Coppens E; Pittet DR; Al-Mayahi M; Cunningham G; Holzer N; Hoffmeyer P; Ladermann A; Uckay I

Research Article

Austin J Infect Dis. 2016; 3(1): 1021.

Characteristics of Propionibacterium acnes Infections in Orthopaedic Surgery: 11 Year Study in a Referral Center, 2004-2014

Coppens E¹*, Pittet DR², Al-Mayahi M¹, Cunningham G¹, Holzer N¹, Hoffmeyer P¹, Ladermann A¹ and Uckay I³

¹Orthopaedic Surgery Service, University of Geneva Hospitals, Switzerland

²Service of Infectious Diseases, University of Geneva Hospitals, Switzerland

³Orthopaedic Surgery Service, Service of Infectious Diseases, University of Geneva Hospitals, Switzerland

*Corresponding author: Coppens Elia, Orthopaedic Surgery Service, University of Geneva Hospitals, Geneva, Switzerland

Received: May 13, 2016; Accepted: June 15, 2016; Published: June 17, 2016

Abstract

Propionibacterium acnes has been associated with late, smoldering and healthcare-associated infections of the shoulder and spine. However, the characteristics of P. acnes with respect of other anatomical locations, patient populations and bacterial culture characteristics remain largely unknown. This retrospective single center observational study compared ten variables (age, sex, type and site of infection, presence and type of hardware, C-reactive protein levels, immunodeficiency state, rate of polymicrobial cultures and type of co-existant germs) of orthopaedic patients suffering from P. acnes infections with those of patients with non-Propionibacterium infections. P. acnes was isolated intraoperatively in 37 of 2716 (1.36%) orthopaedic infections. The lumbar region (odds ratio 7.4, 95% CI 1.2-46.3), the shoulder (OR 9.9, 1.6-60.1) and the presence of hardware (OR 8.2, 2.4-28.4) were significantly associated with P. acnes infection; while sex and age were not. P. acnes was rarely identified (3/1008 vs. 34/1708; p<0.01) among immunodeficient patients, in foot infections, prosthetic joints, and tibia nails and never in septic bursitis. Lower CRP serum levels (< 50 mg/l) were observed more often in P. acnes cases than in controls (49% versus 31%, p<0.01). P. aeruginosa or MSSA were almost never identified as a co-pathogen of P. acnes. Combination of these clinical variables should help attending physicians and surgeons to reinforce suspicion for P. acnes infection, and take appropriate measures when needed. Further research should shed light on P. acnes’ tropism for plates and spondylodesis as opposed to arthroplasties.

Keywords: Propionibacterium acnes; Orthopaedic; Infection; Characteristics; Co-pathogen

Abbreviations

P. acnes: Propionibacterium acnes; CRP: C-Reactive Protein; OR: Odds Ratio; CLSI: Clinical and Laboratory Standard’s Institute; EUCAST criteria: European Committee on Antimicrobial Susceptibility Testing; ROC: Receiver Operating Characteristic; MSSA: Methicilline-Susceptible Staphylococcus Aureus

Introduction

Propionibacterium acnes is a Gram-positive non-obligate anaerobic bacterium colonizing the lipid-rich sebaceous glands of the skin [1]. It belongs to the clinical group of skin commensals, but can occasionally cause serious infections [2]. Determining the presence of P. acnes infection can be difficult. Due to its low virulence, infections are less dolent [3] and markers of infection (fever, sedimentation rate, leucocyte count) are reported of unreliable character in P. acnes infections [4]. Infections may remain latent for several months [5]. Histopathological findings of acute inflammation, chronic inflammation or foreign-body reaction are said not to correlate with P. acnes [6].

Importantly, P. acnes has several particularities, for example, it prefers certain anatomical regions, although it can cause infection in almost any body site [1]. P. acnes can be recovered in the presence or

not of devices. Besides its name-giving affection to acne vulgaris, P. acnes is linked to many chronic orthopaedic and cardiac prosthetic and implant infections [7-10], as well as breast [11] or eye implant infections [12]. P. acnes can produce biofilms [13], and is involved in the mechanism of septic and possibly aseptic loosening of orthopaedic implants [14]. In orthopaedic surgery, P. acnes is historically known to infect shoulder and clavicular infections, especially in the presence of implants [1]. A third prevalent anatomic localization is the lumbar spine, in which anaerobic conditions or long term surgery (associated with cutaneous contamination) are reportedly linked to P. acnes infections [15,16].

