Clinical Significance of Propionibacterium acnes Recovered from Blood Cultures: Analysis of 524 Episodes^▿

Propionibacterium acnes is a microaerophilic, anaerobic, Gram-positive bacillus, and one end product of bacterial fermentation is propionic acid. The organism is a member of the normal flora of the oral cavity, large intestine, conjunctiva, and skin in humans (4). In some patients, however, P. acnes can cause severe infections, including endocarditis, intravascular infections (6, 7), central nervous system infections (11), endophthalmitis (13), and, rarely, arthritis (12, 16).

Identification of P. acnes as a pathogen is difficult because anaerobic conditions and a long incubation time are required for its culture. Moreover, P. acnes is usually considered a contaminant of blood cultures (5). The exact frequency and characteristics of clinically significant P. acnes bacteremia (PAB) have not been well known. We therefore evaluated the clinical significance and characteristics of P. acnes recovered from blood cultures in a tertiary care hospital.

The medical records of all patients, including 151 pediatric patients (less than 16 years old), who gave blood samples of which one or more were positive for P. acnes on culture, between January 1997 and August 2009, at the Asan Medical Center (a 2,700-bed tertiary affiliated hospital in Seoul, Republic of Korea) were reviewed. Recommendations for blood culture practices during the study period were as follows: (i) 3 sets of blood cultures, (ii) sampling from separate venipuncture sites (if a patient had a central venous catheter, 1 set from the central venous catheter), and (iii) 20 ml of blood sampled for each culture set with 10 ml inoculated into 1 aerobic bottle and 10 ml inoculated into 1 anaerobic bottle. One percent chlorhexidine in 70% isopropyl alcohol or 10% povadine (povidone-iodine) was used for skin disinfection, as 2% chlorhexidine is not yet available in Korea. The overall blood culture contamination rate of our hospital has been sustained at less than 1% (unpublished data). All blood cultures were processed by the hospital microbiology laboratory using a standard blood culture system (Bactec 730 or Bactec 9240; Becton Dickinson, Sparks, MD). Clinical isolates were identified using MicroScan (Dade Behring, West Sacramento, CA) or Vitek 2 (bioMérieux, Marcy l'Etoile, France). Antibiotic susceptibility tests were not routinely performed. Antibiotic susceptibilities were assessed only when a physician requested the antibiotic susceptibility information.

P. acnes blood isolates were considered significant if two or more separate blood culture sets were positive on the same day and if systemic inflammatory response syndrome (SIRS) was present without any alternate explanation; that is, patients had to be shown to have at least two of the following four criteria: (i) body temperature of >38.0 or <36°C, (ii) heart rate of >90 beats per minute, (iii) respiratory rate of >20 breaths per minute, and (iv) peripheral white blood cell counts of >12,000/mm³ or >10% bands. Severe sepsis was defined as sepsis with organ dysfunction, hypoperfusion, or hypotension, and septic shock was defined as sepsis-induced hypotension (a systolic blood pressure of <90 mm Hg or a drop in the mean arterial pressure of >40 mm Hg from the baseline) not responsive to an intravenous fluid challenge with signs of peripheral hypoperfusion (1).

Medical records were analyzed to determine underlying disease and medical condition, type of surgery or procedures, choice and duration of antibiotics used to treat P. acnes, and outcome at discharge from the hospital.

During the study period, there were 36,369 positive blood culture results, of which 524 blood culture specimens from 522 patients were positive for P. acnes, thus accounting for 1.4% (524/36,369) of the positive blood cultures. Two patients were positive for P. acnes in each of two separate cultures performed on different days. All 151 positive pediatric patients (<16 years old) were positive in only one blood culture. Forty-six adults (8.8%) were positive in two or more separate blood cultures. Of the 522 positive patients, 20 (3.8%) presented with SIRS, but 2 of these patients were excluded from the study because they had other apparent sources of infection (community-acquired pneumonia). Therefore, clinically significant bacteremia was identified in 18 (3.5%) of 522 patients with PAB. The incidence of clinically significant PAB was 1.72 cases per 100,000 admissions.

The epidemiology and clinical characteristics of the 18 patients with significant PAB are shown in Table 1. Ten patients (55.6%) were male, and the median patient age was 59.5 years (range, 30 to 72 years). Malignancy was the most common underlying disease (13 patients, 72.2%), followed by liver cirrhosis (4 patients, 22.2%) and diabetes mellitus (3 patients, 16.7%). Ten patients (55.6%) had hospital-acquired bacteremia. The sources of bacteremia were identified in only 2 patients (11.1%; both central venous catheters). Six patients underwent diagnostic or therapeutic invasive procedures, including transarterial embolization, radiofrequency ablation, bronchoscopy, cystoscopy, or pericardiocentesis, before developing PAB. One patient with a ventricular septal defect presented with infective endocarditis following inguinal herniorrhaphy surgery 2 months earlier. Two patients had polymicrobial bacteremia, in which P. acnes was associated with Peptococcus species and Staphylococcus epidermidis, respectively. Seventeen patients (94.4%) met the criteria for sepsis, none met the criteria for severe sepsis, and one (5.6%) presented with septic shock. Thirteen patients (72.2%) were treated with antibiotics, and 9 (50%) were treated with expanded-spectrum cephalosporins. The median duration of antibiotic therapy was 8 days (range, 1 to 73 days).

