Periprosthetic joint infections of the shoulder: A 10-year retrospective analysis outlining the heterogeneity among these patients

Abstract

Background

Diagnosis and treatment of shoulder periprosthetic joint infection is a difficult problem. The purpose of this study was to utilize the 2018 International Consensus Meeting definition of shoulder periprosthetic joint infection to categorize revision shoulder arthroplasty cases and determine variations in clinical presentation by presumed infection classification.

Methods

Retrospective review of patients undergoing revision shoulder arthroplasty at a single institution. Likelihood of periprosthetic joint infection was determined based on International Consensus Meeting scoring. All patients classified as definitive or probable periprosthetic joint infection were classified as periprosthetic joint infection. All patients classified as possible or unlikely periprosthetic joint infection were classified as aseptic. The periprosthetic joint infection cohort was subsequently divided into culture-negative, non-virulent microorganism, and virulent microorganism cohorts based on culture results.

Results

Four hundred and sixty cases of revision shoulder arthroplasty were reviewed. Eighty (17.4%) patients were diagnosed as definite or probable periprosthetic joint infection, of which 29 (36.3%), 39 (48.8%), and 12 (15.0%) were classified as virulent, non-virulent, or culture-negative periprosthetic joint infection, respectively. There were significant differences among periprosthetic joint infection subgroups with regard to preoperative C-reactive protein (p = 0.020), erythrocyte sedimentation rate (p = 0.051), sinus tract presence (p = 0.008), and intraoperative purulence (p < 0.001). The total International Consensus Meeting criteria scores were also significantly different between the periprosthetic joint infection cohorts (p < 0.001).

Discussion

While the diagnosis of shoulder periprosthetic joint infection has improved with the advent of International Consensus Meeting criteria, there remain distinct differences between periprosthetic joint infection classifications that warrant further investigation to determine the accurate diagnosis and optimal treatment.

Introduction

Periprosthetic joint infection (PJI) of the shoulder represents a challenging problem for orthopaedic surgeons. The difficulty associated with these complications involves both the diagnosis and treatment because delineating between tissue colonization, contamination, and infection is difficult.^1–10 Historically, there have been numerous ad hoc definitions of shoulder PJI used in the literature. ⁶ In 2018, the International Consensus Meeting (ICM) utilized the Delphi method to develop criteria for the diagnosis of shoulder PJI.^4,5 While this definition is in its early stages and is therefore likely to evolve, it still provides treating physicians with a framework for both diagnostic work-up and discussion of probability of shoulder PJI.^5,6,11,12

Based on ICM criteria, diagnosing a patient with a definitive shoulder PJI requires a major clinical finding such as a sinus tract communicating with the prosthesis, intra-articular purulence, or two positive tissue cultures with the same virulent organism. ⁵ On the other hand, patients with a “probable” shoulder PJI do not present with these overt clinical signs of infection and therefore are subject to diagnosis based on an accumulation of “minor” clinical findings (Figure 1). ⁵ While it is unclear why there is such heterogeneity in the clinical presentations of patients with possible shoulder PJIs, it is clear that there is a difference and this difference may be attributed to the organism causing the infection.^13,14 Cutibacterium acnes is one of the most common organisms identified in shoulder PJI infections.^15,16 Clinical presentations of patients with a PJI associated with C. acnes tend to be less severe and non-specific compared to PJIs caused by other more virulent organisms.^2,15,17

Figure 1.

Major and minor criteria outlined by the 2018 international consensus meeting for diagnosing shoulder periprosthetic joint infection.

The purpose of this study was to employ the ICM criteria to evaluate the clinical presentations among patients diagnosed with a “probable” or “definite” shoulder PJI. Our hypothesis was that there would be an objective difference in the clinical presentation between patients diagnosed with a “definite” and “probable” shoulder PJI based on whether the diagnosis was associated with a virulent or non-virulent organism.

Methods

After institutional review board approval, a retrospective review of all patients at a single institution undergoing revision shoulder arthroplasty surgery in the setting of a prior ipsilateral shoulder arthroplasty (hemi, anatomic, or reverse) was conducted between 2010 and 2020. Our inclusion criterion was to include any patient who underwent a revision shoulder arthroplasty. We excluded patients who did not have all data for the major criteria of the 2018 ICM shoulder PJI criteria. The electronic medical records for all associated patients were reviewed for demographic variables, clinical presentations, laboratory data, treatments provided, and other objective clinical data. It was standard of care at our institution during the study period to take five intraoperative cultures during every revision shoulder arthroplasty that were held in the microbiology lab for 14 days. Clinical presentations and laboratory data were then utilized to classify patients as having an “unlikely,” “possible,” “probable,” or “definite” shoulder PJI based on the 2018 ICM shoulder PJI criteria. ⁵ All patients classified as “definitive” or “probable” shoulder PJI were classified as having a PJI. All patients classified as having a “possible” or “unlikely” PJI were classified as aseptic.

