Human prosthetic joint infections are associated with myeloid-derived suppressor cells (MDSCs): implications for infection persistence

Abstract

Prosthetic joint infection (PJI) is a devastating complication of joint arthroplasty surgery typified by biofilm formation. Currently, mechanisms whereby biofilms persist and evade immune-mediated clearance in immune competent patients remain largely ill-defined. Therefore, the current study characterized leukocyte infiltrates and inflammatory mediator expression in tissues from patients with PJI compared to aseptic loosening. CD33⁺HLA-DR⁻CD66b⁺CD14^−/low granulocytic myeloid-derived suppressor cells (G-MDSCs) were the predominant leukocyte population at sites of human PJI compared to aseptic tissues. MDSCs inhibit T cell proliferation, which coincided with reduced T cells in PJIs compared to aseptic tissues. IL-10, IL-6, and CXCL1 were significantly elevated in PJI tissues and have been implicated in MDSC inhibitory activity, expansion, and recruitment, respectively, which may account for their preferential increase in PJIs. This bias towards G-MDSC accumulation during human PJI could account for the chronicity of these infections by preventing the pro-inflammatory, antimicrobial actions of immune effector cells.

INTRODUCTION

Joint replacement surgeries are often life-changing for millions of people each year, providing pain relief and restored function, which translates into improved mobility and quality of life. The number of total knee and total hip arthroplasty (TKA and THA, respectively) surgeries has continued to increase over the past decade, with numbers projected to reach 572,000 THAs and 3.48 million TKAs by 2030^{1; 2}. Although most surgeries are successful, some patients experience device failure that requires additional surgery. Reasons for aseptic failure include, loosening of the implant at the bone-cement interface, wear, implant malposition, dislocation, instability, or materials fatigue¹. Devices may also fail following prosthetic joint infection (PJI), a complication that can cause sustained disability and increased health care costs attributable to prolonged antibiotic treatment and multiple surgeries³. The average one-year incidence of PJI is estimated at 0.25–1% for primary THA and 0.4–2% for primary TKA^{4; 5}; however, the infection rate following revision surgery is even higher (i.e. 3.2–5.6% for both THA and TKAs)⁶. Currently, the gold-standard for PJI treatment is a two-stage exchange, consisting of irrigation and debridement with complete resection of the infected prosthetic and any surrounding cement before an antibiotic-impregnated cement spacer is inserted and a new device is placed 6–12 weeks later⁷. Unfortunately, this approach is associated with less than 50% likelihood of permanent infection eradication⁸, and new treatment strategies are needed.

Most PJIs are thought to occur by hardware contamination from skin microflora during surgical insertion or peri-operatively, despite extensive antiseptic precautions³. The underlying pathogenesis of PJI involves the formation of biofilm that not only protects the pathogen from the host immune response, but also makes the diagnosis and management of PJI with traditional antimicrobials and surgical debridement or prosthesis removal challenging⁹. Biofilms are a structured aggregate of bacterial cells encased in a self-produced matrix and are adherent to both biotic and abiotic surfaces¹⁰. Indeed, PJIs are often associated with chronic osteomyelitis, reflecting biofilm growth on a native surface^{3; 11}. The biofilm matrix, composed of exopolysaccharides, proteins, techoic acids, lipids and eDNA¹², forms a barrier that is relatively impervious to invasion by phagocytic cells^13–15, while planktonic bacteria that lack this protection are susceptible to phagocyte killing¹⁶. Additionally, it is thought that the existence of slow-growing or dormant cells within the biofilm (i.e. persisters), presence of bacterial subpopulations with different phenotypic levels of antibiotic resistance and gene expression, and stress responses to hostile environmental conditions also contribute to biofilm resistance^{16; 17}. Collectively, the reduced effectiveness of antibiotics and the ability to skew the host immune response toward an anti-inflammatory, pro-fibrotic state, contributes to the chronicity and recurrence of biofilm-related infections^{15; 18–22}.

