Antimicrobial Susceptibility of Propionibacterium acnes Isolates from Shoulder Surgery

Abstract

Orthopedic surgeons at our institution have noticed an increase in the number of infections due to Propionibacterium acnes, especially following operations on the shoulder. We collected P. acnes isolates from our hospital microbiology laboratory for 1 year and performed antimicrobial susceptibility testing on 28 strains from the shoulder. Antibiotics with the lowest MIC values against P. acnes (MIC₅₀ and MIC₉₀) included penicillin G (0.006, 0.125), cephalothin (0.047 and 0.094), and ceftriaxone (0.016, 0.045), while others also showed activity. Strains resistant to clindamycin were noted.

TEXT

Propionibacterium acnes has been recognized as a significant and emerging pathogen in orthopedic surgery over the last 10 years, especially after operations on the shoulder and especially following shoulder arthroplasty with prosthetic material (1–3). At our institution, the number of recognized cases of P. acnes has increased noticeably over the last few years. Formulating treatment recommendations for patients with P. acnes orthopedic infections can be difficult, however, given the paucity of data on antibiotic susceptibility patterns. Many of the early reports on P. acnes antibiotic susceptibilities are from patients with facial acne who were attending dermatology clinics. Other studies are now decades old (4, 5) or originate from countries in which antibiotic usage patterns are much different from those in North America (6, 7), leading us to question whether we can rely on those reports.

We performed testing on 33 strains of P. acnes collected between November 2010 and December 2012, 28 of which were from orthopedic surgeries on the shoulder and 5 of which were isolates from bloodstream or other deep infections. P. acnes was identified using the MicroScan rapid anaerobe identification method (Siemens Healthcare, W. Sacramento, CA). We tested this strain collection against a panel of 10 antibiotics, with an emphasis on antibiotics that might actually be used to treat postoperative orthopedic infections. We used the Etest method (bioMérieux, Durham, NC) on anaerobic brucella blood agar. Two previous studies indicated that MICs obtained by Etest correlated well with results obtained using the agar dilution method (8, 9), and other experts consider the Etest an accepted method (10). Etest results obtained using anaerobic brucella blood agar were compared to those obtained using CDC anaerobe blood agar on a subset of 6 strains. Inocula were prepared to a 0.5 McFarland standard from 48 h of growth on anaerobic blood agar. Blood agars were from Anaerobe Systems (Morgan Hill, CA). We tested only a single isolate from each patient, although most patients had more than one positive culture. P. acnes was cultured anaerobically at 37°C for 48 to 72 h using the GasPak EZ anaerobe container system (BD Corp., Franklin Lakes, NJ).

The results of our study are reported in Table 1. MICs measured on brucella blood agar were the same as those measured using CDC anaerobe agar. The penicillins (penicillin G and amoxicillin) and cephalosporins (cephalothin and ceftriaxone) showed strong activity against P. acnes, with many isolates showing MICs lower than the lowest antibiotic concentration on the Etest strip. For example, for 15 of 28 isolates, the MIC of amoxicillin was less than 0.016, the lowest concentration on the Etest strip. The strong activity of benzylpenicillin (penicillin G) against P. acnes was noted by Oprica et al. (11) but is not often considered an option by orthopedic surgeons. We expected that moxifloxacin would show much greater activity against P. acnes than ciprofloxacin but were surprised to see that moxifloxacin's MIC₅₀ and MIC₉₀ (0.125 and 0.38, respectively) were only about 2-fold lower than those of ciprofloxacin (0.25 and 0.5, respectively). Vancomycin is often viewed as a first-choice drug for postoperative infection in orthopedics due to its activity against staphylococci, but its activity against P. acnes in our study was only fair (MIC₅₀, 0.38; MIC₉₀, 0.5). This is especially relevant since the concentration of vancomycin required to eradicate P. acnes in an established biofilm (biofilm eradication concentration) was ≥128 μg/ml in recent studies (12, 13). Clindamycin was active against many P. acnes strains, but some strains were highly resistant (Table 1 and Fig. 1) (clindamycin MIC₅₀ and MIC₉₀ of 0.032 and 8.5, respectively). A histogram of the distribution of MICs for clindamycin and linezolid showed distributions that appeared to be bimodal (Fig. 1), a pattern not seen with the other antibiotics tested (data not shown). Although our strain collection may be too small to definitely conclude that there is a bimodal distribution, the presence of a subset of highly clindamycin-resistant P. acnes strains has also been noted in other studies of P. acnes, including collections of isolates from skin and deep systemic infection (11, 14).