The virulence of an organism refers to the degree of pathology caused by the organism. Although there is some variability among the same species, infections with certain organisms have a greater effect on the host and are thus more virulent. The PJI cohort was subsequently divided into culture-negative, non-virulent microorganism, and virulent microorganism cohorts based on their culture results. Organism virulence was determined via consultation and consensus among three board-certified infectious disease physicians, one of whom aided in the production of several of the ICM infection definitions (Figure 2). These cohorts were then compared to delineate the differences in their clinical presentations as defined by the ICM criteria.

Figure 2.

Organism virulence classification.

Statistical analysis

Student’s t-tests and ANOVA tests were used to compare continuous variables. Chi-square tests were utilized to compare categorical data. For all statistical tests, a p value <0.05 was used to determine statistical significance.

Results

Patient demographics and revision surgical procedures

After application of inclusion and exclusion criteria, 460 cases of revision shoulder arthroplasty were identified over the course of this 10-year review. Of these, 297 underwent one-stage revision, 105 underwent two-stage revision, and 31 patients underwent component explantation with insertion of an antibiotic spacer and retained the antibiotic spacer throughout the follow-up period. Eighty (17.4%) patients were diagnosed with a definite or probable shoulder PJI based on ICM criteria. Twenty-six (8.8%) patients in the single-stage and 54 (18.2%) patients in the dual-stage revision groups were considered to be definite or probable PJI (p < 0.01). There were no significant differences in patient demographics between patients undergoing revision shoulder arthroplasty for PJI versus aseptic revision (Table 1). Of the 80 patients diagnosed with a PJI, 29 (36.3%) had positive cultures for a virulent organism, 39 (48.8%) had cultures positive for a non-virulent organism, and 12 (15.0%) were culture negative (Table 2). Patients with a virulent PJI were more likely to be female (62.1%) compared to non-virulent PJIs (35.9%) and culture-negative PJIs (33.3%), respectively (p = 0.07, Table 2). There were no significant differences between the virulent, non-virulent, and culture-negative PJI cohorts in age, body mass index (BMI), Charlson Comorbidity Index (CCI), American Society of Anesthesiologist (ASA) physical classification system scores, type of index surgery, or time from index surgery (p > 0.05, Table 2).

Table 1.

Patient demographic comparisons between international consensus meeting unlikely or possible versus probable and definite periprosthetic joint infections.

	Aseptic (n = 380)	PJI (n = 80)	p
Age	65.5 (8.39)	65.8 (10.7)	0.71
Sex, no. (%)
Male	44 (55.0%)	179 (47.1%)	0.25
Female	36 (45.0%)	201 (52.9%)
BMI	31.3 (6.56)	30.1 (6.28)	0.14
CCI	1.00 (1.00)	0.78 (1.17)	0.37
ASA	2.44 (0.53)	2.52 (0.66)	0.53
Laterality
Left	34 (42.5%)	143 (37.6%)	0.49
Right	46 (57.5%)	237 (62.4%)

Aseptic: Unlikely or Possible periprosthetic joint infection based on international consensus meeting criteria.

PJI: Definite or Probable Periprosthetic joint infection based on International Consensus Meeting Criteria; BMI: body mass index; CCI: non-age-adjusted Charlson Comorbidity Index; ASA: American Society of Anesthesiologist score.

Table 2.

Patient demographics according to organism attributed to probable and definite periprosthetic joint infection defined by the ICM criteria.