Staphylococcus aureus (S. aureus) and coagulase-negative Staphylococci are the leading causes of PJI associated with THA and TKA²³. The significant disease burden associated with PJIs and increasing antimicrobial resistance among common PJI pathogens, including Staphylococci, highlights the importance of investigating alternative treatment paradigms to augment immune-mediated clearance. In the current study, we have compared leukocyte infiltrates and inflammatory mediator production profiles in tissues recovered from patients following aseptic revision TKA or THA as well as PJIs caused by distinct gram-positive pathogens. Here we report that CD33⁺HLA-DR⁻ myeloid-derived suppressor cells (MDSCs) are increased in PJI concomitant with reduced T cells relative to aseptic revisions, and have identified that the majority of these cells are CD66b⁺ granulocytic-MDSCs (G-MDSCs). In contrast, the primary granulocytic infiltrate into aseptic implant-associated tissues was CD66b⁺ neutrophils. Based on previous functional analyses in corresponding animal models^20–22, the increase in MDSCs associated with human PJI could contribute to the chronicity of implant-associated biofilm infections by preventing the proinflammatory, antimicrobial actions of immune effector cells.

MATERIALS AND METHODS

Periprosthetic tissue procurement

This study summarizes a Level 3 Case-control study. Informed consent was obtained from eligible patients undergoing arthroplasty revision surgery for infectious (PJI) or aseptic complications during their pre-surgical visit, following prior approval by the Institutional Review Board (IRB) of the University of Nebraska Medical Center (UNMC). Organisms associated with PJIs were identified by the Clinical Microbiology Laboratory at UNMC and samples that were negative for bacterial growth using standard protocols were classified as aseptic. Excess tissues were procured during the surgical procedure as previously described²². Because the approved IRB protocol was restricted to the collection of excess tissue only, information regarding the length of infection, comorbidities, and antibiotic treatment of patients was not available.

Flow cytometry

Immediately following excision, excess tissues were placed in isotonic saline and processed for FACS analysis as previously described²². Cells were stained with anti-human CD66b-FITC (BioLegend, San Diego, CA), CD14-PE, CD16-APC, CD33-PE-Cy5, HLA-DR-PE-Cy7, CD45-eFluor450, and CD3ε-APC-Cy7 (BD Biosciences, San Diego, CA and eBioscience). Dead cells were excluded using a Live/Dead Fixable Blue Dead Cell Stain Kit (Life Technologies, Eugene, OR) according to the manufacturer’s instructions. Analysis was performed using BD FACSDiva software. Single cells were gated from the total events using FSC-A vs. FSC-H, followed by exclusion of dead cells. Live, CD45⁺ leukocytes were separated into granulocytes or non-granulocytes based on SSC-A^highCD66b⁺. Non-granulocyte populations were identified using a series of gates to avoid duplicate counting. First, CD3⁺ T cells were gated, followed by HLA-DR⁺ cells from the non-T cell population to determine the abundance of monocyte populations expressing CD14 and CD16. Although CD11b is often used to identify MDSC populations, our goal was to broadly identify T cells, MDSCs, neutrophils, and monocytes, all of which express CD11b, making this marker less informative to discriminate between these cell types. In addition, both G-MDSCs and neutrophils express CD11b, which does not facilitate the delineation of these populations. Results are presented as the percentage of live cells, unless otherwise noted.

Multianalyte microbead array

Inflammatory mediator expression in aseptic versus PJI tissues was assessed using a human 41-plex microbead array (MILLIPLEX; Millipore, Billerica, MA) according to the manufacturer’s instructions, which included: EGF, FGF-2, Flt-3L, G-CSF, GM-CSF, IFN-α2, IFN-γ, IL-1α, IL-1β, IL-1rα, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12p40, IL-12p70, IL-13, IL-15, IL-17, CX3CL1, CCL2, CCL3, CCL4, CCL5, CCL7, CCL11, CCL22, CXCL1, CXCL10, sCD40L, TGF-α, TNF-α, TNF-β, PDGF-AA, PDGF AB BB, and VEGF. Results were analyzed using a Bio-Plex Workstation (Bio-Rad, Hercules, CA) and normalized to total protein to account for differences in tissue size. The level of sensitivity for most analytes was 3.2 pg/ml. When reporting values, the number of samples presented for some cytokines varied because they fell below the limit of detection. When this occurred, samples were omitted rather than reporting them as zeroes, which could artificially bias the results. No other data were excluded from analysis to reflect the biological variability between patient specimens.

ELISA

IL-10 levels in tissue homogenates associated with aseptic revisions or PJI were quantitated by a sandwich ELISA (BioLegend, San Diego, CA, USA). Results were normalized to the total amount of protein to account for differences in tissue sampling size. The lower limit of detection for the assay was 3.9 pg/ml.

Statistics

Significant differences between aseptic and infected samples were determined by an unpaired two-tailed Student t-test using GraphPad Prism version 6 (La Jolla, CA). For all analyses, p < 0.05 was considered statistically significant.

