Clinical features and outcome of Streptococcus agalactiae bone and joint infections over a 6-year period in a French university hospital

Background Bone and joint infections (BJIs) due to Streptococcus agalactiae are rare but has been described to increase in the past few years. The objective of this study was to describe clinical features and outcomes of cases of S. BJIs. Methods We conducted a retrospective analysis of adult cases of S. agalactiae BJIs that occurred between January 2009 and June 2015 in a French university hospital. The treatment success was assessed until 24 months after the end of antibiotic treatment. Results Among the 26 patients included, 20 (77%) were male, mean age was 62 years ± 13 and mean Charlson comorbidity index score was 4.9 ± 3.2. Diabetes mellitus was the most common comorbidity (n = 14, 54%). Six had PJI (Prosthetic Joint Infections), five osteosynthesis-associated infections, 11 osteomyelitis and four native septic arthritis. Eleven patients had a delayed or late infection: six with a prosthetic joint infection and five with an internal fixation device infection. Sixteen patients (62%) had a polymicrobial BJI, most commonly with Gram-positive cocci (75%) notably Staphylococcus aureus (44%). Polymicrobial infections were more frequently found in foot infections (90% vs 44%, p = 0.0184). During the two-year follow-up, three patients died (3/25, 12%) and seven (7/25, 28%) had treatment failure. Conclusion Diabetes mellitus was the most common comorbidity. We observed an heterogenous management and a high rate of relapse.

Introduction Group B Streptococcus (GBS) or Streptococcus agalactiae is a well-characterised pathogen of infants and pregnant women. However, invasive GBS infections are increasingly observed in non-pregnant adults (two-thirds of patients) and have become a major health concern [1]. Common clinical manifestations in non-pregnant adults include skin or soft-tissue infections, urinary tract infections, pneumonia, bacteraemia with no identified focus, arthritis and osteomyelitis [1]. GBS septic arthritis in a diabetic patient was first reported in 1940 [2], ever since the burden of GBS in invasive infections such as bone and joint infections (BJIs) is increasing [3][4][5][6]. However, few studies have described arthritis and osteomyelitis infections due to GBS in non-pregnant adults. The objective of this study was to describe clinical characteristics and outcomes of all cases of S. agalactiae BJIs that occurred in our hospital between 2009 and 2015.

Study population
We retrospectively reviewed all cases of S. agalactiae BJIs in Nîmes University Hospital in the South of France. Our hospital has a 1,979-bed capacity, which includes one orthopaedic surgery department and one infectious diseases unit. All S. agalactiae BJIs including arthritis, osteomyelitis, internal fixation device infections and prosthetic joint infections were identified from the Microbiology Laboratory database using different codes: "Streptococcus"; "bone samples"; "deep samples"; "orthopaedist surgery"; "osteoarticular infection"; "bone infection". Data were reviewed from January 2009 to June 2015. Those with non-bone and joint infection were excluded.
S. agalactiae BJIs were identified based on past medical history with clinical evidence of infection using biological and/or radiological data, with at least two positive cultures of S. agalactiae identified from deep samples taken during surgery, joint aspirate samples and blood cultures. Prosthetic joint infections were defined according to the IDSA guidelines criteria [7] and classified according to the time from prosthesis implantation to the onset of infection as: early (<3 months), delayed ([3 months-2 years]) and late (>2 years) infections. Internal fixation device infections were classified according to the time of onset after implantation: early (< 3 weeks), delayed ([3-10 weeks]), and late (>10 weeks) [8].

Data collection
All patient files were reviewed between June and September 2017 to collect the following characteristics: sociodemographic, pregnancy and comorbidities (heart failure, chronic liver disease, diabetes mellitus, inflammatory rheumatism, obesity, immunodeficiency, peripheral arterial obstructive disease, and peripheral neuropathy) as well as consumption of alcohol and tobacco. The Charlson comorbidity index (a combined age-comorbidity score used to estimate relative risk of death from prognostic clinical covariates) was calculated for each patient [9].
The location of infection and the presence of prosthetic joint or internal fixation device were recorded. The antimicrobial and/or surgical treatments performed as well as treatment outcomes, assessed at months 3, 6, 12 and 24 after the end of antibiotic treatment, were reviewed for every patient. Outcomes were assessed through phone calls and medical chart.
Treatment failure was defined as either (i) the recurrence of the same prosthetic joint infections by GBS at any time after the first line of medical and surgical treatment (i.e., relapse of infection); (ii) the recurrence of the same prosthetic joint infections because of the presence of a same species with a different antibiogram or another species at any time after the first line of medical and surgical treatment (i.e., reinfection), or (iii) death directly caused by sepsis resulting from active BJI without another known infection.

