Lack of Humoral Immunity Against Glucosaminidase Is Associated with Postoperative Complications in Staphylococcus aureus Osteomyelitis

Abstract

Background:

Glucosaminidase (Gmd) is known to be a protective antigen in animal models of Staphylococcus aureus osteomyelitis. We compared the endogenous anti-Gmd antibody levels in sera of patients with culture-confirmed S. aureus bone infections to their sera at 1 year after operative treatment of the infection.

Methods:

A novel global biospecimen registry of 297 patients with deep-wound culture-confirmed S. aureus osteomyelitis was analyzed to assess relationships between baseline anti-Gmd serum titers (via custom Luminex assay), known host risk factors for infection, and 1-year postoperative clinical outcomes (e.g., infection control, inconclusive, refracture, persistent infection, septic nonunion, amputation, and septic death).

Results:

All patients had measurable humoral immunity against some S. aureus antigens, but only 20 patients (6.7%; p < 0.0001) had high levels of anti-Gmd antibodies (>10 ng/mL) in serum at baseline. A subset of 194 patients (65.3%) who completed 1 year of follow-up was divided into groups based on anti-Gmd level: low (<1 ng/mL, 54 patients; 27.8%), intermediate (<10 ng/mL, 122 patients; 62.9%), and high (>10 ng/mL, 18 patients; 9.3%), and infection control rates were 40.7%, 50.0%, and 66.7%, respectively. The incidence of adverse outcomes in these groups was 33.3%, 16.4%, and 11.1%, respectively. Assessing anti-Gmd level as a continuous variable showed a 60% reduction in adverse-event odds (p = 0.04) for every tenfold increase in concentration. No differences in patient demographics, body mass index of >40 kg/m², diabetes status, age of ≥70 years, male sex, Charlson Comorbidity Index of >1, or Cierny-Mader host type were observed between groups, and these risk factors were not associated with adverse events. Patients with low anti-Gmd titer demonstrated a significant 2.68-fold increased odds of adverse outcomes (p = 0.008).

Conclusions:

Deficiency in circulating anti-Gmd antibodies was associated serious adverse outcomes following operative treatment of S. aureus osteomyelitis. At 1 year, high levels of anti-Gmd antibodies were associated with a nearly 3-fold increase in infection-control odds. Additional prospective studies clarifying Gmd immunization for osteomyelitis are needed.

Level of Evidence:

Prognostic Level IV. See Instructions for Authors for a complete description of levels of evidence.

Osteomyelitis remains the bane of orthopaedic surgery, and a great need exists for novel interventions¹. Osteomyelitis can affect otherwise healthy hosts as well as those who are less healthy. The vast majority of severe cases involve Staphylococcus aureus², primarily methicillin-resistant S. aureus in some regions³, and strains with pan-resistance are emerging⁴. An urgent need exists for non-antibiotic, immune-based approaches to treat resistant infections, as antibiotic resistance is a serious public-health threat⁵. Infection rates following total joint replacement and trauma surgical procedures have remained largely unchanged over the last 50 years¹. Adherence to rigorous prophylactic and surgical protocols (e.g., Surgical Care Improvement Project⁶) cannot reduce infection rates for elective procedures below 1% to 2%⁷. These findings strongly suggest that host factors play an essential role in orthopaedic infections. S. aureus bone infections are caused by pathogenic mechanisms that have developed to achieve immune evasion^8,9. Mechanisms include (1) biofilm formation on the implant¹⁰ and necrotic bone^11,12, (2) generation of staphylococcal abscess communities in soft tissues and bone marrow^13-15, (3) intracellular infection⁸, and (4) ability to colonize the osteocytic-canalicular network of live cortical bone^16,17. Persistence of infection following surgical treatment of S. aureus osteomyelitis is common (15% to 40%) and often requires multiple surgeries^18-22.

To date, 19 S. aureus immunizations have been evaluated in U.S. Food and Drug Administration registration trials, and all failed to demonstrate efficacy^5,23. Expert opinion on the root cause of these failures has focused on the inability to predict the protective role of staphylococcal immune responses in humans based on animal data^5,23. Thus, we departed from this traditional vaccine development approach and aimed to develop an immunotherapy that was based on monoclonal antibodies that have dual-acting mechanisms: (1) direct inhibition of critical S. aureus enzymes, and (2) immunomodulatory activity to stimulate host response and bacterial clearance^14,24. Utilizing a murine tibial osteomyelitis model with an infected transosseous pin that faithfully recapitulates the salient features of implant-associated osteomyelitis²⁵, we identified the glucosaminidase (Gmd) protein subunit of S. aureus autolysin as our lead target^14,15,24,26. An important validation of this discovery was that other groups also identified autolysin as an immunodominant antigen^27-29, including studies of tibial osteomyelitis in a rabbit model²⁸. Autolysin is critical for cell-wall biosynthesis and degradation during binary fission^30-32 and functions as an adhesin³³, a biofilm enzyme³⁴, and a facilitator of host-cellular internalization and immune evasion³⁵. Most importantly, anti-Gmd passive immunization has been shown to synergize with vancomycin therapy in rabbit and murine tibial osteomyelitis and murine peritoneal infection models^15,28,36. Our clinical studies of patients with osteomyelitis from periprosthetic joint infection, trauma, and diabetic foot ulcers have found anti-Gmd antibodies in patients who have recovered from these serious infections^26,37,38.