RESULTS

PJI is a debilitating complication of arthroplasty and as the number of arthroplasties continues to increase annually, it is expected that the infection rate will concur^{1; 9}. To determine whether PJIs are biased towards immunosuppressive responses to account for infection persistence in immune competent patients, leukocyte infiltrates and inflammatory mediator expression profiles were compared in patients with PJI versus aseptic revisions. The majority (~70%) of tissues obtained for this study were from patients undergoing revision surgery for presumed aseptic complications, as determined during a pre-operative visit (Table 1). The remaining 30% of consented patients presented with suspected infection associated with the implanted device, which was confirmed following post-operative culture. Of the 9 infected samples, 6 were Staphylococcal species (methicillin-resistant and –susceptible S. aureus, coagulase-negative Staphylococcus, and S. lugdunensis), 2 were Streptococci (Groups B and C) and one was attributed to Actinobacteria (Corynebacterium diptheriae).

Table 1.

Summary of aseptic and infected tissues analyzed in this study from patients undergoing total knee or total hip arthroplasty revision surgery.

Age	Gender	Ethnicity	Revision Type	Diagnosis/Reason for Revision Surgery	Pathogen
55	M	Caucasian	Hip	Aseptic; Mechanical failure
74	M	Caucasian	Hip	Aseptic; Mechanical loosening
61	F	Caucasian	Hip	Aseptic; Mechanical loosening
58	F	Caucasian	Knee	Aseptic; Failed device
61	F	African-American	Hip	Aseptic; Failed device
73	F	Caucasian	Hip	Infected	S. lugdunensis
73	F	Caucasian	Hip	Aseptic; Mechanical loosening
74	M	Caucasian	Knee	Aseptic; Mechanical loosening
83	F	Caucasian	Hip	Aseptic; Mechanical loosening
56	F	Caucasian	Knee	Aseptic; Failed device
86	F	Caucasian	Hip	Infected	Group B Streptococcus
79	F	Caucasian	Knee	Infected	C. diptheriae
54	F	Caucasian	Knee	Infected	Coagulase-negative Staphylococcus
58	F	Caucasian	Knee	Infected	S. aureus
NA*	NA	NA	Knee	Infected	Group C Streptococcus
76	F	Caucasian	Knee	Aseptic; Mechanical loosening
67	M	Caucasian	Knee	Aseptic; Knee pain, stiffness
64	F	Caucasian	Knee	Aseptic; Mechanical loosening, stiffness
61	M	Caucasian	Knee	Infected	Methicillin-resistant S. aureus
64	M	Caucasian	Knee	Aseptic; Mechanical loosening
61	M	African-American	Hip	Aseptic; Mechanical loosening
80	M	Caucasian	Hip	Aseptic; Mechanical loosening
64	M	Caucasian	Knee	Infected	Methicillin-susceptible S. aureus
51	M	Caucasian	Knee	Aseptic; Mechanical loosening
75	M	Caucasian	Knee	Aseptic; Instability
80	M	Caucasian	Knee	Infected	Coagulase-negative Staphylococcus
76	M	Caucasian	Knee	Aseptic; Mechanical loosening
63	F	Caucasian	Knee	Aseptic; Mechanical loosening
67	M	Caucasian	Knee	Aseptic; Mechanical loosening

^*NA; not available

Representative images of aseptic and infected tissues recovered during arthroplasty revision surgery demonstrate the disparities in gross appearance of these two cohorts (Fig 1A). There were no differences in the percentages of live cells recovered from either aseptic or infected explanted tissues; however, CD45⁺ leukocyte infiltrates were significantly increased in PJIs, indicating their active accumulation in response to signals at the site of biofilm infection (Fig 1B). MDSC infiltrates (CD33⁺HLA-DR⁻) were significantly increased during PJI compared to aseptic tissues (Fig 2A). In humans, two MDSC subsets have been described, namely CD33⁺HLA-DR⁻CD66b⁻CD14⁺ monocytic-MDSCs (M-MDSCs) and CD33⁺HLA-DR⁻CD66b⁺CD14^−/low granulocytic-MDSCs (G-MDSCs)²⁴. This distinction between G- and M-MDSCs was observed in both infected and aseptic tissues isolated from knee and hip prostheses (Fig 2B and C, respectively). CD33⁺HLA-DR⁻ MDSCs (Fig 2A) were gated using CD66b and CD14, resulting in two distinct cell populations represented by CD66b⁺CD14^low G-MDSCs (Fig 2B) and CD66b⁻CD14⁺ M-MDSCs (Fig 2C). Quantitation of these populations expressed as a percentage of MDSCs revealed a significant increase in G-MDSC infiltrates in PJI tissues compared to aseptic (Fig 2B), whereas no differences in M-MDSCs were observed (Fig 2C). After gating on CD66b⁺SSC-A^high cells from the total live population, the CD33⁺HLA-DR⁻ MDSC population was excluded, resulting in the percentage of non-MDSC granulocytes, or neutrophils, in implant-associated tissues. This strategy revealed that the percentage of neutrophils was not different in PJI patients compared to aseptic revisions (Fig 2D).