Table 1.

MICs of P. acnes strains

Antibiotic	MIC (mg/liter) for orthopedic isolates from the shoulder (n = 28)		CLSI breakpoint (mg/liter)	% of strains resistant
	MIC50	MIC90
Amoxicillin	0.028	0.117	–a	–
Penicillin G	0.006	0.125	≤0.5	4
Cephalothin	0.047	0.094	–	–
Ceftriaxone	0.016	0.045	–	–
Clindamycin	0.032	8.5	≤2	7
Ciprofloxacin	0.25	0.5	–	–
Moxifloxacin	0.125	0.38	≤2	0
Ertapenem	0.032	0.141	≤4	0
Vancomycin	0.38	0.5	–	–
Linezolid	0.25	0.93	–	–

^a–, no interpretive standards from the Clinical and Laboratory Standards Institute (CLSI), and therefore, the percentage of resistant strains cannot be determined.

Fig 1.

Histogram of MIC values for clindamycin and linezolid from the orthopedic isolates. MIC values for the strains are displayed as a histogram, with the number of isolates falling into each bin of the MIC ranges shown. (A) Clindamycin; (B) linezolid.

Ertapenem (MIC₅₀, 0.032; MIC₉₀, 0.141) also showed good activity against P. acnes, which is notable because its once-daily administration makes home intravenous antibiotic administration more feasible. In addition, many patients who are reportedly allergic to penicillin are able to tolerate carbapenems.

The MIC₅₀ values of the shoulder isolates were less than or equal to the MIC₅₀ from the nonshoulder isolates for all antibiotics tested. For example, the shoulder isolates showed an MIC₅₀ value to vancomycin of 0.38 mg/liter, compared to an MIC₅₀ of 1.5 mg/liter for the nonshoulder isolates. (The nonshoulder isolates included one isolate each from bloodstream, tibia, hip prosthesis, cornea, and mediastinal lymph node.) However, larger numbers of nonshoulder isolates are needed to make definitive comparisons of antibiotic susceptibilities between different body sites.

Although MICs obtained with CDC anaerobe agar and brucella blood agar were identical, we noticed one difference between the two agar types. Some P. acnes strains were strongly hemolytic on brucella blood agar, but these strains did not show hemolysis on CDC anaerobe agar, even though both agars contain 5% sheep blood. P. acnes strains showing hemolysis on brucella blood agar appeared to have a more aggressive clinical course than the nonhemolytic strains. Our investigation of the role of hemolysis in P. acnes is being submitted for publication elsewhere.

Guidelines for treatment of deep-seated P. acnes infections, including prosthetic joint infections, are few and are based mostly on anecdotal experience. In most reports, clindamycin and vancomycin are mentioned as the first drugs to consider for deep-seated P. acnes infections (15). Our data, like those in the study by Oprica et al. (11), suggest that penicillins and first-generation cephalosporins (cefazolin and cephalothin) show promise and should have an expanded role in studies of clinical efficacy. The variability in P. acnes susceptibility to clindamycin argues for routine antimicrobial susceptibility testing for clindamycin and other antibiotics. In addition, an expanded role for oral antibiotics should be considered in treatment of P. acnes infections, since many of the drugs that we found to be highly active in vitro, including amoxicillin, moxifloxacin, and linezolid, are also highly bioavailable. Levy et al. advocated for a greater use of oral antibiotics, such as an oral β-lactam plus rifampin, for treatment of P. acnes prosthetic infections (2), but actual practice has lagged behind these concepts, partly because of widely held biases in favor of intravenous treatment and also due to the lack of controlled clinical studies. Further work in this area may result in improved outcomes of P. acnes infections at substantially lower costs than with traditional approaches.