Prosthetic joints are a major advance in the practice of modern medicine and have revolutionised the life of many patients. At least 50,000 total hip replacements are performed each year in the UK [1]. The incidence of hip replacements worldwide is expected to increase from 1.66 million in 1990 to 6.26 million in 2050 and, in the European Union, an increase from 414,000 to 972,000 cases per annum is expected over the next 50 years [2]. Unfortunately the risk of infection is a significant problem, resulting in high rates of morbidity and creating a massive economic burden.

Prosthetic joint infections of total hip arthroplasties (THAs) reportedly occur with an incidence of 1.5% for the primary THA and 3.2% for the revision THA [3]. However, one group demonstrated that up to 15% of hip replacements in their study were infected [4]. Several of these infections are characterised by biofilms, adherent communities of bacteria attached to the prosthetic hip joint components that are resistant to antibiotic challenge and host immunity [5].

One of the major problems in accurately determining the infection rate is the difficulty in isolating, by traditional culture methods, the bacteria from the surface of the prosthetic hip joint [6]. The reasons for this include strongly adherent bacteria in the biofilm and the presence of antibiotic-containing cement. One method developed to improve the microbial yield was to place the hip prosthesis directly into an anaerobic jar after surgical removal, followed by mild ultrasonication of the prosthesis to remove adherent microbes and processing of the specimens within an anaerobic cabinet [7]. Another reason for the low yield of microbial growth may be that the joint is infected with highly fastidious and non-cultivable, or viable but non-cultivable, bacteria that cannot be isolated with standard techniques. This problem can be overcome by the use of molecular techniques to detect the microbial DNA from bacteria present on the prosthesis. All culture-independent, molecular-based studies that have investigated the microflora in a wide range of environmental and clinical samples have identified a greater diversity of bacteria than culture methods alone [8, 9]. More specifically, this was found to be true in one study that identified bacteria associated with failed prosthetic hip joints [10]. Using PCR, these workers detected bacteria in 72% of the prosthetic hip joints removed during revision arthroplasty, whereas there was only a 22% detection rate by conventional culture. In addition, these workers were able to reveal bacteria directly by immunofluorescence confocal microscopy of sonicates from previously uncultured specimens. Overall, this indicated that the incidence of prosthetic hip joint infection is grossly underestimated by conventional culture methods.

The purpose of this study was to identify bacteria within the biofilms on the surface of clinically infected and non-infected prosthetic hip joints by using both 16S rRNA-based molecular detection methods and conventional microbiological culture. Ten prosthetic hip joints were analysed for the presence of bacteria by PCR amplification, cloning, and sequence analysis of bacterial 16S rRNA genes. The results obtained were compared with data obtained from aerobic and anaerobic microbiological culture of the same samples. The clinical interest of this study is the presence of any organism on the prosthetic hip joints and the role, if any, that they have in initiating, prolonging or activating simultaneous or subsequent clinical infections.

The risk of infection after hip replacement surgery remains unacceptably high. A greater understanding of which microorganisms may be involved in the infective process will be necessary for an improvement in infection rates and subsequently an improvement in treatment methods. In addition to the uncertainty over the true prevalence of prosthetic hip joint infection, in many cases there is also debate over the source of the infection. The skin microbiota of hospital staff or patients has been assumed to be a likely reservoir of infection. For some patients it has been demonstrated that the oral cavity is the source of prosthetic joint infection [14–17]. However, there is continuing debate over the need for antibiotic prophylaxis when patients with joint prostheses undergo dental treatment procedures that stimulate a bacteraemia [18].

Previous studies have shown that PCR amplification of the 16S rRNA gene, a highly conserved region within the bacterial genome, is invaluable in the detection of the bacterial types involved in prosthetic hip joint infections [10, 19, 20]. However, it has been claimed that PCR assays cannot be used to identify each pathogen in cases of mixed infection [21] and havepoor positive predictive value for hip joint infection [22]. However, we have shown in the present study that gene amplification and sequencing of 16S rRNA is useful in identifying single bacterial species isolated by standard culture techniques as well as in defining the mixed bacterial flora found on the surface of the prosthetic hip joints by using a direct PCR and sequencing approach. It is important to note that the necessary precautions were taken to avoid contamination in the clinical and laboratory settings. The prosthetic hip samples were collected by a surgical team wearing body exhaust suits in an operating theatre with a clean-air enclosure and were packaged in sterile bags. In addition, PCR was performed under stringent conditions and with the use of appropriate controls to prevent false-positive results. Processing of the prosthetic hip joints and subsequent DNA extractions were conducted in a separate laboratory from the PCR assays. All of the reagents for PCR were also stored separately from the positive DNA samples, with the reagents being aliquoted before use to avoid contamination. Finally, a negative control was included with each PCR assay to rule out possible contamination by bacterial DNA. A sterile hip, autoclaved and processed in an identical manner to the 10 test hip joint components, yielded a negative PCR result.