	Virulent PJI (n = 29)	Non-virulent PJI (n = 39)	Culture-negative PJI (n = 12)	p
Age (years)	66.4 (9.84)	64.4 (7.83)	66.7 (6.29)	0.55
Gender
Female	18 (62.1%)	14 (35.9%)	4 (33.3%)	0.07
Male	11 (37.9%)	25 (64.1%)	8 (66.7%)
BMI (m/kg 2 )	29.8 (5.49)	31.4 (6.99)	34.3 (6.93)	0.14
CCI	1.30 (1.16)	0.33 (0.58)	0.83 (0.75)	0.32
ASA	3.00 (0.00)	2.25 (0.50)	2.33 (0.58)	0.26
index surgery
Anatomic TSA	11 (37.9%)	14 (35.9%)	2 (16.7%)	0.35
Reverse TSA	11 (37.9%)	19 (48.7%)	5 (41.7%)
Hemiarthroplasty	7 (24.1%)	6 (15.4%)	5 (41.7%)
Time from index to PJI (days)	1235 (1222)	1380 (1681)	1426 (1816)	0.92
Type of revision				0.07
Single stage	6 (20.7%)	18 (46.2%)	2 (16.7%)
Two stage
Anatomic TSA	1 (3.45%)	2 (5.13%)	2 (16.7%)
Reverse TSA	6 (20.7%)	6 (15.4%)	5 (41.7%)
Hemiarthroplasty	1 (3.45%)	0	0
Permanent spacer	15 (51.7%)	13 (33.3%)	3 (25.0%)

PJI: definite or probable periprosthetic joint infection based on international consensus meeting criteria; BMI: body mass index; CCI: non-age-adjusted Charlson Comorbidity Index; ASA: American Society of Anesthesiologist score; TSA: total shoulder arthroplasty.

Aseptic versus PJI revision

Patients undergoing revision shoulder arthroplasty for PJI had significant higher erythrocyte sedimentation rate (ESR, 34.1 vs. 22.3, p = 0.02), C-reactive protein (CRP, 14.6 vs. 7.1, p = 0.02), and ICM scores (7.09 vs. 0.79, p < 0.01) compared to uninfected patients undergoing revision shoulder arthroplasty (Table 3). In addition, patients diagnosed with a PJI were more likely to have a sinus tract communication with the prosthesis (16.2% vs. 0.0%, p < 0.01), wound drainage (21.5% vs. 1.3%, p < 0.01), intraoperative purulence (43.8% vs. 0.0%, p < 0.01), and humeral component loosening (26.6% vs. 3.16%, p < 0.01) compared to non-infected patients (Table 3).

Table 3.

Comparison of clinical presentations of patients undergoing revision shoulder arthroplasty for a periprosthetic joint infection versus an aseptic clinical reason.

	Aseptic (n = 380)	PJI (n = 80)	p
ESR (mm/h)	22.3 (23.0)	34.1 (32.4)	0.02
CRP (mg/dL)	7.13 (13.8)	14.6 (20.5)	0.02
Sinus tract	0	13 (16.2%)	<0.01
Wound drainage	5 (1.32%)	17 (21.5%)	<0.01
Purulence	0	35 (43.8%)	<0.01
Humeral loosening	12 (3.16%)	21 (26.6%)	<0.01
ICM Score	0.79 (1.40)	7.09(3.23)	<0.01

Aseptic: unlikely or possible periprosthetic joint infection based on international consensus meeting criteria.

PJI: definite or probable periprosthetic joint infection based on international consensus meeting criteria; ESR: erythrocyte sedimentation rate; CRP: C-reactive Protein; ICM: international consensus meeting.

PJI analysis by organism virulence

The most common organism in the virulent PJI cohort was Staphylococcus aureus (n = 16, 55.2%) and the most common organism in the non-virulent PJI cohort was C. acnes (n = 32, 82.1%; Figure 2). On subgroup, analysis there was significant heterogeneity among the PJI subgroups (Table 4). Eleven (91.7%) patients with culture negative, 5 (12.8%) patients with non-virulent, and 26 (89.7%) patients with virulent PJIs had major criteria that defined the infection (p < 0.01). Preoperative CRP (p = 0.02), ESR (p = 0.05), sinus tract presence (p = 0.01), and intraoperative purulence (p < 0.01) were all significantly different between the cohorts. The total ICM criteria scores were also significantly different between the PJI cohorts (<0.01, Table 4).

Table 4.

Comparison of clinical presentations of patients undergoing revision shoulder arthroplasty for a periprosthetic joint infection based on the virulence attributed to the microorganism causing the infection.

