Background

The current ‘gold standard’ for detection of bacteria associated with OM/SA is bacterial culture, although in the UK the pathogen detection rate from culture in children’s bone and joint infections is low.¹ This means that the true spectrum of organisms causing bone and joint infections in the UK is not known, and treatment is usually empirical.

New molecular techniques can enhance diagnostic detection rates. The use of 16S real-time PCR has become available in some specialist centres; however, in the UK it is available at one centre with a turnaround time of days to weeks. 16S real-time PCR is expensive and, although capable of detecting many pathogenic and non-pathogenic organisms, it has low sensitivity because the load of bacteria in the primary sample needs to be sufficiently high to allow detection.

Species-specific reverse transcription polymerase chain reaction (RT-PCR) to identify bacteria associated with OM/SA has been used in specialist centres to enhance the detection of certain organisms, in particular K. kingae, which is difficult to grow in routine culture. However, although it is much cheaper than 16S PCR, species-specific PCR is not routinely available in most UK or European Union laboratories.

Aims and objectives

This study evaluated the potential for the species-specific RT-PCR to be used both to establish the baseline UK microbiology of paediatric bone and joint infections more rigorously than would have been possible by culture alone and to assess the use of the test in a larger RCT.

We also aimed to understand the pathophysiology of K. kingae by using PCR to test throat swabs (taken from patients who have provided consent) for K. kingae to determine how many had concurrent pharyngeal colonisation with this organism.

Methods

The culture and preparation of control organisms

Sample handling and preparation

Samples from routinely collected blood and tissue cultures were taken in the local microbiology laboratory and processed in accordance with the study standard operating procedure (see Report supplementary material 4) before storage and transfer to the central laboratory (Southampton Public Health England South East Regional Laboratory) for PCR analysis.

All samples were taken for PCR after initial bacterial culture had been carried out. For blood cultures, an aliquot of sample was removed after at least 19 hours’ incubation (unless the sample had already been identified as positive) using a sterile needle and syringe, from which 1 ml was placed into a sterile nuclease-free skirted 2-ml tube before being capped and labelled and placed in storage at –80 °C until extraction and PCR was to be carried out. For liquid samples such as synovial fluid, pus and blood or joint washouts, 500 µl was taken from the original sample (where limited sample volume remained, 200 µl was used, thus allowing for repeat culture or further analysis using another method). Solid samples were divided and, where sample quantity allowed, a minimum of two aliquots of tissue or bone weighing around 10 mg each were taken and placed in to a sterile nuclease-free skirted 2-ml tube and stored at –80 °C. All samples were handled in a class 1 cabinet to prevent contamination.

For throat swabs, after swabbing, the cotton end was snipped off using sterile scissors into a flat-bottom 2-ml tube containing 1 ml of skimmed milk, tryptone, glucose and glycerine media, used to preserve bacteria while being frozen at –80 °C.

Polymerase chain reaction analysis

Control bacteria used in this study were sourced from Pro-Lab Diagnostics (Birkenhead, UK) and the UK National Culture Collection (Salisbury, UK). Bacterial deoxyribonucleic acid was extracted by standard methods and multiplex PCR used for analysis. Table 57 shows the multiplex panels used for all samples.

TABLE 57

Elements to each of the multiplexes

Table 57 shows the constituent elements to each of the multiplexes. Two separate targets were used to detect S. pneumoniae as the pneumolysin (ply) gene is also common to other closely related streptococcal species including Streptococcus mitis, Streptococcus oralis and Streptococcus pseudopneumoniae. The N-acetylmuramoyl-L-alanine amidase (lytA) gene was later chosen as it occurs only in S. pneumoniae.

Results

Although seven sites took part in the study, samples from only six of the sites were processed because of incorrect sampling at one site, an error subsequently acknowledged by the local principal investigator. The total number of samples and patients consented to the study from each site is shown in Table 58.

TABLE 58

Total number and patients consented to the study according to geographical site

Table 58 shows the number of patients consented to the study at each site and the number of samples (both throat swabs and native bone and joint samples). Site number refers to tables below.