Clinical diagnosis of the 10 cases studied identified five hip replacement devices removed because of bacterial infection and the other five because of aseptic loosening. Comparison of the bacterial species identified in both clinical situations showed a large number of species that may be involved in infection. The microflora associated with each prosthetic hip joint studied was very similar, irrespective of the clinical reason for prosthesis removal. No specific bacterial species that can be associated with clinical infection or aseptic loosening were found. The bacterial species identified take the form of a biofilm attached to the surface of the removed prosthesis in both infected and non-infected cases. It may be that one organism alone or several bacterial species have a role in initiating, prolonging or activating simultaneous or subsequent joint infections. They may also have a role in rendering the joint more susceptible to clinical infections.

The predominant bacteria identified by culture-independent methods from the surface of all the prosthetic hip joints (both infected and non-infected cases) were Lysobacter enzymogenes and Lysobacter sp. C3. Other members of the Lysobacter clade [23] identified were Lysobacter sp. IB-9374, iron-oxidising lithotroph ES-1 and hydrothermal vent eubacterium, in addition to the closely related species Stenotrophomonas maltophilia. These species, which have not previously been reported to be involved in prosthetic hip infection, were not identified by standard culture techniques. The role of Lysobacter-type species in prosthetic hip joint infections is unknown and further research will be required to study the virulence factors involved in infection and the effects on the human immune system. However, Lysobacter-type species have been shown to be important pathogens in hospital-acquired infections [24]. In fact, it has recently been demonstrated that various Lysobacter-type species have the ability to form biofilms readily on various substrates. These include Stenotrophomonas maltophilia, Xylella fastidiosa and Xanthomonas axonopodis [25–27]. It is therefore perhaps unsurprising that these species were identified on the prostheses of the patients in our study.

S. maltophilia was first reported as an environmental species but is now known to be an emerging hospital-acquired pathogen that, among others, has been isolated from gentamicin-loaded polymethylmethacrylate beads in orthopaedic revision surgery [28]. A previous report of a strain of this organism, which was positively charged, demonstrated favourable adhesion kinetics to surfaces such as glass and Teflon [29], and S. maltophilia is now known to adhere avidly to medical implants and catheters to form a biofilm [30]. Furthermore, the organism has been implicated in a range of human infections, including septic arthritis in a patient with AIDS [31].

Sullivan and colleagues [23] described the evolutionary relationship between members of the Lysobacter clade. Lysobacter sp. strain C3 was initially identified as Stenotrophomonas maltophilia [32]. S. maltophilia has been identified by 16S rRNA gene sequencing as the predominant species in advanced noma lesions [33] and has been isolated from a case of acute necrotising gingivitis in an immunocompromised individual [34]. Whether Stenotrophomonas/Lysobacter species are natural members of the oral flora or are merely transient would require further study. However, a high oral carriage of S. maltophilia in a Tibetan population has been reported [35].

From the cultured bacterial isolates sequenced, nine different species were identified that are thought to be involved in prosthetic hip joint infections; Staphylococcus species have previously been described in this context [7, 20, 22], but the other bacteria identified are not commonly associated with prosthetic hip joint infections. Leifsonia species are known to favour moist environments, and in association with other bacterial species they cause infections of the central venous catheter used as vascular access for haemodialysis [36]. Proteus mirabilis has been described in joint infections [20] but is more commonly associated with urinary tract infections [37]. Brevundimonas species are rarely isolated from clinical samples; the role of this species in human disease needs further research, but it has been associated with two cases of bloodstream infections [38]. Zimmermannella alba has been isolated from human blood [39] but has not been reported to be involved in prosthetic hip joint infections. Methylobacterium radiotolerans [40] and Salibacillus species (GenEMBL accession number AY121439) are environmental bacteria more commonly found in plants and salt water lakes, respectively.

A further 21 species of bacteria were identified by culture-independent methods. As stated previously, Staphylococcus and Proteus [7, 20, 22] species are associated with prosthetic hip joint infections and were isolated by microbiological culture and culture-independent methods in the current study. Several other species that differ from those identified by microbiological culture were identified by culture-independent methods. Bacteroides fragilis has previously been associated with hip joint infections and has been isolated in cases of septic arthritis [41]. Many of the other species identified have been more commonly isolated from plants and soil; these include gamma proteobacterium [23], Methylobacterium [42], Bradyrhizobium [43], Acidobacteria [44] and Xyella [45]. For example, Xyella fastidiosa is a phytopathogenic bacterium responsible for diseases in many economically important crops [45]. The uncultivable species identified in the present study are environmental bacteria [46, 47]. Although these bacteria could not be cultured by standard microbiological techniques in the present study, this might have been due to the fastidious growth requirements of these organisms, or to the fact that bacteria growing within a surface-associated biofilm displayed viable but non-culturable tendencies. A recent review has highlighted the plethora of clinical and environmental bacteria that have this characteristic, but whether they are capable of pathogenic traits has yet to be determined [48]. In addition, the use of prophylactic antibiotics during the surgical procedure would hinder the ability to culture and isolate bacteria. However, in our current study none of the patients received antibiotic prophylaxis before surgery.