Figure 1. CD45⁺ leukocyte infiltrates are increased in response to PJI.

Tissues surrounding human aseptic and infected knee and hip prostheses were analyzed by flow cytometry. (A) Gross appearance of tissue from a single aseptic revision and a coagulase-negative Staphylococcal-infected knee prostheses, which are representative of each tissue collected for this study. (B) Quantitation of live cells recovered from implant-associated tissues, and the percentage of those cells expressing CD45 are presented as the mean ± SEM of all tissues examined (* p < 0.05; unpaired Student t-test).

Figure 2. Tissues from PJIs display increased numbers of CD33⁺HLA-DR⁻ G-MDSCs.

Tissues surrounding human aseptic and infected knee and hip prostheses were analyzed by flow cytometry. Results were calculated after gating on live cells and contour plots depict an individual aseptic and infected sample that are representative of all tissues collected. Quantitation of (A) CD33⁺HLA-DR⁻ MDSCs, (B) CD66b⁺ G-MDSCs, (C) CD14⁺ M-MDSCs, and (D) neutrophils of total CD66b⁺SSC-A^high non-MDSC granulocytes that represent the mean ± SEM of all tissues collected (** p < 0.01, *** p < 0.001; **** p < 0.0001; unpaired Student t-test).

MDSCs are recognized for their suppression of T cell proliferation and activation, in part, through arginine depletion from the extracellular milieu^{25; 26}. Since G-MDSCs were more abundant in PJIs, we next determined whether this would translate to reduced T cell numbers at the site of infection. Approximately 15% of live cells isolated from aseptic tissues were CD3⁺CD45⁺; and while T cell infiltrates were decreased in infected tissues, this did not reach statistical significance (Fig 3). Unfortunately, the number of MDSCs recovered from implant-associated tissues was not sufficient to perform proliferation assays and assess suppressive properties.

Figure 3. PJI tissues have reduced T cell infiltrates compared to aseptic revisions.

Tissues collected during knee and hip arthroplasty revisions were analyzed by flow cytometry and gated on live cells. (A) Contour plots of CD3 and CD45 staining are shown from an individual aseptic and infected sample that are representative of all tissues collected. (B) Quantitation of CD3⁺CD45⁺ cells that represent the mean ± SEM of all tissues collected.

Distinct monocyte populations can be identified in human blood, which migrate to sites of infection in response to chemokine signals; however, the frequency of monocyte infiltrates in human periprosthetic tissues has not yet been examined. Human monocytes can be divided into three populations based on their relative surface expression of the LPS co-receptor CD14 and the FCγIII receptor CD16²⁷. Classical monocytes (CD14⁺CD16⁻) account for the majority of the total monocyte population, while CD14⁺CD16⁺ intermediate monocytes and CD14⁻CD16⁺ non-classical monocytes are less abundant²⁷. To identify these different monocyte subsets associated with human arthroplasty revisions, we gated on HLA-DR⁺ cells that were non-granulocytes and non-T cells (Fig 4A) before differentiation of CD14 and CD16 surface expression. The exclusion of both cell types was important, as T cell populations have been reported to express HLA-DR²⁸, and granulocytes express CD16²⁹. This strategy allowed us to specifically identify monocyte subsets in implant-associated tissues, where no differences were observed in either the percentage of HLA-DR⁺ cells infiltrating aseptic and infected tissues (Fig 4A) or classical, intermediate, or non-classical monocyte infiltrates during PJI (Fig 4B, C, and D, respectively).

Figure 4. MHC class II⁺ cells and monocyte infiltrates are similar between aseptic and PJI-associated tissues.

