Introduction

Propionibacterium spp. is a facultative anaerobic Gram-positive bacterium, which is commonly found in the pilosebaceous follicles of the human skin, oral cavity, conjunctiva, respiratory and intestinal tract, and external ear canal [1]. Although it is often considered nonpathogenic, Propionibacterium spp. can cause severe infections including endocarditis, meningitis, brain abscess, endophthalmitis, conjunctivitis, and osteomyelitis or spondylodiscitis [2]. Propionibacterium spp. was also isolated in atherosclerotic lesions and several inflammatory conditions, but its pathogenetic role in these clinical situations is less clear.

The capability of forming a biofilm on the surface of medical devices is a predisposing factor for infection, especially in relation to artificial heart valves, ventriculoperitoneal shunts, joint prostheses, spinal hardware, and fracture-fixation devices [3]. However, perioperative antibiotic prophylaxis, new implant designs, and an improvement of surgical techniques and the operating room environment have considerably reduced the risk of intraoperative infection [4]. The frequency of periprosthetic joint infection (PJI) continuously increases as the number of implanted prostheses and indwelling time rise. In addition, improved diagnostic procedures, such as sonication of the removed prosthesis, inoculation in blood culture bottles, and molecular diagnostics, have improved the detection rate [3, 5–9].

Propionibacterium acnes PJI typically causes chronic infection manifesting several months after surgery, rarely in the early postoperative course [10–13]. Data on P. acnes shoulder PJI are limited, predominantly originating from case reports or smaller case series. We therefore performed a retrospective cohort study investigating the epidemiology, clinical characteristics, diagnostic pathway, treatment, and outcome of patients with shoulder PJI caused by P. acnes.

Patients and methods

Hospital setting.

The study was conducted at two university institutions, a primary and a tertiary health-care center. They are major providers of medical care, together serving about 2 million inhabitants. The study was approved by the local Ethics Committees.

Study population.

From 1 January 2010 through 31 December 2015, consecutive episodes of P. acnes shoulder PJI were included. Episodes of PJI were identified using the microbiology database and the infectious diseases consultation list. Each episode was evaluated by an orthopedic surgeon and infectious diseases specialist according to predefined criteria (see next section).

Definitions.

PJI was diagnosed when at least one of the following criteria was present [3, 14]: (1) visible purulence of a preoperative aspirate or intraoperative periprosthetic tissue (as determined by the surgeon); (2) presence of a sinus tract communicating with the prosthesis; (3) acute inflammation in intraoperative permanent tissue sections at histopathology (as determined by the pathologist); (4) microbial growth in preoperative joint aspirate, at least two intraoperative periprosthetic tissue samples, or sonication fluid of the removed implant (>50 CFU/ml); (5) synovial fluid with >2,000 leukocytes/µl or >70 % granulocytes. Time to infection was defined as the interval from arthroplasty (or last surgical intervention of prosthesis) to the diagnosis of PJI. According to the time of manifestation, PJIs were classified in early (<3 months after surgery), delayed (3–24 months after surgery), and late (>24 months after surgery) infections.

Data collection.

Hospital charts were reviewed with a standardized case-report form to retrieve demographic, clinical, and laboratory data. The following data were extracted: age, sex, underlying joint condition, prosthesis type, date of primary implantation, microbiology findings, histology findings, antimicrobial and surgical therapy, and treatment outcome at follow-up. We assessed radiological images at the time of diagnosis for signs of loosening (defined as a line zone greater than 1 mm on one or both components).

Results

Demographics and prosthesis characteristics.

The demographic and prosthesis characteristics of 20 episodes of Propionibacterium shoulder PJI are shown in Tab. 1. The median age at time of PJI diagnosis was 65 years (range, 54–77 years). The median time from implantation to diagnosis of PJI was 34 months (range, 2–60 months). The diagnosis of PJI was preoperatively confirmed in 50 % of patients and suspected in 20 %; in 30 %, PJI was not considered before surgery. In nine patients (45 %), a previous revision was performed on the index shoulder prosthesis (four septic revisions, five presumed aseptic revisions). The shoulder pain was the primary clinical sign (90 %), followed by local signs of inflammation (60 %), whereas sinus tracts (15 %) and fever (10 %) were less common. Most patients (55 %) presented more than 24 months after arthroplasty, followed by delayed presentation during the period of 3–24 months after arthroplasty (30 % of patients). Radiological loosening was present in 14 patients (70 %).

Tab. 1 Demographics and prosthesis characteristics
Full size table

The laboratory and microbiological results are shown in Tab. 2.

Tab. 2 Laboratory and microbiological characteristics
Full size table

The preoperative C‑reactive protein (CRP) value was normal (<10 mg/l) in 13 patients (65 %). Among ten patients in whom aspiration of the joint fluid was performed, leukocyte count was increased by >2,000 leukocytes/µl in nine patients (90 %) and granulocytes were increased by >70 % in all patients.

P. acnes was detected in preoperative joint aspirate in four of 12 patients (33 %) in whom the joint was punctured. Intraoperative tissue samples revealed the pathogen in 60 % and sonication of the prosthesis in 89 %. In four patients (20 %), co-infection with coagulase-negative staphylococci was present.

Surgical and antimicrobial treatment.