Surprisingly, the characteristics of P. acnes with respect to all orthopaedic infection locations, the presence and type of orthopaedic material, patients’ characteristics and microbiologic findings remain little documented. The purpose of this study was to compare P. acnes infections with other bacteriological infections in an orthopaedic unit and determine whether there is a difference between groups in the proportion of: anatomical sites; the presence and type of hardware; age, sex and immune deficiency status; CRP levels, and identified germs. Identification of these clinical variables should help attending physicians and surgeons to reinforce suspicion for P. acnes infection, hence allowing them to take appropriate measures. This retrospective single center observational study tries to address these questions.

Materials and Methods

Study design and setting

The Orthopaedic Service of University of Geneva Hospitals has 132 acute care beds (24 on the Septic Ward). Dedicated infectious diseases consultants (since the year 2000) have established a number of databases regarding orthopaedic infections as approved by our hospital Ethics Committee [15]. We retrospectively analyzed these data over a period of 11 years in regards to the epidemiology of P. acnes.

Participants/study subjects

All first clinical infection episodes in adult patients, hospitalized in our orthopaedic and traumatology center between January 2004 and December 2014 were included. All infection types (osteoarticular, cutaneous, subcutaneous, bursitis, with or without hardware or prosthesis) were included. The diagnosis of P. acnes infection was based upon the presence of P. acnes in at least two intraoperative samples [17], together with clinical signs of infection (novel pain, fever, sinus discharge) and/or radiographic signs [18]. We excluded recurrences, episodes with insufficient data, infections without intraoperative micriobiological samples and paediatric cases.

Description of experiment

The databases were cross-matched with the Hospital’s Coding Office files for patient identification and retrospectively analyzed was performed. We processed all specimens in our clinical microbiology laboratory and the procedures corresponded mainly to CLSI (Clinical and Laboratory Standard’s Institute) recommendations [19] remained unchanged throughout the entire study period except for switching to EUCAST criteria (European Committee on Antimicrobial Susceptibility Testing) in spring 2014 [20]. Sonication was not performed. A median of 5 days of microbiological culture time was performed in CDC Anaerobe 5% Sheep Blood Agar media, enriched with brain-heart infusion.

Variables, outcome measures, data sources, and bias

Ten variables (age, gender, type and site of infection, presence and type of hardware, C-reactive protein levels, immunodeficiency state, rate of polymicrobial cultures and type of co-existant germs) were collected per episode on an EXCEL™ sheet, and compared between cases and controls. To confront our findings with data from other centers, we performed a literature review. A computerized strategy using MEDLINE, EMBASE, EFFORT and AAOS, from January 1946 until March 2015, searched for the following key words: ‘infections’, ‘Propionibacterium acnes’, ‘orthopaedic’, ‘trauma’, and ‘incidence’. The search was limited to in vivo human studies of adults published in the English language concerning the field of orthopaedic surgery. We also hand searched the reference lists of the retrieved studies. After having eliminated all duplicates, we retained 28 original articles. 14 articles concerned shoulder infections, 4 concerned spine infections, 4 articles discussed hip and knee prosthesis infections and 3 articles related to positive P. acnes cultures in orthopaedic implants. Of particular interest, 3 papers addressed the overall incidence of P. acnes in post-operative osteoarticular infections and associated risk factors [7-9].