Soft tissues collected during revisions of knee and hip arthroplasties were analyzed by flow cytometry. Results were calculated after gating on live cells and removing granulocyte and T cell populations from the analysis. Contour plots of HLA-DR vs. FSC-A, followed by CD14 vs. CD16 expression of HLA-DR⁺ cells are shown from an individual aseptic and infected sample that are representative of all tissues analyzed. (A) HLA-DR⁺ cells, (B) CD14⁺CD16⁻ classical monocytes, (C) CD14⁺CD16⁺ intermediate monocytes and (D) CD14^lowCD16⁺ non-classical monocytes presented as the mean ± SEM of all tissues collected.

To better understand the inflammatory milieu that may dictate patterns of leukocyte invasion in aseptic versus PJI tissues, we evaluated clarified supernatants from homogenized explanted tissues for a panel of 41 human cytokines and chemokines. Of the mediators examined, only IL-10, IL-6 and CXCL1 were significantly increased in PJI tissues compared to aseptic revisions (Fig 5A–C, respectively). However, CXCL8 (Fig 5D), CCL7 (Fig 5E), and TGF-α (Fig 5F) were slightly elevated during infection, whereas VEGF (Fig 5G), and Flt-3L (Fig 5H) were lower than in aseptic tissues. Differential IL-10, IL-6, and CXCL1 expression and G-MDSC infiltrates between PJI and aseptic tissues may prove to be useful to diagnose human PJIs and assess responsiveness to treatment.

Figure 5. Differential cytokine and chemokine expression is observed between aseptic and PJI tissues.

Quantitation of (A) IL-10 (B) IL-6, (C) CXCL1, (D) CXCL8, (E) CCL7, (F) TGF-α, (G)VEGF, and (H) Flt-3L expression in aseptic and PJI tissues. Results were normalized to the amount of total protein to correct for differences in tissue size and are expressed as the mean ± SEM of all tissues examined (** p < 0.01; unpaired Student t-test).

DISCUSSION

S. aureus and coagulase-negative staphylococci are the primary etiologic agents of human PJIs, accounting for approximately 50–60% of cases¹. Indeed, of the nine PJIs examined in our study, six were Staphylococcal species (~66%, Table 1). The mode of infection is not known in many cases; however, the majority of PJIs are thought to occur during surgery, originating from skin commensals³. The anterior nares of 20–30% of the general population and 2–6% of orthopedic patients in the US are colonized by methicillin-susceptible and methicillin-resistant S. aureus (MSSA and MRSA, respectively)⁵, and preoperative colonization by S. aureus strains is associated with a 2–9 times greater risk of surgical site infection during orthopedic surgery compared to patients who are not colonized^{30; 31}. This has resulted in increased screening for S. aureus carriage to ensure that patients undergo decolonization regimens before their surgeries in an attempt to minimize infection risk³². While S. aureus is recognized as a common cause of healthcare-associated infections, other less virulent coagulase-negative species are also implicated in PJI. S. epidermidis is the most frequently identified member of this group and causes PJI primarily through its ability to adhere to prosthetic materials and form a biofilm^{33; 34}. S. lugdunensis was recovered from one PJI in this study (Table 1), and is another coagulase-negative staphylococcal species capable of eliciting severe systemic and local infections similar to those caused by S. aureus, including PJI³⁵. Gram-positive cocci are responsible for the majority of hip and knee PJIs; and although largely attributed to staphylococci, streptococci account for < 10% of cases¹. In the current study, Group B and C Streptococcus were identified in two out of the nine PJIs examined (Table 1). Streptococcal infections are typically acute in nature, and most patients have one or more comorbidities that may have contributed to joint arthroplasty infection³⁶. Other anaerobic bacteria have been reported in PJIs, including Corynebacterium diptheriae, a member of Actinomyces known to cause monomicrobial infections. However, these occurrences are rare as most anaerobes are present as part of a polymicrobial infection¹.

Since the immune status of PJIs remains relatively underexplored, the objective of the current study was to identify differences in immune cell infiltrates and inflammatory mediator production associated with aseptic or infected explanted tissues from human subjects following post-arthroplasty revision surgery. A secondary outcome was to substantiate changes in human specimens with recent findings in our mouse model of S. aureus orthopedic biofilm infection^20–22. This report is the first to identify the preferential accumulation of CD33⁺HLA-DR⁻ G-MDSCs in human PJI tissues relative to aseptic samples. This affords translational relevance to our recent findings in a mouse model of S. aureus orthopedic implant infection, where MDSCs represented the primary infiltrate and played a critical role in transforming the local milieu toward an anti-inflammatory environment that favored biofilm persistence^20–22. Subsequent to our initial reports, others have documented MDSC infiltrates in mouse models of S. aureus infection^{37; 38}. Neutrophils are a critical anti-bacterial effector population; however, in human PJIs neutrophil influx was unchanged compared to aseptic tissues. The failure to augment neutrophil accumulation during infection is likely attributed to the expanded MDSC population associated with PJIs that are CD66b⁺ G-MDSCs, and thus could be neutrophil precursors that have become arrested in an immature, immunosuppressive state. The exact combination and sequence of infection-derived factors that regulate MDSC mobilization, proliferation, and differentiation are not understood; however, it appears that in both humans and mice an infection-induced signal arrests granulocyte precursors in an immature G-MDSC state.