Specimen collection and microbiological analysis
Deep biopsies samples were obtained during surgery or at patient bedside. Joint fluids, tissue samples, or bone biopsies were crushed, plated on different agar media (Columbia blood agar with 5% sheep blood agar, chocolate, Mueller-Hinton, trypticase soy, MacConkey agar plates and Schaedler tube (BioMerieux, Marcy L'Etoile, France)) and incubated at 37˚C in 5% CO 2 for up to seven days. Bacterial cultures were identified with the Vitek 2 identification card (2009-2013) and the Vitek-MS system (2013-2015) (BioMérieux). The antibiotic susceptibilities of S. agalactiae isolates to amoxicillin, penicillin G, tetracycline, erythromycin, clindamycin, rifampicin, cotrimoxazole, vancomycin and teicoplanin were determined by agar disk diffusion method and interpreted according to the recommendations of the French Society for Microbiology and the European Committee on Antimicrobial Susceptibility Testing 2017 (http://www.sfm-microbiologie.org).

Statistical analysis
Results are expressed as n (%) for categorical variables and median (minimum; maximum or Interquartile Range (IQR)) or mean (standard deviation (SD)) for continuous variables. Comparisons were done using the Fisher's exact tests for categorical variables. A p-value <0.05 was considered statistically significant. Data analyses were performed using GraphPad Prism version 6.00 for Mac OS X, (GraphPad Software, La Jolla California USA).

Ethics approval and consent to participate
Approval from the Institutional review board of our university hospital was obtained (IRB Centre Hospitalier Universitaire Régional Caremeau, Nîmes, France N˚15/06.07). All procedures performed in the study were in accordance with the ethical standards and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Written informed consents have been obtained from all the patients included. No identification data are disclosed.

Sociodemographic (Table 1)
Overall, 26 patients with S. agalactiae BJIs were identified. Twenty patients (77%) were male, median age was 62 years (range 27-80 years) and mean Charlson comorbidity index was 4.9 (SD 3.2). Most patients (22/26, 85%) had one or more underlying conditions. Diabetes mellitus was the most common comorbidity (14, 54%). The most common risk factors were tobacco use (10, 38%) followed by alcohol abuse (5, 19%). Four patients had solid cancer (15%) among whom three had radiotherapy on a bone that was infected afterwards. None of the patients was pregnant.
Overall, 15 (58%) patients had native BJI (11 osteomyelitis and four native septic arthritis), six (23%) patients had a prosthetic joint infection (five hip and one knee) and five (19%) patients had an internal fixation device infection (two in the ankle, two in the foot and one in the tibia). All patients with a prosthetic joint or internal fixation device had a delayed or late infection. The mean time between orthopaedic device implantation and infection onset was 137 months (range 4-300). Infections occurred between 4 months to 8 years after implantation for prosthetic joint infections, and between 4 months to 6 years for internal fixation devices. Native BJIs were located to shoulder, sacrum (pressure ulcer) or sternum joint (n = 1), tibia or knee (n = 2), the majority were diabetic foot infections (8, 31%).
Ten of the polymicrobial infections (10/16, 63%) were foot osteomyelitis: two patients with an internal fixation device infection and eight patients with diabetic foot infection.

PLOS ONE
Polymicrobial infections were more frequent in foot infections than in other locations (90% vs 44%, p = 0.0184). Six polymicrobial infections occurred in patients with prosthetic joint or internal fixation devices. There was no difference in polymicrobial infections rate between native (6/11, 54%) and prosthetic joint or internal fixation device infections (10/15, 67%) (p = 0.53). All S. agalactiae isolates were susceptible to amoxicillin, benzylpenicillin, vancomycin and teicoplanin. Sixty percent of isolates were resistant to tetracycline, 35% to erythromycin and 20% to clindamycin. Only 14 isolates were tested for rifampicin: 8 (57%) were susceptible, four (29%) intermediate and two were resistant (14%). Rifampicin resistance was associated with cotrimoxazole resistance in both isolates, in addition to erythromycin resistance in one case and tetracycline resistance in the other.
In the three patients who died, one S. agalactiae strain was susceptible to all the antibiotics tested, one strain was only resistant to tetracycline and one strain was resistant to tetracycline, erythromycin and intermediate to rifampicin.

