougkos et al., 2007). As PVL-positive United States-300 clone is widely spread in the United States, it is difficult to separate the effects of United States-300 and those truly caused by the pvl gene (Sheikh et al., 2015). This can be avoided in China because of low prevalence of United States-300 (Song et al., 2013).

One of the most formidable challenges of OM therapy is its refractoriness and recurrence during long duration time which can continue for several decades (Lew and Waldvogel, 2004; Rao et al., 2011; Wang et al., 2017), but few studies focused on the microbiological factors affecting OM duration. Kalinka et al. (Kalinka et al., 2014) attempted to find differences on virulence factor profiles between S. aureus OM isolates from 11 patients with less than 2-month duration and 10 patients with more than 12-month duration, but no significant difference was found.

In consideration of controversial viewpoints on microbiological risk factors associated with S. aureus OM and few data on the relationship of molecular epidemiology and clinical manifestations of S. aureus OM, we conducted this study to explore microbiological factors that may affect the prevalence, duration, and severity of S. aureus OM. The results will provide information for risk factors of S. aureus OM, identify potential biomarkers for early diagnosis and warning for long duration and serious S. aureus OM.

Although pathogenic microorganism, especially S. aureus, is the primary cause of OM (Lew and Waldvogel, 2004), the relationship between microbiological molecular epidemiology and clinical characteristics of OM is largely unknown.

It is not well known if any specific molecular types of S. aureus are associated with OM infection. Some studies indicated higher prevalence of MRSA in OM patients (Gadepalli et al., 2009). In this study, multiple molecular typing revealed that neither MRSA/MSSA, nor specific spa types were more likely to cause S. aureus OM. We found that STs in some cluster seem to be specific for OM infection (Figures 1C,D), and agr group III/IV may play as a risk factor of S. aureus OM (p = 0.0040). MLST is principally employed genetic typing method. It well reflects the underlying genealogical information and provides an accurate assessment of species and even strains, and therefore widely used in evolutionary and population studies for pathogens (Maiden, 2006). The clustering of some STs that are related to OM infections suggests that some genetic factors of theses STs make them more susceptible to OM infections. More detailed mechanisms underlying this phenomenon need further in-depth investigation, and would be meaningful for understanding the relationship between bacteria genetic makeup and infectious susceptibility. The significant difference amongst agr groups may be attributed to some virulence factors usually carried by specific agr group as previously reported (Xiong et al., 2009). Actually, in this study, the bbp-carrying isolates all belonged to agr group III/IV, and agr group III/IV isolates were more likely to carry pvl gene (p = 0.00020).

Antimicrobial resistance is becoming a serious public health problem, especially for ESKAPE pathogens (Enterococcus faecium, S. aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacteriaceae) (Rice, 2010; Cinquin et al., 2015). Antimicrobial resistance may cause recurrent infection and prolonged treatment duration (Lew and Waldvogel, 2004; Proctor et al., 2006). However, no specific antimicrobial resistance profile was found to be associated with prevalence or longer duration of S. aureus OM in this study. Several antimicrobials showed relatively high resistance rate in both groups. Amongst them, clindamycin, one of the preferred choices for treatment of S. aureus OM (Feigin et al., 1975; Bamberger and Boyd, 2005), presented a resistance rate of 45% in OM group. This may form a potential challenge to anti-infection therapy of S. aureus OM.

Previous studies had suggested that several virulence factors might contribute to S. aureus OM, but controversies exist. In the current study, we found that the pvl gene was associated with S. aureus OM in both univariate analysis (p = 0.014) and multivariate analysis (p = 0.042). This result is partly consistent with the investigation in S. aureus-related pediatric acute hematogenous OM in the United States (Bocchini et al., 2006; Dohin et al., 2007), and further indicates the significant role of PVL in the prevalence of S. aureus OM. PVL was few related to bone-associated infections previously, although the contribution of PVL to clinical diseases remains controversial by now (Bakthavatchalam et al., 2017), the relationship between PVL and bone-associated infections deserves more attention and needs more basic experimental studies. The S. aureus Bbp protein is predicted to be important in the localization of bacteria to bone tissues, and thus related to the pathogenicity of OM (Tung et al., 2000). In this study, we also found that the bbp gene was associated with S. aureus OM, but only in the univariate analysis (p = 0.010), not in the multivariate analysis (p = 0.374). Our results support the view of Tristan et al.’s (2003) study, but contrary to Kalinka et al.’s (2014) results. More studies combining clinical investigation and basic research are needed to clearly illustrate the role of Bbp in S. aureus OM. The ebps gene was first implied as a risk factor of S. aureus OM (p = 0.019) in this study. However, in consideration that the ebps gene could often be detected in the bbp-positive isolates, especially in some specific STs, such as ST121 and ST30 (Aung et al., 2011; Montanaro et al., 2016), and all the bbp-positive isolates in this study carried the ebps gene. So, the significantly higher carrying rate of the ebps gene in OM group may be related to the carriage of the bbp gene. Although fnbB gene was once indicated to be associated with S. aureus OM (Kalinka et al., 2014), no significantly higher prevalence of fnbB gene or other virulence factors were observed in OM isolates. The sed gene was found significantly more frequent in non-OM group (p = 0.045) in this study. This was partly because more ST59 clone, which was reported tending to carry sed gene (Yu et al., 2015), was involved in non-OM group.

