Background: Propionibacterium is commonly recovered from explants or surrounding tissues in revision shoulder arthroplasty. Rather than attempting to differentiate a true infection from a false-positive result on the basis of the number of positive cultures, we characterized the amount of these bacteria in each specimen and shoulder.
Methods: The study included 137 revision shoulder arthroplasties from which a minimum of 4 specimens had been submitted for culture and at least 1 was positive for Propionibacterium. Standard microbiology procedures were used to assign a semiquantitative value (0.1, 1, 2, 3, or 4), called the Specimen Propi Value, to the amount of growth in each specimen. The sum of the Specimen Propi Values for each shoulder was defined as the Shoulder Propi Score, which was then divided by the total number of specimens to calculate the Average Shoulder Propi Score.
Results: The number and percentage of positive specimen-specific cultures (of material obtained from the stem explant, head explant, glenoid explant, humeral membrane, collar membrane, other soft tissue, fluid, or other) per shoulder ranged from 1 to 6 and 14% to 100%. A high percentage of specimens (mean, 43%; median, 50%) from the culture-positive shoulders showed no growth. Only 32.6% of the fluid cultures were positive in comparison with 66.5% of the soft-tissue cultures and 55.6% of the cultures of explant specimens. The average Specimen Propi Value (and standard deviation) for fluid specimens (0.35 ± 0.89) was significantly lower than those for the soft-tissue (0.92 ± 1.50) and explant (0.66 ± 0.90) specimens (p < 0.001). The Shoulder Propi Score was significantly higher in men (3.56 ± 3.74) than in women (1.22 ± 3.11) (p < 0.001). Similarly, men had a significantly higher Average Shoulder Propi Score (0.53 ± 0.51) than women (0.19 ± 0.43) (p < 0.001).
Conclusions: This investigation suggests that Propionibacterium is unevenly distributed within culture-positive revised shoulders. As a result, the specimen number and source (explant, soft tissue, or fluid) have major influences on the culture results for a revised shoulder arthroplasty. We found no evidence that suggested useful threshold values for defining a true infection.
Level of Evidence: Diagnostic Level III. See Instructions for Authors for a complete description of levels of evidence.
Propionibacterium is commonly grown on culture of specimens harvested during revision shoulder arthroplasties performed for pain, stiffness, or component loosening, even though a patient may not show obvious clinical, laboratory, or radiographic signs of periprosthetic infection1-5. Some authors have tried to develop criteria for distinguishing “true infections,” “false-positive cultures,” and “contamination” in such cases on the basis of clinical signs and number of positive cultures6. However, because the clinical signs lack sensitivity and specificity for predicting Propionibacterium infection and because often as few as 2 specimens are submitted for culture, these criteria have proven difficult to apply consistently7-12.
Our experience indicates that there is no obvious, practical threshold for culture results obtained at the time of revision shoulder arthroplasty that can differentiate a true Propionibacterium infection from a false-positive result. Instead, the results vary from a minimal presence of Propionibacterium (a small amount of growth from one of several specimens submitted) to an overwhelming presence (heavy growth in a high percentage of the specimens). Thus, rather than attempting to refine a definition of periprosthetic Propionibacterium infection for shoulder arthroplasty akin to what has been done for hip and knee arthroplasty (based, for example, on the presence or absence of pathogens in at least 2 separate tissue or fluid samples)13, we suggest that studies of revision shoulder arthroplasty would be better informed if they (1) routinely included culture data from at least 4 tissue or explant specimens and (2) used measures of the bacterial load in the specimens.
In this investigation of patients who had had at least 1 positive culture of specimens obtained during a revision shoulder arthroplasty, we tested the hypotheses that (1) the degrees of culture positivity vary widely among specimens harvested during same shoulder revision, (2) the degrees of positivity are higher for cultures of humeral and glenoid component explant and soft-tissue specimens than for fluid cultures, (3) the total of the degrees of positivity for all specimens from a shoulder are associated with the patient’s sex, and (4) there is no obvious threshold for culture results that allow a distinction between a true-positive and a false-positive finding of infection.