Management (Tables 2 and 3)
Surgical treatment was performed in all six prosthetic joint infections: four had two-stage reimplantation (one died before the second prosthetic implantation); one had a one-stage reimplantation; one had conservative treatment (debridement only). All internal fixation devices were removed surgically with bone debridement. Half of the native bone/joint infections (7/ 15, 47%) had a surgical debridement.
All but two patients with prosthetic joint infections underwent probabilistic antibiotic regimen via a combination of vancomycin with a beta-lactam (ceftriaxone, cefotaxime or imipenem) and/or aminoglycoside (gentamicin). One patient had an initial ofloxacin monotherapy for a sepsis of unknown origin but switched to a combination of amoxicillin and gentamycin when S. agalactiae bacteraemia was diagnosed at day 4. The second patient had linezolid initial monotherapy having recently been treated with this antibiotic for a haematoma infected by E. faecalis and S. aureus in front of the prosthetic joint.
In patients with internal fixation device infections, two had empiric treatment of intravenous vancomycin (monotherapy or combined with ceftriaxone) and the other three patients

PLOS ONE
received adapted oral therapy. Among the 15 patients with native bone/joint infections, adapted antibiotic treatment was given to seven patients and an initial empiric treatment was given to the other eight patients of intravenous vancomycin combined with gentamicin or piperacillin/tazobactam, amoxicillin combined with gentamicin, piperacillin/tazobactam in monotherapy or amoxicillin/clavulanic acid in monotherapy.

Clinical outcomes (Tables 2 and 3)
One patient was lost to follow-up. Follow-up was complete for the 25 remaining patients. Among them, three died (3/25, 12%), 15 were in remission (15/25, 60%), and seven had treatment failure (7/25, 28%) of whom five had polymicrobial infection on a diabetic foot. Treatment failure were relapse in five patients (including polymicrobial infection with S. agalactiae in two cases) and reinfection in two cases. The latter two patients had hip prosthesis infection, both whom had had their prosthesis removed via a 2-stage revision strategy. In the first patient, M. morganii and S. epidermidis were identified during the second stage re-implantation. The other patient relapsed eight months after the second re-prosthesis implantation from E. faecalis infection. Neither diabetes mellitus (p = 0.13) nor concomitant bacteraemia at the time of diagnosis (p = 0.28) were statistically associated with treatment failure.
All three patients who died had a high Charlson score (from 8 to 11) and a monomicrobial S. agalactiae infection. One had a native knee infection and two had a hip prosthesis infection. Both dead when their prosthetic joint were removed.

Discussion
S. agalactiae is a growing cause of invasive infections in adults. We report 26 S. agalactiae BJIs from January 2009 to June 2015. Most of the infections occurred in patients with

PLOS ONE
comorbidities. Diabetes mellitus is a known major risk factor for GBS BJIs, present in 20 to 48% of patients [6,[10][11][12] and associated with unfavourable clinical outcome [10]. Two-thirds of the patients had polymicrobial infection (mainly with S. aureus and coagulase-negative staphylococci) which differs from the series of Seng et al. including 37 cases of BJIs due to S. agalactiae that were usually monomicrobial infections [10]. The rate of bacteraemia did not differ between monomicrobial and polymicrobial infections and was similar to that reported in the series of Seng (15%) [10]. In contrast to results found by Fiaux et al. [13], concomitant bacteraemia at the time of diagnosis did not negatively impact the outcome for streptococcal prosthetic infections.
The management of GBS BJIs was very heterogeneous in our patients depending on the time until infection, the presence of orthopaedical device or multiple bacteria. Fiaux et al. suggested that rifampicin combined with another agent, especially levofloxacin, was an effective and well-tolerated treatment in their series of 95 streptococcal prosthetic joint infections [13]. In our study, among monomicrobial S. agalactiae BJI, two combinations with rifampicin were used, once with a good outcome and the second with a relapse with other bacteria. In polymicrobial BJI with S. agalactiae and S. aureus, four oral combined antibiotic treatments with rifampicin and a second agent were given with a good outcome. Previous studies have shown that rifampicin was active against the majority of Streptococcus strains [10,13]. In our experience, only 57% of tested strains were susceptible to this antibiotic. This result suggests that rifampicin should be tested in order to preferentially propose this antibiotic in the GBS prosthetic BJIs therapeutic regimen.
The rate of relapse observed in our study was high (27%), in accordance with rates reported in the literature (18-32.7%) [10,13,14] which is probably due to the fact that 5 out of 7 relapses occurred in patients with diabetic foot infection and a polymicrobial infection. Our mortality rate was high (12%) but comparable to other studies [14,15]. Interestingly, the three patients who died with a S. agalactiae BJI in our study had a Charlson score > 8. S. agalactiae related BJIs are known to be more severe with a poorer outcome as showed by Zeller et al. In 2009, this team described 24 cases of S. agalactiae prosthetic hip infections, compared to 115 otherpathogens cases. Their findings showed that S. agalactiae prosthetic hip infection had a poorer outcome compared to other-pathogens [14].
This work had some limitations. First, this was a monocentric study in mainland France, with an observational retrospective design and a small sample size. Second, the studied population was heterogeneous with monomicrobial and polymicrobial infections, native and prosthetic joints infections, including diabetic foot infections, which makes management recommendations difficult.

Conclusion
We observed an heterogenous management and a high rate of relapse in our patients with S. agalactiae BJIs probably due to the fact that infections were often polymicrobial diabetic foot infection.