One of the most concerned issues in OM is the long duration time, which makes the cure difficult and brings heavy burn to both patients and the society (Lew and Waldvogel, 2004). It is widely reported that clinical treatment in the course of OM has an important influence on OM duration (Lew and Waldvogel, 2004; Rao et al., 2011). In other infections, some microbiological factors had been related to infection durations (Lindberg et al., 2000; Nowrouzian et al., 2011), while microbiological factors were seldom investigated in OM. A study with only 21 S. aureus OM patients of different durations found no significant difference in virulence factors (Kalinka et al., 2014). The current study suggested that the prevalence of pvl gene was significantly higher in patients with OM duration of more than 24 months (p = 0.0003) and 20 years (p = 0.00030) (Figure 3B). Furthermore, we first proposed that the carriage of bbp (p = 0.016) or sei gene (p = 0.041) might also increase the risk of long duration OM with more than 20 years duration (Figure 3C). The bbp gene had not been related to infection duration previously. Because, Bbp may be related to bacteria colonization (Tung et al., 2000), so it is reasonable to result in longer infection time in OM infections, but to make this conclusion more convincing, more epidemiological studies should be conducted in the future. Several previous studies suggested that egc cluster (including seg, sei, sem, sen, and seo) might be responsible for long-term and persistent infections (Lindberg et al., 2000; Nowrouzian et al., 2011), but we did not find significant differences on other members of egc cluster between OM patients with different durations in this study. These results suggest that pvl, bbp, and sei genes may provide as potential precautionary and diagnostic biomarkers for long-duration S. aureus OM, and therapeutic strategies targeting theses genes may help to shorten the OM infection duration.

Besides the crucial role of PVL in S. aureus OM mentioned above, we found some characteristics of PVL-positive OM cases and their isolates. Consistent with the result of a previous case control study enrolling S. aureus isolates from various infections (Boan et al., 2015), OM patients infected with PVL-positive S. aureus also tended to be younger (p = 0.012) and males (p = 0.01). But intriguingly, these characteristics are OM-specific in the current study (Figures 5A–C). The PVL-positive S. aureus strains were more likely to be agr group III/IV for OM patients, but not for non-OM patients (Figure 5D), this result suggests that OM infection may be more susceptible to agr group III/IV S. aureus carrying the pvl gene. In addition, we found that PVL-positive S. aureus-related OM patients showed more serious inflammatory responses reflected by significantly higher levels of CRP, WBC, and ANC values, and these significant differences were not found in non-OM patients infected with PVL-positive S. aureus (Figures 4A,C,D). Therefore, clinicians should be more vigilant against PVL-related OM when high values of these data were acquired from S. aureus OM patients. However, as the reason of functional redundancy, the lab experiments may not completely mean the actual function in the patient’s body, so further studies are still needed to verify the role of PVL in patients with S. aureus OM (Lebughe et al., 2017).

The present study has several limitations. First, the investigation was done in a single hospital. However, the data were representative because our hospital is one of the biggest hospitals in western China, taking patients from provinces mainly including Chongqing, Sichuan, Guizhou, and Yunnan. A multicentre study should be done in the future if possible. Second, not every patient’s laboratory data in the study could be obtained, but at least the data from 49/60 patients (for ESR data) in the OM group and 61/79 patients (for CRP data) in the non-OM group were collected, so it is able to reflect the tendency.

Our study reveals that STs belonged to certain cluster are specific to S. aureus OM infection. While antimicrobial resistance is not responsible for prevalence or duration of S. aureus OM, some virulence factors play important roles in S. aureus OM. The pvl, bbp, and ebps genes may represent as risk factors for prevalence of S. aureus OM. Presence of pvl, bbp, and sei genes may prolong the duration of S. aureus OM. We also observed that PVL-related S. aureus OM showed more serious inflammatory responses. Our study suggests that PVL may play as a potential indicator for prevalence, duration, and severity of S. aureus OM. Special attention should be paid on PVL-related S. aureus OM in clinical diagnosis and treatment.

XH and ZX conceived and designed this study. BJ and YW carried out the experiments, collected, and analyzed the data. ZF, LX, LT, and SZ helped with the data collection and interpretation. YG, CZ, and XL helped with BioNumerics analysis and construction of phylogenetic tree. SL, XR, and YP drafted the manuscript. All authors have read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.