Materials and Methods
When we revise a failed arthroplasty because of pain, stiffness, or component loosening (but without obvious clinical evidence of infection), we routinely harvest specimens of explants, soft tissues, and/or joint fluid with individual sterile instruments before intravenous antibiotic administration. These specimens are submitted for culture according to a defined protocol to optimize the opportunity of recovering Propionibacterium2,14. Specifically, all specimens are cultured for 3 weeks in blood agar (Trypticase soy agar with 5% sheep blood), chocolate agar, Brucella agar (with blood, hemin, and vitamin K), and brain-heart infusion broth. All media, with the exception of the Brucella agar, are incubated at 37°C with 5% CO2 for 28 days. Brucella agar plates are incubated anaerobically at 37°C for 28 days. Media are visually examined daily for growth, and the culture plates are opened only if growth is noted.
This study was approved by our institutional review board (#51461). In a review of the 223 revision arthroplasties performed by the senior author between April 10, 2007, and February 9, 2016, 168 shoulders (75%) were found to have at least 1 culture that was positive for Propionibacterium. For 31 shoulders, 3 or fewer specimens were submitted for culture, and these shoulders were omitted from this analysis, leaving 137 culture-positive shoulders, in 34 women (average age [and standard deviation], 58.2 ± 13.2 years; range, 27 to 78 years) and 103 men (average age, 61.4 ± 10.8 years; range, 22 to 81 years; p = 0.27), from which at least 4 specimens had been submitted for culture2.
As is the case at many centers, our microbiology laboratory reports positive cultures in semiquantitative terms (growth in broth only, 1 colony, 1+, 2+, 3+, or 4+) on the basis of the findings at the end of the 3-week culture period. “Broth only” indicates that only the broth was culture positive (the streaked plate did not show growth), and “1 colony” indicates that only 1 bacterial colony was noted on the streaked plate. The scale of 1+ to 4+ indicates the number of quadrants (1 to 4) that showed growth on a plate streaked with the standard technique15. As reported previously, we assigned numerical values (Specimen Propi Values) to the semiquantitative Propionibacterium culture results: 0.1 (broth only), 0.1 (1 colony), 1, 2, 3, and 4 (1+, 2+, 3+, or 4+, respectively)2.
The Specimen Propi Values were summed for each type of specimen (humeral stem explant, humeral head explant, glenoid explant, collar membrane [between the modular head and stem], humeral membrane [between the humeral stem and humeral bone], other soft tissue, fluid, or “other”) from each shoulder. Each tissue specimen was harvested with a previously unused sterile rongeur, with care taken to avoid contact with other tissue or implants. The explant specimens were obtained by vortexing the retrieved implant in 3 mL of sterile saline solution. The fluid specimens were obtained with a sterile syringe. The Specimen Propi Values for all of the specimens from a particular shoulder were summed to derive the Shoulder Propi Score for that shoulder2. If ≥2 different colony types or species were reported for the same specimen, the assigned values were added. We have noted that prior clinical studies commonly refer to all species of Propionibacterium as “P. acnes,” even though other species of Propionibacterium are found in specimens retrieved during shoulder revisions16. In this analysis, laboratory reports of P. acnes, P. avidum, P. granulosum, P. humerusii, P. species, and “presumed” Propionibacterium were all counted as Propionibacterium. Because of the small numbers of organisms other than P. acnes, we did not separately analyze the data for the different species. Because the Shoulder Propi Score was influenced by the number of culture specimens submitted, we used a third metric, the Average Shoulder Propi Score, defined as the Shoulder Propi Score divided by the total number of specimens from that shoulder submitted for culture.