All 20 patients underwent surgical intervention (Tab. 3). In four patients (20 %), a one-stage exchange was performed, for whom the pathogen was known preoperatively through joint aspiration (n = 2) or during previous revision surgery for suspected aseptic loosening (n = 2). In most patients, a two-stage exchange was performed (n = 15, 75 %) with spacer implantation in the prosthesis-free interval. A short interval of 2–3 weeks until re-implantation was performed in five patients (25 %), whereas a long interval of 6–11 weeks was followed for ten patients (50 %). Temporary spacers of bone cement were impregnated with vancomycin (2 g per 40 g of polymethylmethacrylate). One patient refused re-implantation surgery and the spacer remained in place. Antimicrobial treatment included intravenous treatment for at least 1 week, followed by oral treatment for a planned total duration of 12 weeks.

Tab. 3 Treatment and outcome
Full size table

Outcome evaluation.

Patients were evaluated at regular follow-up visits with a median follow-up time of 26 months (range, 6–47 months). Evaluation of clinical signs and symptoms of infection, CRP values, and x‑ray analysis was performed 3–6 months after termination of antimicrobial treatment. At last follow-up, 18 patients (90 %) showed no signs or symptoms of infection. One patient refused re-implantation of the prosthesis and retained the spacer. At follow-up 17 months after explantation this patient was infection-free.

Discussion

While growth of P. acnes was previously often considered contamination, its pathogenic role was revealed in the past decade [2]. Improved diagnostic tests, higher awareness of possible PJI, and a better understanding of the pathogenesis in PJI have increased the detection of P. acnes PJI, especially after shoulder arthroplasty [12]. The slow growth rate and low virulence of P. acnes are the reason for delayed clinical presentation of infection [10]. The typical latency period from prosthesis implantation until clinical manifestation ranges from 1 to 3 years, but can extend to 10 years or more after implantation [15]. In the present study, the median latency between implantation and clinical manifestation of PJI was 2.8 years (34 months) and the longest period was 5 years. However, many patients probably had nonspecific signs and symptoms several months before diagnosing the PJI, such as chronic shoulder pain or decreased mobility. In the present study, in 30 % of patients PJI was not suspected preoperatively because of subtle clinical symptoms. This finding underlines the importance of the low threshold of differentiating “normal” postoperative pain or limited function from low-grade PJI [3].

For the correct diagnosis of chronic PJI, joint aspiration should be performed, although arthrocentesis may be difficult in shoulder joints. In synovial fluid, the leukocyte count and differential should be determined and the specimen sent for microbiological examination [4, 9, 16]. In P. acnes PJI, synovial fluid cultures are often false negative, as this facultative anaerobic pathogen is difficult to culture, requiring anaerobic culture media and prolonged incubation time (10–14 days) [2]. Owing to the slow replication rate and adherence to the implant surface, microbiological detection is challenging and contributes to the false-negative preoperative cultures, thus causing delay in the diagnosis of infection [2]. In four patients in our study (27 %), PJI was only confirmed by sonication of the implant and could not be diagnosed in preoperative joint aspirate or intraoperative periprosthetic tissue. Tunney et al. [17] demonstrated, that by using tissue cultures most PJI were missed, in particular P. acnes, which were only found by sonication of the removed prosthesis. Several other investigators confirmed the added value of sonication in the diagnosis of PJI [5, 6, 18].

Positive culture with isolation of low-virulent skin flora may be misinterpreted as contamination if the diagnostic work-up is not performed properly and does not include corresponding histopathological examination and cell count. Therefore, a meticulous diagnostic sampling of tissue and synovial fluid, which allows for confirmation of inflammation, is crucial in interpreting the microbiological results.

In PJI with P. acnes, CRP and blood leukocyte count are often normal and therefore do not allow exclusion of the PJI [3]. In our study, CRP levels were normal in the majority of patients (80 %).

Almost half of the patients in our study (40 %) had undergone at least one previous revision surgery for nonspecific symptoms (with partial or complete exchange of the prosthesis) before diagnosis of P. acnes PJI. Even if P. acnes was not found during previous revisions, the symptoms may have already been caused by this pathogen. In other studies, previous revision has been found to be a risk factor for PJI with P. acnes [19, 20].

Owing to its broad antimicrobial susceptibility, including rifampin as biofilm-effective antibiotic, P. acnes-caused PJI can be treated with a one-stage revision (absence of a sinus tract provided) or a two-stage revision with a short interval (2–3 weeks) [4].

In summary, P. acnes PJI typically manifests months to years after implantation and the infection is often not suspected before surgery. In patients with persistent pain or prosthesis loosening, low-grade PJI should be excluded, in particular infection caused by P. acnes. Only a complete diagnostic work-up with histopathological examination and joint aspiration with determination of cell count and differential allow for the discrimination between infection and contamination of this skin-resident organism.

Conclusion

  • Prosthetic shoulder infections caused by P. acnes often show a subtle clinical presentation due to the low-virulence of the organism, which causes delay or missing of the diagnosis.

  • Owing to the slow replication rate and adherence to the implant surface, as well as the need for specific culture media and incubation time, microbiological detection is challenging and contributes to the false-negative preoperative cultures.

  • Only a complete diagnostic work-up with histopathological examination and joint aspiration with determination of cell count and differential allow for the discrimination between infection and contamination of this skin-resident organism.

  • Because of its broad susceptibility including rifampin as biofilm-effective antimicrobial agent and rare formation of sinus tracts, P. acnes-caused PJI can be treated with a one-stage revision or a two-stage revision with a short interval, depending on the bone and soft tissue condition.