A total of 151 samples were received from 50 patients. Of these, 113 were bone, fluid and blood culture samples, which were collected from 44 patients across six geographical sites. The remaining 38 samples were from throat and wound swabs. For five participants only throat swab samples were obtained, and for one participant only skin swab samples were obtained.

Of the 113 native tissue samples collected, 39 were blood cultures, in 16 of which pathogenic bacteria were identified using the RT-PCR panel. Sixty of the total native samples collected were joint fluids including pus, of which 46 were positive using the RT-PCR panel. Eleven tissue samples and three bone samples were collected, of which 10 and 1, respectively, were positive using the RT-PCR panel.

Of the 16 blood culture samples that were found to be positive, nine were identified as S. aureus whereas no K. kingae was detected in this sample type. Other notable results identified in blood cultures were three positives for Staphylococcus epidermidis, two positive for GAS, one positive for H. influenzae and one for S. pneumoniae (ply).

Of the 46 positive joint fluid samples that included aspirates and pus, 19 were identified as positive for S aureus. K. kingae was identified in 10 and GAS in 11 different samples. S. epidermidis and S. pneumoniae (lytA) were also identified in four and one sample, respectively. In three samples of this type that were positive for S. epidermidis, another more pathogenic organism was actually identified, and in these cases S. epidermidis was assumed to be a contaminant.

Of the 11 tissue samples received as part of this study, 10 were positive by RT-PCR, of which eight were positive for S. aureus with one sample each positive for K. kingae and GAS. Three bone samples were also received, of which one was positive for S. aureus.

Overall, excluding any data from skin or throat swabs, 32 of the 44 patients (72.7%) were PCR positive for one or more of the target organisms included in the RT-PCR panel (see Tables 59–67 for detailed results).

Table 59 shows the number of positive samples detected for each target organism from all of the six sites for all sample types (including skin and throat swabs), expressed as a percentage of the total sample number from each location.

TABLE 59

Number of organisms detected by PCR according to geographical site

Table 60 identifies the number of positives for each PCR target detected based on all sample types across all geographical sites.

TABLE 60

Number of organisms detected by PCR according to sample type

Table 61 shows the number of positive organisms in samples taken from usually sterile sites (excluding skin and throat swabs) from each geographical site expressed as a percentage of the total number of fluid, tissue, bone and blood samples collected at each location.

TABLE 61

Number of organisms detected by PCR from usually sterile sites according to geographical site

Table 62 shows that a small number of tissue and fluid samples were positive for more than one of the target organisms in the bone and joint PCR panel.

TABLE 62

Patients in which multiple organisms were identified

Table 63 shows the number of positives for each PCR target detected based on blood, tissue and fluid sample types only across all geographical sites.

TABLE 63

Number of organisms detected by PCR from usually sterile sites according to sample type

Table 64 shows the total number of samples positive for any of the target organisms by sample type; these data exclude throat swabs and skin/wound swabs.

TABLE 64

Total number of positive samples according to sample type

Table 65 shows the number of positives detected for each of the multiplex targets based on patient number rather than the actual number of samples from each geographical location (i.e. combining results from duplicate samples from individual patients). These data exclude throat and skin swabs.

TABLE 65

Number of patients with organisms detected by PCR according to geographical site

PCR results according to simple and complex cases are summarised in Table 66.

TABLE 66

Summary of the results of the PCR

Table 67 shows the number of positive throat swabs for each of the target organisms taken from the six geographical sites, also expressed as a percentage of the total of throat swabs collected at each location.