MDSCs are known for their ability to augment arginase-1 (Arg-1) expression^{25; 26} and arginine depletion from the environment by Arg-1 results in defective TCR signaling through inhibiting CD3ζ expression, cell cycle, and cytokine production²⁵. In the current study, a trend toward reduced CD3⁺ T cells was associated with PJI compared to aseptic revisions, which agrees with the preferential accumulation of G-MDSCs in PJI tissues. Previously, we have found few T cell infiltrates associated with different mouse models of S. aureus biofilm infection^{13; 22; 39}, which corroborates the findings in human PJIs. However, in this study, limited cell numbers prevented analysis of G-MDSC suppressive properties.

Interestingly, there was no enhanced recruitment of HLA-DR⁺ cells or apparent upregulation of HLA-DR expression in response to PJI. HLA-DR is a major histocompatibility complex (MHC) class II surface receptor that is critical for initiating a signaling cascade to activate adaptive immunity. In the current study, no differences in classical, intermediate, or non-classical monocyte populations were observed between human aseptic and infected post-arthroplasty revision tissues. The lack of HLA-DR up-regulation and monocyte recruitment to human PJIs concomitant with increased MDSCs suggests that signals originating from the site of infection induce an immunosuppressive environment that prevents sufficient activation of antimicrobial effector mechanisms, thereby promoting bacterial persistence. This is supported by the observation that IL-10 levels are significantly increased in patents with PJI compared to aseptic loosening. IL-10 is an anti-inflammatory cytokine known for its role in controlling inflammatory responses, including inhibiting T cell activation and programming macrophages toward an anti-inflammatory phenotype^{40; 41}. Recent studies from our laboratory with a mouse model of S. aureus orthopedic implant infection have shown that IL-10 production by MDSCs is one mechanism used to promote biofilm persistence²¹. These observations align with elevated IL-10 expression in human PJI; however, it remains unclear how much IL-10 is produced by MDSCs versus the other cell types present during infection. Although IL-6 is significantly increased during human PJI and is classically considered an inflammatory cytokine, it can also induce MDSC expansion^{26; 42}. Furthermore, CXCL1 was significantly increased in human PJI tissues and has been reported to induce MDSC recruitment⁴³, suggesting that it could play a role in G-MDSC influx into implant-associated tissues.

It is important to note that despite the variety of bacterial species responsible for PJI that were identified in this study, the overall immune profile of infected implant-associated tissues was similar. Namely, G-MDSC infiltrates were increased in all infected samples, whereas no changes in neutrophils or monocyte populations were observed. Based on previous functional analyses in corresponding animal models^20–22, the increase in MDSCs could be a mechanism to control the initial inflammatory response to bacteria and inadvertently set the stage for biofilms to form and persist on these devices for protracted periods.

In conclusion, this analysis of tissues from patients undergoing revision surgeries for PJI has revealed the preferential accumulation of inhibitory G-MDSCs concomitant with the failure to augment phagocytic effector cells (i.e. neutrophils and monocytes) compared to aseptic revisions. Elucidating the mechanisms whereby bacterial biofilms evade protective immunity could lead to novel immune-mediated approaches to facilitate PJI clearance in combination with conventional antibiotics. This becomes increasingly important as the rates of joint replacement surgery continue to rise along with the incidence of infection and the economic burden placed on both patients and the healthcare system.

Clinical Significance.

Animal models of PJI have revealed a critical role for MDSCs and IL-10 in promoting infection persistence; however, whether this population is prevalent during human PJI and across distinct bacterial pathogens remains unknown. This study has identified that granulocytic-MDSC infiltrates are unique to human PJIs caused by distinct bacteria, which are not associated with aseptic loosening of prosthetic joints. Better defining the immune status of human PJIs could lead to novel immune-mediated approaches to facilitate PJI clearance in combination with conventional antibiotics.