The average Specimen Propi Values (and standard deviation) for the different types of specimens (e.g., explant, soft tissue, and fluid) were determined, and the significance of the differences between the results of fluid cultures and (1) explant-specimen cultures and (2) soft-tissue cultures were analyzed with an unpaired t test. The Shoulder Propi Score and Average Shoulder Propi Score were determined for male and female patients, and the significance of the differences was analyzed with an unpaired t test. Significance was defined as p < 0.05. Because of the possibility that the load of Propionibacterium was related to age-dependent hormonal changes, we also compared the distribution of Shoulder Propi Scores by patient age and sex.
An average of 6.62 ± 2.61 specimens from female patients and 6.69 ± 2.06 specimens from male patients were cultured (p = 0.88). On average, 3.1 of the specimens were of soft tissue, 2.5 were from explants, and 1.1 were of joint fluid. The number and percentage of positive specimen-specific cultures per shoulder ranged from 1 to 6 and 14% to 100% (Fig. 1). Neither of these parameters demonstrated a threshold that might be used to identify true-positive culture results. Propionibacterium was found to varying degrees in specimens of different types and from different locations. Shoulders with positive cultures had a high percentage of specimens that showed no growth (mean, 43%; median, 50%). Figure 2 and Table I show an example of uneven Propionibacterium distribution. In this case, the Propionibacterium was refractory to multiple single-stage revisions; the results of the preoperative laboratory studies of serum inflammatory markers were all normal.
Only 32.6% of the fluid cultures were positive in comparison with 66.5% of the soft-tissue cultures and 55.6% of the cultures of the explant specimens. The average Specimen Propi Value for fluid (0.35 ± 0.89) was significantly lower than that for soft tissue (0.92 ± 1.50) and explant specimens (0.66 ± 0.90) (p < 0.001). Men had a significantly higher mean Shoulder Propi Score (3.56 ± 3.74) than women (1.22 ± 3.11) (p < 0.001), but patient age did not have a significant effect on the score (Fig. 3). Similarly, men had a significantly higher Average Shoulder Propi Score (0.53 ± 0.51) than women (0.19 ± 0.43) (p < 0.001), but patient age did not have a significant effect (Fig. 4). The Shoulder Propi Score and Average Shoulder Propi Score varied among the shoulders that were culture-positive for Propionibacterium, without a clear threshold above which a true-positive result could be defined with confidence.
The Workgroup of the Musculoskeletal Infection Society has developed criteria for periprosthetic infections of the hip and knee13. Investigators have had difficulty applying similar criteria to periprosthetic infections of the shoulder, largely because most shoulder infections are of the “stealth” type, presenting as pain, stiffness, and apparent aseptic loosening without the features characteristic of obvious infection. This difference is likely due to the fact that the most common bacteria recovered from the sites of failed shoulder arthroplasties are low-virulence organisms, such as Propionibacterium4. The presence of Propionibacterium may be overlooked because the presentation is subtle, special culture techniques are required for reliable detection2, surgeons often collect a small number of specimens for culture2, and the infections often present years after the index arthroplasty3. For these reasons, the development of a practical definition of periprosthetic shoulder infection has been difficult. Thus, the evaluation and management of failed shoulder arthroplasties have been complicated by concerns about “unexpected positive cultures,” “false-positive cultures,” and “contamination.”5,6,10,11,17-21
Up to this point, efforts to interpret the results of cultures of specimens obtained at the time of revision surgery have been based on the number of cultures that were positive (for example, it has been suggested that ≥2 cultures positive for Propionibacterium is clinically relevant)7-12. The problem with this approach is that the number of specimens that are culture-positive depends on (1) the nature of the specimens (explant, tissue, or fluid), (2) how many specimens are submitted for culture, (3) what media are used for culture, and (4) how long the cultures are observed2. Inconsistency among these variables confuses attempts to compare data among patients and among studies and confounds efforts to establish guidelines for diagnosis and management of shoulders with possible periprosthetic infection.