Statistical analysis, study size

Group comparisons were performed using the Pearson-χ2, the Fisher-exact or the Wilcoxon-ranksum-tests, as appropriate. An unmatched logistic regression analysis determined associations with the outcome “P. acnes infection”. Independent variables with a p value ≤0.30 in univariate analysis were introduced stepwise in the multivariate analysis. We included 5-10 outcome events per predictor variable [21]. Key variables were checked for confounding, co-linearity and interaction, the latter by Mantel-Haenszel estimates and interaction terms. Age and CRP levels were analyzed as continuous and categorized variables. P values =0.05 (all two-tailed) were significant. STATA™ software (9.0; College Station, USA) was used. The goodness-of-fit-value was 0.99 and the ROC-value 0.85, indicating a high accuracy of our final model.

Results

Demographics, description of study population

2716 infection episodes were retained for analysis. The median age of all patients was 57 years (range 17-99 y); 1845 (69.93%) were males and 1008 (37.11%) were immunodeficient. 665 surgical procedures (24%) involved hardware. Hardware were arthroplasties (n=319; 11.75%), plates (n=149; 5.49 %), nails (n=54; 2%), spondylodesis material (n=30; 1.10%), screws (n=20; 0.74%), cerclage cables or wires (n=14; 0.51%) or hip screws (n=5; 0.18%). 1055 (38.84%) infectious episodes were associated with abscesses, and 468 (17.23%) were septic bursitis cases. 567 infections (20.87%) were polymicrobial. Common pathogens were MSSA (n=1040; 38.29%), Streptococcus (n=434; 15.97%) and P. aeruginosa (n=253; 9.31%).

P. acnes population

P. acnes was isolated intraoperatively in 37 of 2716 (1.36%) infection episodes (Table 1).

Table 1: Comparison of P. acnes infections with other infections in orthopaedic surgery.




  
    n =    2716
    P. acnes infection 
      n = 37 
    Comparison
      p value*
    non-P. acnes infection (control)
      n =    2679
  
  
    Male gender
      Median age
      Age group = 50 years
      Age group 51 - 80 years
      Age group > 80 years
      Immune deficiency+
    29    (78%)
      46    (range 17-84)
      20    (54%)
      16    (43%)
      1 (3%)
      3 (8%)
    >    0.05
      -
      >    0.05
      >    0.05
      >    0.05
      < 0.000
    1827    (68%)
      57    (range 17-99)
      1065    (40%)
      1296    (48%)
      318    (12%)
      1005    (37%)
  
  
    Spine
      Shoulder
      Foot
      Other
    10    (27%)
      6 (16%)
      1 (3%)
      20    (54%)
    < 0.00
      < 0.00
      0.029
      >    0.05
    69 (3%)
      87 (3%)
      423    (16%)
      2100    (78%)
  
  
    Presence of hardware
      Plates
      Nails
      Spondylodesis
      Prosthesis
      Other
    24    (65%)
      6 (16%)
      1 (3%)
      5 (14%)
      5 (14%)
      7 (19%)
    <0.000
      0.004
      >    0.05
      <0.000
      >    0.05
      0.004
    634    (24%)
      143    (5%)
      53 (2%)
      25 (1%)
      313    (11%)
      100    (4%)
  
  
    Joint and bone
      Bursitis
      Abscess formation
      Serum CRP level <50 mg/L
      CRP 51 - 200 mg/L
      CRP > 200 mg/L
    32    (86%)
      0
      12    (32%)
      18    (49%)
      12    (32%)
      7 (19%)
    -
      0.005
      >    0.05
      0.033
      0.033
      0.033 
    1662    (61%)
      468    (17%)
      1043    (39%)
      838    (31%)
      845    (32%)
      996    (37%)
  
  
    Polymicrobial
      MSSA
      Pseudomonas aeruginosa
    15    (41%)
      1 (3%)
      0 (0%)
    0.015
      <0.000
      0.035
    552    (21%)
      1039    (44%)
      253    (11%)
  
  
    * Pearson-?² or Fisher exact-test;    Variables in bold are statistically significant (two-tailed p value <0.05).    MSSA = Methicillin-Sensitive Staphylococcus aureus; CRP= C-Reactive Protein    level on admission.⁺Immune-suppressive therapy, dialysis, cirrhosis Child C, human    immunodeficiency virus infection, active malignancy, pregnancy, splenectomy, agranulocytosis,    diabetes mellitus.



