Corynebacterium periprosthetic joint infection: a systematic review of 52 cases at 2.5 years follow-up

Introduction While large progress has been achieved in identifying and treating the most common pathogens involved in periprosthetic joint infections (PJI), there remains limited knowledge on atypical pathogens such as Corynebacterium. For that reason, we analyzed infection and diagnostical characteristics, as well as treatment outcome in Corynebacterium PJI. Methods A systematic review was performed based on a structured PubMed and Cochrane Library analysis using the PRISMA algorithm. The search was performed by 2 independent reviewers, and articles from 1960 to 2022 considered eligible for inclusion. Out of 370 search results, 12 studies were included for study synthesis. Results In total, 52 cases of Corynebacterium PJI were identified (31 knees, 16 hips, 4 elbows, 1 shoulder). Mean age was 65 years, with 53% females, and a mean Charlson Comorbidity Index of 3.9. The most common species was Corynebacterium striatum in 37 cases (71%). Most patients were treated with two-stage exchange (40%), isolated irrigation and debridement (21%), and resection arthroplasty (19%). Mean duration of antibiotic treatment was 8.5 weeks. At a mean follow-up of 2.5 years, there were 18 reinfections (33%), and 39% were for Corynebacterium. Initial infection by Corynebacterium striatum species was predictive of reoperation (p = 0.035) and reinfection (p = 0.07). Conclusion Corynebacterium PJI affects multimorbid and elderly patients, with one in three developing a reinfection at short term. Importantly, the relative majority of reinfections was for persistent Corynebacterium PJI.


Introduction
Periprosthetic joint infections (PJIs) remain a devastating complication following arthroplasty [1,2]. Although large progress was made in identifying the most common pathogens involved, such as Staphylococcus aureus in acute infections, as well as coagulase-negative Staphylococci (CNS) in chronic infections, there is limited knowledge on characteristics of atypical pathogens such as Candida or gram-negative bacteria [3,4].
Gram-positive bacilli are an atypical cause of PJI and often considered a contaminant in microbiology findings. The gram-positive bacillus Corynebacterium spp. is a facultatively anaerobically growing, gram-positive rod, and part of the standard flora of human skin and mucosa [3,4]. Given its primary consideration as a contaminant in the context of PJI, as well as difficulties in cultivation, and oftentimes missing standardized diagnostical tools for assessment, there are limited reports on PJI caused by Corynebacterium [3][4][5].
As such, this systematic review aimed to summarize all cases of Corynebacterium PJI, diagnostical and clinical characteristics, as well as possible treatment strategies, and reported outcomes.

Material and methods
A systematic review was performed based on a structured PubMed and Cochrane Library analysis using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) criteria. Search terms were as followed: "Corynebacterium PJI OR Corynebacterium periprosthetic joint infection OR Corynebacterium joint infection OR Corynebacterium hip infection OR Corynebacterium knee infection OR Corynebacterium shoulder infection OR Corynebacterium elbow infection OR Corynebacterium finger infection OR Corynebacterium foot infection OR Corynebacterium septic arthritis OR Corynebacterium osteomyelitis ".
Final inclusion criteria were: (1) studies published from 1960 to September 2022, (2) PJI caused by Corynebacterium, and (3) clinical and diagnostical case description as well as an available follow-up. Exclusion criteria were: (1) non arthroplasty infections (soft tissue, osteosynthesis material, native joints, osteomyelitis), (2) animal and experimental studies without patients, (3) Corynebacterium PJI in the context of new diagnostical tests without detailed clinical follow-up and case description, and (4) Corynebacterium PJI as part of larger PJI cohorts or combined with other infection types without a detailed and separate description of the Corynebacterium PJI. Although cases of osteomyelitis and septic arthritis in native joints without a prosthesis in situ were excluded in the study, search criteria included these terms as older studies partially used a different terminology than PJI.
Search was divided into two phases (Fig. 1). Phase one included the identification of publications using the upper search terms. In phase two, screening of abstract was performed based on the eligibility criteria of this study. The search was performed by two independent reviewers (YS, DK). Outcome parameters included infection characteristics (affected joint, corynebacterium species, coexisting microbes), patient characteristics (age, sex, Body Mass Index (BMI), Charlson Comorbidity Index (CCI) [6], secondary diseases), arthroplasty details (year and indication for primary implantation), diagnostical work-up (CRP, ESR, preoperative joint aspiration, intraoperative tissue samples, symptoms, histopathology), surgical and antibiotic treatment, as well as outcome evaluation (follow-up, reinfection rates, revision rates, perioperative complications, mortality).
Statistical analysis was performed using SPSS (SPSS Inc., Chicago, IL, USA), with T-and Mann-Whitney U tests for continuous variables, as well as Fisher exact test for categorical variables. A p-value < 0.05 was considered significant whereas a p-value < 0.1 was interpreted as a trend to significance.