It was interesting to note that the bacterial species found in the preoperative/perioperative samples by culture (coagulase-negative Staphylococcus, patients 3 and 6; P. mirabilis, patient 7) were also found on the surface of the corresponding prosthetic hip joints by both culture and culture-independent methods. This is suggestive of potential involvement of these species in the infective process. Bacteria were cultured from preoperative/perioperative samples in only 3 out of 10 cases analysed, whereas all 10 corresponding prosthetic hip joints were positive for the presence of bacteria by both culture and culture-independent methods. Furthermore, bacteria were isolated from preoperative/perioperative samples in only three of the five cases classified as being clinically infected. This clearly suggests that standard methods for determining infection in these cases are unreliable.

Several unusual bacterial species have been isolated during this study that have not previously been described as human pathogens and have not been implicated in human infections of prosthetic hip joints. Most of the unusual species identified are environmental bacteria isolated more commonly from plants. Further research is required into the pathogenicity of these bacterial species. It may be that they show pathogenic potential only when they are part of a biofilm in association with other bacterial species. This is indeed the case with Leifsonia species, which are reported to cause infections of the central venous catheter when they are in a biofilm with two other unusual bacterial species [36]. Furthermore, Duan and colleagues [49] demonstrated that key virulence factors from the biofilm-forming organism Pseudomonas aeruginosa were upregulated in the presence of oropharyngeal commensal flora. These studies indicate the key role of polymicrobial biofilms in clinical biofilm diseases, and how cell-cell interactions from non-pathogenic organisms may promote the progression of disease.

There were differences in the bacterial species identified by the microbiological culture and culture-independent methods used in the current study. The species identified that were common to both methods were from the genera Staphylococcus and Proteus, which have previously been associated with hip joint infections. One possible reason for this is the culture techniques used: in this study we used standard culture media and incubation conditions, which broadly enabled us to maximise the culture of bacteria. However, this approach may not have been specific enough for other fastidious organisms. Because the species present on the surface of the prosthetic hip joints are unknown, it is not possible to use specialised media and conditions at this stage, primarily because of cost implications and the time required to process specimens on a vast array of media. Now that it is known that Lysobacter-type species are predominant in prosthetic hip joint infections it will be possible to use specialised media to culture them from the hip sonicate. This exemplifies the validity for culture-independent methods to be conducted in parallel with the culture techniques, so as to identify the entire array of infecting bacteria in each clinical sample.

Another reason for the differences observed may be that primer bias occurs during the PCR procedure. Primer bias results in the unequal amplification of PCR products, resulting in distortion in the product numbers for each bacterial type. PCR primer bias is thought to be caused by inhibition of amplification by self-annealing of the most abundant templates in the late stages of amplification [50] or as a result of differences in the amplification efficiency of templates [51]. We have shown that neither method on its own can isolate all bacteria involved in prosthetic hip joint infections. The vast majority of the bacteria identified that had previously been characterised were Gram-negative species, with the only Gram-positive species being Staphylococcus epidermidis, Staphylococcus pasteuri, Leifsonia aquatica, Leifsonia shinshuensis, Salibacillus sp. and Zimmermannella alba. Some studies, which used 16S rRNA gene sequencing to identify bacteria in a relatively small number of clinical specimens, adopted the approach of sequencing about 50 clones from each library generated per sample [33, 52]. Because of the relatively large number of samples analysed in our study we sought to minimise the sequencing of identical clones by screening with RFLP analysis, and sequencing a single representative clone from each RFLP group. This approach has been used successfully in many studies to avoid sequencing redundancy and to estimate bacterial diversity within clinical specimens [53–55].

From our findings it can be seen that a wide range of bacteria are potentially associated with prosthetic hip joint infections. Further research is required to identify other bacteria involved in infections, because other species have been reported in the literature that were not identified in this study. The knowledge of the bacteria involved in infection can further our research into biofilm formation, into the signalling patterns between the bacteria within the biofilm and into the effects on the human immune system of these infecting pathogens that lead to prosthetic hip joint infections. Although the immediate clinical significance of the present study is somewhat limited, it nonetheless represents a useful preliminary investigation into the microbiology of the aseptic loosening and infection of prosthetic hip joints. Ideally, the study should be expanded to include a larger number of specimens to determine whether true differences exist with regard to the microflora associated with both groups. Because the present study detected only microbial DNA, it is unknown which of the bacteria identified were viable. This could be overcome in future studies by the detection of bacterial mRNA rather than DNA, which would identify only the transcriptionally active (viable) bacteria present on infected and uninfected prosthetic hip joints. This would give a greater insight into which species may be of clinical significance. Ultimately, such data could inform both antibiotic usage in prosthetic joint surgery (both prophylactic and therapeutic) and the relative merits of one-stage and two-stage revisions. However, it should be noted that the antibiotic susceptibility profiles of the bacterial species identified by culture-independent techniques cannot be determined, and this may hinder the development of improved antimicrobial therapy regimes. This study may also have a potential impact on improving the laboratory diagnosis of prosthetic hip joint infections.

A wide range of bacteria can be found on the surface of prosthetic hip joints removed at revision arthroplasty. No significant differences were observed in the microflora associated with infected and non-infected cases. However, the predominant species were members of the Lysobacter genus.