Table 1:  Comparison of P. acnes infections with other infections in orthopaedic surgery.

Hardware was present in 24 cases (64.86%) and mainly involved plates (n=6; 25.00%), spondylodesis (5; 20.83%) and arthroplasties (5; 20.83%). 32 cases (86, 48%) were joint or bone infections, 5 were cutaneous or subcutaneous infections. 12 episodes (32.43%) presented an abscess. We found no septic bursitis cases linked to P. acnes. P. acnes was more frequently identified during procedures in the presence compared with the absence of foreign orthopaedic material (24/665 vs. 13/2051; p<0.01). After multivariate logistic regression analysis, the presence of hardware (OR 8.2, 2.4-28.4) remained significantly associated with P. acnes infection. A significant association was found between P. acnes, plates (Odds Ratio 3.4, 95% CI 1.40-8.36) and spondylodesis (OR 16.6, 95% CI 5.97-46.06). In contrast, differences concerning prosthetic joints, tibia nails and the presence of abscesses were found not to be significant (p>0.05).

Anatomical sites of P. acnes infections concerned mostly the spine (n=10; 27.02%) and shoulder (n=6; 16.22%) region (Table 1). Few P. acnes infections were caudal to the lumbar region. After group comparison between P. acnes (cases) and non-P.acnes infections (controls) (Tables 1 & 2), proportion of P. acnes in the spine (12%) and shoulder (6%) was found to be significantly different. Due to differences in crude group comparisons, a multivariate logistic regression analysis adjusted for case mix was performed (Table 2). The lumbar region (Odds Ratio 7.4, 95% CI 1.2-46.3) and the shoulder (OR 9.9, 1.6-60.1) remained significantly associated with P. acnes infection. Among P. acnes infections, 29 (78%) were male; median age was 46 years (range 17-84 y) and 3 patients (8.10%) were immunodeficient. P. acnes was rarely identified among immunodeficient patients (3/1008 vs. 34/1708; p<0.01). After multivariate logistic regression analysis, sex and age were not associated with P. acnes infection. Lower CRP serum levels (< 50 mg/l) were observed more often in P. acnes cases than in controls (49% versus 31%, p<0.01). 15 (40, 54%) infections were polymicrobial. Microbiological cultures were more often polymicrobial in P. acnes infections than compared to controls (p=0.015), and P. aeruginosa or MSSA were almost never associated to polymicrobial cultures of P. acnes. Co-existing microorganisms in polymicrobial cultures (Table 3) were S. epidermidis (n=8), Corynebacterium (n=4), S. agalactiae (n=1), S. warneri (n=1) and MSSA (n=1).

Table 2: Unmatched logistic regression analyses with outcome “P. acnes orthopaedic infection”.




  
    
    Univariate    analysis (OR)
    Multivariate    analysis (OR)
  
  
    Female gender
      Age (continuous variable)
      Immune deficiency⁺
      CRP levels (continuous variables)
    0.6    (0.26-1.30)
      1    (0.97-1.00)
      0.1    (0.04-0.48)
      1    (0.99-1.00)
    n.a.
      n.a.
      n.a.
      n.a.
  
  
    Spine
      Shoulder
      Foot
    7.3    (2.96-18.12)
      5.7    (2.34-14.28)
      0.1    (0.02-1.08)
    7.4    (1.18-46.06)
      9.9    (1.61-60.6)
      1.0    (0.13-9.40)
  
  
    Hardware
      Plates
      Nails
      Spondylodesis
      Prosthesis
      Abscess
    6    (3.01-11.76)
      3.4    (1.40-8.36)
      1.4    (0.18-10.22)
      16.6    (5.97-46.06)
      1.5    (0.60-3.53)
      0.7    (0.38-1.50)
    8.2    (2.35-28.33)
      n.a.
      n.a.
      n.a.
      n.a.
      1.4    (0.40-4.72)
  
  
    Results as OR (odds ratio) with 95%    confidence interval. n.a.= not applicable due to interaction.
      Variables in bold are statistically    significant (two-tailed p value <0.05).
      ⁺Immune-suppressive therapy, dialysis,    cirrhosis Child C, human immunodeficiency virus infection, active malignancy,    pregnancy, splenectomy, agranulocytosis, diabetes mellitus.