Results
In total, 370 studies were identified using the aforementioned search terms. After exclusion of duplicates, 365 titles were screened for study inclusion, and 20 articles analyzed in detail. Of these 20 studies, 2 investigations were excluded as they were experimental studies without clinical details [7,8], 3 studies as they reported of Corynebacterium PJI as part of a larger epidemiological cohort without specification of the Corynebacterium PJI itself [9][10][11], and another 3 as they were not differentiating between orthopedic implant types or referred to general surgical site infections without implants [12][13][14].
The initial diagnosis of Corynebacterium PJI was confirmed via intraoperative tissue samples in 33, and preoperative joint aspiration in 4 cases. Panuu et al. [26] did not clarify the methods used for initial diagnosis (15 cases). A polymicrobial Corynebacterium PJI was identified in 6 cases, with co-existing pathogens including Mycobacterium avius, Peptoniphilus asaccharolyticus, Prevotella bivia, Enterobacter asburiae, Staphylococcus aureus as well as multiresistant Staphylococcus aureus (MRSA) and CNS. Among 33 cases with a description of symptom onset, 29 were of chronic entity (symptoms longer than 4 weeks), whereas 4 were acute infections (shorter than 4 weeks). Mean preoperative CRP was 18.4 mg/dl (± 15.3), mean ESR was 85.88 mm/hr (± 32.5).
Surgical treatment included debridement, antibiotics and implant retention with exchange of mobile components (DAIR) in 9 joints, permanent resection arthroplasty in 10 cases, isolated irrigation and debridement (I&D) in 11 cases, and two-stage exchange in 21 patients. One Corynebacterium PJI was identified incidentally in the course of an elbow revision due to loosening, but considered PJI by the authors, as two positive intraoperative samples of Corynebacterium spp. were found.
Mean follow-up was 30.1 months (± 35 months). There were 20 reoperations, including one for periprosthetic fracture requiring open reduction and internal fixation, one for dislocation treated with open reduction and head exchange, and 18 for reinfections. Mortality was low, with one patient dying from sepsis at 11 months. A total of 4 nonoperative complications occurred, including knee hematoma, chronic pain syndrome, and darkening of the toes in 3 TKAs, as well as one central humerus necrosis in case of a shoulder PJI.
Treatment for the 18 reinfections included DAIR in 4 cases, resection arthroplasty with spacer insertion in 4 cases, irrigation and debridement in 4 cases, above knee amputation in 3 cases, spacer exchange in one case, unspecified revision arthroplasty with extensor mechanism reconstruction in one case, and two-stage exchange in another case. In total, there were 7 reinfections by Corynebacterium striatum, including 2 mixed infections with Enterococcus faecium and Enterobacter cloacae, 6 reinfections by a different pathogen (Enterobacter asburiae; Enterococcus faecium and Candida albicans; Proteus mirabilis; Vancomycin Resistant Enterococcus (VRE); VRE; MRSA), and 5 culture negative cases.
The most common Corynebacterium species, Corynebacterium striatum, demonstrated a significantly higher rate of reoperations when compared to other Corynebacterium species (p = 0.035), and trended towards increased rates of reinfection (p = 0.07). The remaining baseline and surgical characteristics including age, CRP, ESR, presence of coexisting pathogens, duration of antibiotic treatment, and follow-up did not show a statistically significant difference among species (Table 2).