TABLE 67

Number of organisms detected by PCR from a throat swab according to geographical site

Discussion

The microbiology substudy has demonstrated the diagnostic yield of the relatively cheap species-specific RT-PCR compared with routinely used microbiology culture based techniques in children’s bone and joint infections, achieving a definitive bacterial diagnosis in 32 of 44 patients [24/32 simple cases (75%) and 8/12 complex cases (67%)] in whom the test was used (see Table 66). Although the substudy was not designed to estimate formal sensitivity and specificity, a positive diagnosis in around 65–70% overall is higher than the reported yields of 40–50% (where tissue samples are achieved at surgical intervention) or 9–22% (where only blood cultures are achieved).^15,27

The study has also demonstrated the feasibility of the test being performed centrally or locally/regionally in a future randomised clinical trial. The protocol of the easy-to-perform, species-specific PCR panel is such that routine NHS molecular microbiology laboratories could adopt the test, or it could be performed on a regional or national basis for UK practice. A future trial could use results from locally or nationally performed tests depending on availability, as long as results were obtained in NHS/Public Health England-accredited laboratories performing routine molecular microbiological investigations.

The microbiology substudy has also provided important data on the current UK molecular epidemiology of children’s bone and joint infections. S. aureus remains the most commonly detected pathogen. Although pilot data (and anecdotal use of the species-specific panel on an ongoing basis at the Southampton site in routine clinical practice) suggested that K. kingae is almost as common in young children as S. aureus, these data also suggest that overall, during the period of this study, GAS (four cases detected by PCR) was as common cause of OM/SA as K. kingae (seven cases detected by PCR). In this study, GAS was identified by PCR in children aged between 2.3 and 12.4 years, whereas K. kingae was identified only in children aged 0.8–2.3 years. GAS caused simple (two cases) and complex (two cases) disease, whereas K. kingae caused only simple OM or SA. Although numbers are not high, our data confirm previous reports that K. kingae bone and joint infection primarily affects very young children. As expected, other pathogens were found more rarely.

The S. aureus data correlate with the existing literature and clinical experience as, historically, S. aureus has been known to be the most common cause of infection in both native and prosthetic bone and joint infections.⁶⁹ Thirty-seven native tissue samples from 16 patients (11 simple disease, five complex disease; age range 2.5–15.9 years, median 11.4 years) were positive for S. aureus using PCR, including eight tissue samples, nine blood cultures, one bone sample, one wound swab and 19 fluid samples (this includes joint fluids, pus and aspirates). S. aureus is a known coloniser of the upper respiratory tract, so to identify it in throat swabs is not uncommon. One patient (site 1) had significant levels of S. aureus in eight out of nine samples that had been taken, including pus and blood culture samples. Another patient (site 4) had high levels of S. aureus in both blood culture and pus, whereas in another (site 5) S. aureus was identified in three out of six samples tested.

A total of 11 samples taken from seven patients (all simple disease, age range 0.8 months to 2.3 years, median 1.15 years) were positive for K. kingae. These included 10 samples of joint fluids and aspirates and one other tissue sample. Although one group in Israel has reported that culture of tissue in blood culture bottle systems can increase the culture yield of this fastidious organism,¹⁷ this study has demonstrated again that species-specific PCR on routine blood culture samples or directly on tissue samples can formally diagnose K. kingae.¹⁹

Four children (two with simple disease and two with complex disease; age range 2.3–12.4 years, median 6.7 years) had OM/SA caused by GAS detected by PCR from a total of 14 native tissue samples. Eleven samples were fluids including pus, two were blood cultures and one was tissue. Although a higher frequency than previously reported,¹ this study was conducted during the time of a reported outbreak of GAS infections nationwide in the UK.⁷⁰

Two patients (ages 5.2 years and 0.8 months) were found to have OM/SA due to S. pneumoniae. One blood culture sample from one patient was positive for the ply gene of S. pneumoniae. One other joint sample from a different patient was positive for the more specific target for the lytA gene. The species-specific panel uses two targets to identify the presence of S. pneumoniae (the ply gene and the lytA gene). This is because the ply gene shows a degree of cross-reactivity between S. pneumoniae and S. mitis, S. oralis and S. pseudopneumoniae. The lytA target was chosen⁷¹ as it is common only in S. pneumoniae.

Haemophilus influenzae was identified in one blood culture from one patient aged 5.2 years but not from any PCR test. As expected, this was an unusual organism but potentially an important factor in the potential antibiotic choice in the ≥ 5 years age group¹ in a future RCT.