	Virulent PJI (n = 29)		Non-virulent PJI (n = 39)	Culture-negative PJI (n = 12)	p
ESR (mm/h)		40.1 (28.9)	25.1 (28.3)	60.2 (49.9)	0.05
CRP (mg/dL)		20.3 (16.5)	7.16 (8.27)	28.0 (42.5)	0.02
Sinus tract		6 (20.7%)	2 (5.13%)	5 (41.7%)	0.01
Wound drainage		9 (31.0%)	5 (13.2%)	3 (25.0%)	0.22
Purulence		20 (69.0%)	4 (10.3%)	11 (91.7%)	<0.01
Humeral loosening		6 (20.7%)	11 (28.2%)	4 (36.4%)	0.56
ICM score		8.83 (3.48)	7.05 (2.06)	3.00 (1.81)	<0.01

PJI: definite or probable periprosthetic joint infection based on international consensus meeting criteria; ESR: erythrocyte sedimentation rate; CRP: C-reactive protein; ICM: international consensus meeting.

When individually comparing the subclassified PJI subgroups to the aseptic cohort, there was again a significant degree of heterogeneity between the cohorts. While the virulent PJI cohort differed significantly from the aseptic cohort in almost all ICM criteria, the non-virulent cohort only differed from the aseptic cohort in overt clinical signs of infection and culture positivity but essentially none of the “minor criteria” used in the ICM definition (Table 5).

Table 5.

Side-by-side comparison of the clinical presentations of patients undergoing revision shoulder arthroplasty for an aseptic clinical reason versus those undergoing revision arthroplasty for a presumed or definitive periprosthetic joint infection based on the organism attributed to the infection.

	Aseptic vs. virulent PJI (n = 29)	Aseptic vs. non-virulent PJI (n = 39)	Aseptic vs. culture- negative PJI (n = 12)
ESR (mm/h)	0.03	0.64	0.17
CRP (mg/dL)	0.01	0.99	0.24
Sinus tract	<0.01	0.01	<0.01
Wound drainage	<0.01	<0.01	<0.01
purulence	<0.01	<0.01	<0.01
Humeral loosening	<0.01	<0.01	<0.01
ICM score	<0.01	<0.01	<0.01

Aseptic: unlikely or possible periprosthetic joint infection based on international consensus meeting criteria.

PJI: definite or probable periprosthetic joint infection based on international consensus meeting criteria; ESR: erythrocyte sedimentation rate; CRP: C-reactive protein; ICM: international consensus meeting.

Discussion

PJIs of the shoulder represent a heterogeneous group of postoperative complications. Several publications have suggested that patients with shoulder PJIs attributed to virulent microorganisms present with a more “classic” inflammatory picture, whereas those patients with shoulder PJIs attributed to low virulence microorganisms tend to have a slightly more indolent or even aseptic presentation.^{3,4,6,9,12–14} Unfortunately, the ICM criteria were not available until 2018, and therefore the classification methodology utilized in the above-mentioned studies has led to inconsistent conclusions drawn about varying populations. ⁶ Therefore, the purpose of our study was to highlight the heterogeneity in patients diagnosed with shoulder PJIs based on the current ICM criteria utilizing organism virulence as a stratification tool. The results of our study were consistent with prior evaluations in that organism virulence impacts the clinical presentation associated with a given shoulder PJI. Those patients with shoulder PJIs attributed to a virulent microorganism or “no organism” were similar to one another, whereas those patients with shoulder PJIs attributed to low virulence microorganisms were more similar to their aseptic counterparts.

The organisms classified as virulent in our investigation caused overt signs of PJI, like that of a draining sinus or intraoperative purulence, over five times more frequently than those organisms classified as non-virulent. Similarly, serum biomarkers associated with acute inflammation, more specifically CRP, were also significantly higher in the virulent PJI cohort relative to the non-virulent PJI cohort. While not all virulent microorganisms are associated with the same pathophysiology, it is clear that the inflammatory reactions that occur as a consequence of bacterial invasion from organisms like S. aureus or Escherichia coli are significantly more dramatic than those reactions caused by organism like C. acnes or Staphylococcus epidermidis. Interestingly, although draining sinus tracts and purulence were significantly more common in the virulent PJI cohort, 13% of patients with non-virulent PJIs also presented with overt symptoms of infection. While our investigation clearly highlights the differences in clinical presentation between PJIs based on the virulence attributed to the organisms causing them, it also highlights the similarities between those patients with low virulence PJIs and those patients undergoing revision shoulder arthroplasty for an aseptic reason.