Table 2:  Unmatched logistic regression analyses with outcome “P. acnes
orthopaedic infection”.

Discussion

This retrospective study confirmed a strong association between P. acnes infections and bone and joint infections, plate and spondylodesis infections, mainly in shoulder and spine regions. P. acnes infections were associated with lower CRP level elevation than other infections. Of interest, immunodeficient patients were more likely to have infections due to other pathogens than P. acnes. Microbiological cultures of P. acnes were more often polymicrobial, but co-infection with P. aeruginosa or MSSA was rare. Age and sex were found not to be correlated with P. acnes.

P. acnes deep tissue infection in orthopaedic surgery remains a rare complication outside of eventual nosocomial outbreaks. Our study identified 37 (1.36%) P. acnes infections in 2716 orthopaedic infection episodes, which is comparable to literature: Walter et al. [9] retrospectively reviewed anaerobic bone and joint infections of orthopaedic implants in 3144 patients and noted a P. acnes infection risk of 0.63% (20/3144). Titécat et al. [7] attributed P. acnes to 5.8% of all orthopaedic infections, with a predilection to prosthetic infections of the upper limb. Kanafani et al. [8] analyzed the global incidence of P. acnes after orthopaedic surgery of major joints over a period of 11 years and identified P. acnes as responsible for infection in 1.6/1000 procedures.

The median of 5 days microbiological culture time of P. acnes in this retrospective study, and the absence of sonication [22] could have resulted in an underestimation of P. acnes infection rate: Between 2000 and 2011, a significant increase in the number of P. acnes isolates in orthopaedic surgery has been observed [7,23], probably inherent to the prolonged incubation time of the samples. Ideal incubation time observation periods longer than one week are found to have a higher sensitivity for P. acnes detection [24,25] and some colleagues propose an optimal incubation period of 13 day [5]. The strength of this study is the requirement for two intraoperative samples with the same P. acnes to be counted as infection, together with clinical signs of infections. Deep tissue infection specimens are 6.3 times more likely to have =2 culture media positive for P. acnes than specimens from nondiagnostic events [17]. Gram stain is currently not recommended because of its low sensitivity and negative predictive value [4]. The relative lower CRP levels (<50 mg/l) associated with P. acnes infections, as observed in this study, can be explained by a lower inflammatory response of the host to P. acnes, inherent to its indolent nature. In this study, P. acnes microbial cultures were more often polymicrobial than cultures from controls, compatible with P. acnes’ tolerance for other germs (Table 3). The primary identified co-existing germ was S. epidermidis (22%), known to live in close proximity with P. acnes on human skin. Of interest, antagonistic interactions between these two species have been described and it could be envisaged that anti-P. acnes activity of S. epidermidis could be exploited, for example as a probiotic treatment approach against P. acnes-associated diseases [26]. Selective competition in favour of fast growing bacteria like P. aeruginosa or MSSA, might make identification of P. acnes in presence of these pathogens less likely and can lead to a misidentification or underestimation of P. acnes in culture.

Download PDF

Citation: Coppens E, Pittet DR, Al-Mayahi M, Cunningham G, Holzer N, Hoffmeyer P, et al. Characteristics of Propionibacterium acnes Infections in Orthopaedic Surgery: 11 Year Study in a Referral Center, 2004-2014. Austin J Infect Dis. 2016; 3(1): 1021.

Instruction for Authors

Submit Your Article

	*Cases (P. acnes) 37*	*Control (non-P. acnes) 2679*
*Propionibacter acnes*	37	0
*Methicilline-susceptible Staphylococcus aureus*	1	1039
*Streptococcus*	0	434
*Pseudomonas aeruginosa*	0	253
*Methicillin-resistant Staphylococcus aureus*	0	172
*Staphylococcus epidermidis*	8	97
*Corynebacterium*	4	12
*S. agalactiae*	1	70
*S. warneri*	1	7