Clinical outcomes are summarized in Table 1. The outcome of one patient (number 16) was not available because that patient was transferred to another hospital. Of the 17 remaining patients, 1 (5.9%) died. This patient (number 17) was a 39-year-old male with hepatitis B virus-associated liver cirrhosis and a malignant mesenchymal tumor who had undergone a right posterior segmentectomy of the liver.

P. acnes, a common skin organism, belongs to the genus Propionibacterium along with P. granulosum, P. avidum, P. propionicum, and P. innocuum. Among the propionibacteria, P. acnes is the most frequent cause of human infections. Such infections are usually associated with predisposing factors, such as surgery, the presence of a foreign device, or trauma (10). Because of the increased use of indwelling foreign bodies, such as central venous catheters, prosthetic joints, prosthetic heart valves, ventriculoperitoneal shunts, and intraocular lenses, it is necessary to understand the epidemiology, clinical characteristics, and outcomes of P. acnes infection.

Few reports to date have addressed the infection characteristics and outcomes of patients with PAB. Data on such infections are usually reported when series of patients with anaerobic bacteremia are studied (3, 8, 15), although one series included nonbacteremic children (2). To the best of our knowledge, our PAB series is the largest studied to date. We found that P. acnes was responsible for 1.4% of all positive blood cultures in our hospital, within the range of previous reports (0.76 to 3.0%) (8, 14). The percentages of patients positive for coagulase-negative staphylococci (CoNS), Bacillus species, and Corynebacterium species, all of which are common blood culture contaminants, were 19.5%, 2.5%, and 1.3%, respectively (unpublished data).

Few studies have addressed the clinical significance of bacteremia caused by Propionibacterium spp. One study, of 843 instances of positive blood cultures, regarded all 48 Propionibacterium strains isolated from blood cultures as contaminants (14). A second study, of 166 patients whose blood cultures yielded anaerobic bacteria, found that none of the 53 patients with Propionibacterium bacteremia had clinically significant disease (15). A recent investigation of the clinical significance of anaerobic bacteremia in 140 patients found that, of the 46 Propionibacterium strains identified, all isolates were contaminants (8). Although we applied more stringent criteria for clinically significant PAB here than in previous studies, we found that the incidence of clinically significant PAB was 3.5% (18/522), which was higher than the 0% rate noted in previous studies. The differences may be explained in several ways. First, because P. acnes bacteremia is relatively rare and the numbers of patients in individual studies have been small, significant bacteremia caused by Propionibacterium spp. may not have been detected earlier. As our study included about 10-fold more patients with Propionibacterium bacteremia than did previous reports, it is more likely that we would encounter patients with clinically significant PAB. Second, differences in PAB frequency may be partly attributable to variation in the criteria used to define clinically significant bacteremia. In previous studies, P. acnes may have been considered a contaminant in patients with SIRS who experienced rapid recovery. We believe it erroneous, however, to regard all patients positive for Propionibacterium bacteremia as having been infected with contaminants. Several case reports of Propionibacterium bacteremia in patients with endocarditis and osteomyelitis, in which propionibacteria were indeed the true pathogens, have appeared (6, 9).

We were able to identify entry portals in only 2 of the 18 patients with clinically significant bacteremia; in both instances, central venous catheters were responsible. Seven patients (38.9%) had undergone invasive procedures or surgery before PAB, suggesting that transient bacteremia associated with such treatments may explain a substantial proportion of patients with clinically significant bacteremia. Except for the patient with endocarditis (number 3), the other six patients recovered quickly.

Among the 17 patients with clinically significant PAB, only 1 (5.9%) died of bacteremia-associated causes. The mortality rate that we observed was similar to that (5%) previously reported in patients infected with P. acnes, including children with bacteremic and nonbacteremic P. acnes (4).

Our study had several limitations. First, during the study period, we did not routinely assess the antibiotic susceptibilities of P. acnes isolates, making it difficult to determine whether patients recovered spontaneously or responded to antibiotics. Second, our study was performed on patients in a single large tertiary care center, and results may thus differ in other settings. For example, the incidence of PAB may be higher in hospitals where prosthetic implantations are frequently performed. Our hospital is one of the largest medical centers in Korea, and many surgeries, including organ transplantations (heart, lung, liver, kidney, etc.), and invasive procedures are performed. In our hospital, approximately 300 liver transplants, 200 kidney transplants, 50 heart transplants, 450 prosthetic knee arthroplasties, and 180 prosthetic hip arthroplasties have been performed annually. Therefore, our findings may not be generalizable to other hospitals.

In summary, our results suggest that P. acnes rarely causes significant bacteremia and may be associated with a low risk of mortality. Nevertheless, PAB has pathogenic potential and may play a significant role in some patients with underlying malignancy or hospital-acquired infections and especially in those with implanted devices or who have undergone invasive procedures.