Defining the virulence of a given organism represents a difficult process for several reasons. While the orthopaedic and infectious disease communities generally view low virulence organisms as those associated with more indolent presentations, there is no binary definition of virulence, as it occurs on a spectrum of diseases. ¹⁸ In fact, virulence is defined as the relative capacity of a microorganism to cause damage to a host and ultimately reflects the ability of a specific organism to multiply within a specific environment. ¹⁹ Bacteria have evolved to become pathogens by acquiring genetic material encoding virulence determinants that can be turned on or off to change behavior dependent on the host environment. Variation in bacterial virulence relates to this ability.^19,20 S. aureus, an aggressive pathogen, harbors a wide array of virulence factors that facilitate tissue adhesion, immune evasion, and tissue injury via destructive toxin production, proinflammatory proteins and cell wall components that elicit host responses similar to the lipopolysaccharide toxin structure of virulent aerobic gram-negative bacteria.^21,22 In contrast, C. acnes and coagulase-negative staphylococci, both associated with indolent infection, produce fewer tissue-damaging exoenzymes and toxins while evading host defenses.^23,24 While S. aureus and other virulent pathogens do form implant-related biofilm, low-virulence organisms such as C. acnes and S. epidermidis may utilize biofilm formation as a primary virulence determinant. Biofilm formation downregulates chemokines and the inflammatory response and can lead to chronic infection without the promotion of acute, purulent inflammation. This dichotomy may reflect a passive defense strategy in low virulence bacteria rather than the range of inflammatory virulence factors associated with S. aureus.^21,24,25 Virulence attributed to an organism that is associated with a given PJI may also be dependent on the patient and inoculum size, which is likely why some patients have a greater risk of infection relative to others and why some PJIs attributed to low virulence organisms present with overt signs of infection. ²⁰ With that being said, the organisms in our study were classified on a binary basis and the associated clinical presentations that corresponded with them were relatively uniform. Consequently, PJIs that are caused by non-virulent organisms tended to be more indolent, relative to those PJIs attributed to their more virulent counterparts.

Shoulder PJIs caused by low virulence microorganisms clinically resemble that of their aseptic counterparts. Our analysis demonstrates that there is no significant difference between the cohorts with regard to systemic biomarkers and less than 15% of patients with non-virulent PJIs have overt signs of infection. Therefore, it is not surprising that Padegimas et al. reported an unexpected “positive” culture incidence of nearly 25% in the setting of revision shoulder arthroplasty. ¹² While it is largely accepted that a single positive culture in this setting does not indicate the presence of an infection, as over 20% of sterile operating room “air swabs” and 38% of primary shoulder arthroplasty swabs are culture positive, it does highlight the difficulty associated with making a PJI diagnosis in the setting of low virulence organisms.^26,27 As such, having a universally accepted criteria like that of the ICM not only provides treating physicians with a framework for making difficult decisions but also for performing consistent research to evaluate these decisions. Although it is imperative to diagnose shoulder PJIs via a standard set of criteria, it is equally important to isolate the organism(s) responsible for causing these complications due to the ramifications associated with having a culture-negative PJI.^{4,7,11,12,28,29}

Currently, bacterial culture represents the “gold standard” for organism identification in the setting of synovial fluid or tissue sampling. ³⁰ Nonetheless, several studies have highlighted the pitfalls associated with this technique, namely its lack of sensitivity.^11,31–37 In fact, several studies suggest that culture can fail to isolate an organism in up to 50% of PJI cases. In our study, culture-negative PJI represented 15% of all patients diagnosed with shoulder PJIs and 3% of all patients undergoing revision shoulder arthroplasty for an atraumatic reason. While this obviously represents an opportunity for more sensitive technology, it also highlights the importance of proper tissue sampling, as some studies demonstrate that reinfection rates are up to 4.5 times higher in culture-negative PJIs.^29,37,38

There are several limitations associated with this retrospective study design, including inability to assess any minor ICM criteria that were not documented. Another limitation of this study, although previously mentioned, was the binary classification of organisms as virulent or non-virulent regardless of the host. While this undoubtedly represents an issue with our evaluation, because there was no significant demographic difference between the cohorts, the potential impact of this issue may have been minimal. Finally, the influence of organism virulence on outcome of treatment of shoulder PJI was not assessed in this study.

Conclusion

In conclusion, shoulder PJI represents a heterogenous group of complications. While those infections caused by more “virulent” organisms like S. aureus or E. coli are typically associated with an inflammatory state, those PJIs associated with organisms considered to be “non-virulent,” like C. acnes or S. epidermidis, tend to be associated with a more indolent presentation. As such, the primary factor that differentiates a virulent shoulder PJI is culture positivity. Further host-specific prospective evaluations are necessary to validate our findings.