Discussion
Nondiphtheria Corynebacteria are widely considered an opportunistic commensal of the human skin and mucosa with little to unknown potential to cause infections [3]. As such, little attention has been paid to this bacterium in the context of PJI with a limited number of reports to date. This systematic review analyzed all existing Corynebacterium PJIs to date while including a total of 52 infections at a mean follow-up of 2.5 years. Our results demonstrated Corynebacterium PJIs to primarily affect total knee arthroplasties (60%) in old and multimorbid patients, with one in three joints developing recurrent infection at short term.       Knowledge on epidemiological characteristics is essential, as certain pathogens are known to be attributable to certain risk groups [27,28]. Our cohort demonstrated no tendency to a certain sex, although patients were multimorbid. In fact, more than 50% of patients showed secondary diseases, with rheumatoid arthritis and diabetes mellitus being most prevalent. Moreover, the mean age of our patients was high, falling in line with previous reports on gram-positive PJIs, and representing another risk factor for PJI [29,30]. Importantly, the knee was the most common joint affected. We believe this to be attributable to grampositive Corynebacterium being a part of the normal skin flora, whereas gram-negative pathogens were described to more frequently affect the hip, possibly due to its proximity to the gastrointestinal tract [31].
Diagnostical work-up of Corynebacterium PJI is challenging. This is due to the fact that Corynebacterium may not be part of the standard work-up, is universal part of the human microbiome, and the identification process itself is expensive [3,4,32]. Our results showed most cases to be identified out of intraoperative tissue samples. As previously acknowledged, the identification of Corynebacterium spp. as commensals of the human skin and mucosa must always be considered in the context of a possible contamination before drawing a final conclusion. Importantly, only six cases were of polymicrobial nature, and CRP and ESR were significantly increased, reducing the likelihood of possible contamination.
All but 4 infections showed a chronic symptom onset. Accordingly, the majority of cases were treated without implant retentions attempts, given a high likelihood of completed biofilm formation [33]. However, not all patients were treated according to current guidelines with a substantial number of patients undergoing DAIR and/or isolated irrigation in case of chronic infections. We believe this to be associated with a multimorbid and old patient cohort being at high risk of perioperative complications in case of complete prosthesis removal and later reimplantation. In fact, Ferry et al. attempted a pure antibiotic treatment attempt given a high risk of bleeding in their patient, before deciding upon resection arthroplasty 3 months later [21].
In addition to an adequate surgical strategy, the selection of a correct antimicrobial therapy plays an essential role [34]. In our cohort, the majority of cases were treated with Vancomycin and Cephalosporines. This is important, as Corynebacterium spp., especially the jeikeium species, has shown a resistance rate of up to 60% against various groups of antimicrobials, including Aminoglycosides, Penicillin, and Cephalosporines [12]. Although no detailed resistance pattern was reported in included studies, the use of the aferomentioned antibiotics might indicate a low rate of resistance against standard antibiotics in cases of Corynebacterium PJI.
With respect to the outcome, one in three joints developed reinfection at short term. We believe this devastating outcome to be caused by a number of factors, including a substantial number of secondary diseases, an old age, as well as a number of implant retention attempts in chronic infections [35,36]. Importantly, Corynebacterium striatum had a significantly higher rate of reinfections as opposed to other species, resulting in a failure rate of nearly 50%. While this might lead to the assumption that striatum species are a risk factor for failure in Corynebacterium PJIs, we acknowledge mean age, CRP, and ESR to be higher in this group of patients, although the effect was statistically not significant.
This systematic review had several limitations that were primarily attributable to its included studies. Foremost, we included a small number of cases with inconsistent information, resulting in a highly heterogenous group of patients. Moreover, PJI among 4 different joint types was inconsistently defined between studies, and treatment occurred over nearly 3 decades with substantially different strategies used. Finally, our results represent short-term outcomes only.
In conclusion, Corynebacterium PJI is a rare, yet severe complication occurring in the elderly and multimorbid, while resulting in significant treatment failures. One in five patients will experience Corynebacterium persistence at short term. Further studies will be necessary to draw additional conclusions on the midterm outcomes, as well as the role of the different species involved.