This study presents three straightforward metrics that can characterize the results of cultures, assuming that multiple specimens are harvested and that the specimens are cultured for Propionibacterium on aerobic and anaerobic media. The Specimen Propi Value is assumed to be a reflection of the bacteria burden in each specimen submitted for culture, with an indication of whether the specimens were obtained from explants, soft tissue, or joint fluid. The Shoulder Propi Score indicates the total of the Specimen Propi Values for all specimens obtained from a specific shoulder. The Average Shoulder Propi Score indicates the total of the Specimen Propi Values for specimens obtained from a specific shoulder (the Shoulder Propi Score) divided by the number of specimens submitted for that shoulder. While not proven in this study, shoulders with a high Average Shoulder Propi Score seem likely to have larger numbers and concentrations of bacteria (i.e., a higher bacterial load) than those with a lower score.
It is important to note that, within the same shoulder, the Specimen Propi Values may be 0 for some specimens and highly positive for others collected during the same surgery. Soft-tissue and explant specimens had higher rates of culture positivity than joint fluid; this observation is consistent with the tendency of Propionibacterium to reside in a biofilm rather than existing in planktonic form in joint fluid that is accessible by aspiration. Conclusions about the presence of Propionibacterium in a shoulder need to be based on the number and nature of the specimens submitted for culture.
While it has been previously noted that cultures of specimens taken during revisions of failed shoulder arthroplasties are more likely to be positive for male patients than for female patients2-4, our study provides the first evidence that the average bacterial load (as reflected by Shoulder Propi Scores and Average Shoulder Propi Scores) is almost 3 times higher in men than in women with at least 1 positive culture of a specimen taken during shoulder revision.
The Shoulder Propi Scores and Average Shoulder Propi Scores for the shoulders for which at least 1 culture was positive for Propionibacterium varied without a clear threshold above which a true-positive result could be defined with confidence.
These results need to be viewed in light of certain limitations. These retrospective data representing one of several possible approaches to specimen harvest are from the practice of an individual surgeon and the microbiology laboratory at one institution and thus the results of this study may not be generalizable. Although our sampling protocol has become more standardized over time, there was some variability in the number and type of specimens submitted for each revision case as well as variability in the sample size among the different sample types and among the samples of the same type. We also did not validate the Specimen Propi Score by comparing it with a validated quantitative measure of the number of bacteria present. We currently have yet to investigate the relationship of these culture metrics and treatment protocol to the clinical outcomes of revision shoulder arthroplasty.
In conclusion, we presented a method for standardizing the collection and reporting of culture data obtained at the time of revision shoulder arthroplasty. The Specimen Propi Values provide a means of demonstrating the differences in results among different types of specimens submitted for culture. They showed that Propionibacterium is not evenly distributed through a culture-positive shoulder. The Shoulder Propi Score and Average Shoulder Propi Score provide a means for characterizing the load of Propionibacterium in a revised shoulder and showed highly significant differences between men and women. The metrics used in this study demonstrate a continuous range of variability among patients, without evidence of a threshold that would clearly distinguish shoulders that are truly infected. We believe that, rather than viewing a specimen culture result as positive or negative, there is value in noting the degree of positivity as a reflection of the bacterial burden in the specimen and in the shoulder. Attention to the degree of positivity, or bacterial load, may provide clinically useful information about the importance of culture results in the evaluation and management of failed shoulder arthroplasty. As is the case for all laboratory tests, culture results need to be considered in the context of all other information available about the patient and the shoulder.
Investigation performed at the Department of Orthopaedics and Sports Medicine, University of Washington, Seattle, Washington
A commentary by Grant Garrigues, MD, is linked to the online version of this article at jbjs.org.
Disclosure: There was no external funding for this study. On the Disclosure of Potential Conflicts of Interest forms, which are provided with the online version of the article, one or more of the authors checked “yes” to indicate that the author had a relevant financial relationship in the biomedical arena outside the submitted work.
- Copyright © 2017 by The Journal of Bone and Joint Surgery, Incorporated