No native tissue samples in this study were positive for GBS, included in the panel because of their frequency in neonatal OM/SA.¹ As no children aged ≤ 2 months were included in this microbiology substudy, this finding is not unexpected as GBS infections are rare in older infants and children.

No samples were found to be positive for Salmonella spp., an organism included in the PCR panel because of the reported frequency in OM/SA in children with sickle cell anaemia.⁷²

Staphylococcus epidermidis is generally considered a contaminant in skin and surface swabs, and in OM/SA infections overall in the absence of prosthetic material.⁷³ However, in the absence of other pathogens detected from usually sterile sites, and when S. epidermidis is detected in samples from those previously sterile sites, the role of this organism in the pathophysiology of bone and joint infections is undergoing more detailed investigation. In this study three blood cultures and four fluid samples were positive for S. epidermidis. One patient positive for this organism was also positive for GAS in 12 different samples, whereas another patient was positive for S. aureus but at low levels. Our data concur with the overall concept that in children without prosthetic joints or artificial implants of any kind, S. epidermidis is an unlikely pathogenic cause of SA/OM even when detected in sterile samples.

Our study also investigated K. kingae and other pathogens detected by PCR in throat (oropharyngeal) swabs⁷⁴ to further investigate the relationship between OM/SA and respiratory tract carriage reported by a French group during the conduct of this study.⁷⁵ Thirty-eight throat swabs were received in total, among which K. kingae was identified in none. Thirty swabs were positive for S.pneumoniae (ply), of which only one was positive for the lytA gene, indicating that in 29 throat swabs the RT-PCR was detecting S. mitis, S. oralis or S. pseudopneumoniae as a result of ply gene cross-reactivity. Fourteen swabs were positive for S. aureus, 14 were positive for S. epidermidis, 12 were positive for H. influenzae, six were positive for GAS and one was positive for GBS.

The ability to perform species-specific RT-PCR in a local laboratory setting can benefit both patient and health-care system. By enabling antibiotic therapy targeted to a known organism, the switch from i.v. to oral therapy is facilitated, resulting in a shorter hospital stay. Although this study was not big enough for us to conduct a formal health-economic analysis, it is intuitive that being able to use directed therapy will result in shorter hospital stays as a result of earlier oral switch, as well as a reduction in treatment failure, cost, time and use of unnecessary antibiotics, and minimising the risk of antimicrobial resistance and nosocomial infections (the acquisition of which is proportional to length of stay, among other things).

Rapid identification of a bacterial organism causing OAI is important for a number of reasons. The use of conventional culture and microscopy can take a minimum of 48 hours, of which 24 hours is needed to establish organism growth (providing that it is a non-fastidious species) and 24 hours for identification and sensitivity profiling in order to correctly target antibiotic therapy. This process can be speeded up slightly by the use of matrix-assisted laser desorption/ionisation time of flight spectrometry, which can identify bacteria to a species level, providing that a pure culture can be grown initially.⁷⁶ The use of species-specific RT-PCR on bone and joint samples can speed up the identification process to a matter of hours.⁷⁷

Conclusions

This study evaluated the potential for the species-specific RT-PCR to be used both to establish the baseline UK microbiology of paediatric bone and joint infections better than would have been obtained by culture alone and to assess the use of the test in a larger RCT. We have established that the test identifies a pathogenic organism more frequently than conventional culture methods alone. The multiplex assay used has provided additional information on the UK molecular epidemiology of OM/SA in children in addition to the conventional culture results collected as part of the service evaluation (see Chapter 2). The test is feasible to use in a larger RCT and would be cheap and useful to implement in local microbiology laboratories for routine clinical care in the UK.

Data gained in the service evaluation and PCR substudy showed that the predominant pathogens in bone and joint infections in children in the UK are antibiotic-sensitive S. aureus in simple and complex disease in children of all ages outside the neonatal period; K. kingae in simple disease in young children (aged < 4 years in this study but < 6 years if currently unpublished data collected outside the time frame of the service evaluation and substudy is included); and GAS in simple and complex disease in children of all ages.