Recently, we reported on the safety and pharmacokinetics of a candidate anti-Gmd monoclonal antibody (1C11) passive immunization in a sheep model³⁹. In addition to reporting the favorable profile of 1C11, we described the behavior of endogenous human anti-Gmd antibodies analyzed from sera collected in a unique biospecimen registry (AO Trauma Clinical Priority Program [CPP] Bone Infection Registry) of 297 patients with culture-confirmed S. aureus osteomyelitis⁴⁰. The results of that study demonstrated that anti-Gmd antibody levels ranged from undetectable (<1 ng/mL) to 300 μg/mL, and the mean concentration was 21.7 μg/mL³⁹. We estimated that the circulating half-life of endogenous anti-Gmd antibodies was 120.4 days, which is roughly equivalent to ~4 passive immunizations over a 1-year course of treatment. However, as critical questions regarding the relationships between endogenous anti-Gmd antibodies in these patients and the clinical outcome following standard-of-care surgical and postoperative treatment remained open, we performed a post-hoc analysis of the AO Trauma CPP Bone Infection Registry to test 3 hypotheses: (1) most osteomyelitis patients with normal immune status (commonly referred to as “type-A hosts”⁴¹) do not have high levels of anti-Gmd antibodies (>10 ng/mL) in their serum, (2) osteomyelitis patients with low levels of anti-Gmd antibodies (<1 ng/mL) at the time of the surgical procedure have an increased risk of an adverse event during the first postoperative year, and (3) high levels of anti-Gmd antibody are associated with a high rate of “infection controlled” outcomes at 1 year postoperatively. In the present study, we describe our biostatistical analyses of the AO Trauma CPP Bone Infection Registry data and the results supporting the aforementioned hypotheses. To our knowledge, this is the first clinical evidence directly associating the humoral immune response of a patient with osteomyelitis against a specific S. aureus bacterial surface antigen with clinical outcomes following surgical treatment of the infection.

Materials and Methods

AO Trauma CPP Bone Infection Registry

This study was part of an international, prospective, observational case series of patients with long-bone S. aureus infection, conducted according to ISO (International Organization for Standardization) 9001 guidelines and registered at ClinicalTrial.gov (NCT01677000). Details on patient enrollment, data and sample collection, clinical outcome measures, and end points have previously been published⁴⁰; additional results are pending publication. In brief, patients ≥18 years old with culture-confirmed S. aureus infection of a long bone (i.e., femur, tibia, fibula, humerus, radius, ulna, or clavicle) following fracture fixation, arthroplasty, or injury were enrolled into the registry prior to the surgical treatment (i.e., debridement or 1-stage or 2-stage revision). Patient demographic data (i.e., age, sex, race, place of residence at admission, and body mass index [BMI]) and medical information (i.e., comorbidities and prior treatment) were collected prior to the surgical procedure (baseline). Patients were categorized as A, B, or C-type hosts according to the methodology previously described by Lazzarini et al.⁴¹. Charlson Comorbidity Index scores were also calculated for each patient⁴².

Serology

Although clinical blood-laboratory data were collected in the registry at baseline, 6 months, and 12 months postoperatively⁴⁰, the present study only analyzed data on baseline immunoglobulin G (IgG) antibodies against 8 immunodominant S. aureus antigens (Gmd, aminidase [Amd], iron scavenging determinant A [IsdA], iron scavenging determinant B [IsdB], iron scavenging determinant H [IsdH], chemotaxis inhibitory protein of staphylococci [CHIPS], staphylococcal complement inhibitor [SCIN], and staphylococcal hemolysin [Hla]) with use of a custom multiplex Luminex assay that reports titers as median fluorescent intensity (MFI) in arbitrary units, as previously described^37,38. Quantification of antibody concentration (ng/mL) was only performed for anti-Gmd in the baseline sera, determined by interpolation to a standard curve generated with mouse:human chimeric 1C11 anti-Gmd monoclonal antibodies as previously described⁴³. This analysis determined that the lower limit of detection of the custom Luminex assay is 1 ng/mL⁴³. We considered a tenfold increase over this limit of detection (10 ng/mL) to be the high threshold of anti-Gmd antibodies in human serum. Using thresholds for the assay level of detection (MFI = 1,550, 1 ng/mL)⁴³ and using a 10 ng/mL high level of circulating anti-Gmd antibody (MFI = 9,000) as prospectively defined⁴³, we stratified the cohort into 3 anti-Gmd groups: low (MFI <1,550), intermediate (MFI = 1,550 to 9,000), and high (MFI >9,000). All patients were studied who had informed consent and complete baseline demographic and serology data (297 patients, including 292 from the AO Registry⁴⁰ and 5 who were excluded from the AO Trauma CPP Bone Infection Registry because they did not have a final positive wound culture for S. aureus, 297 total). The inclusion criteria were age of ≥18 years and deep culture-confirmed S. aureus infection (methicillin-resistant or sensitive). Infections involved a long bone (i.e., femur, tibia, fibula, humerus, radius, ulna, clavicle) and followed fracture fixation or arthroplasty. Prisoners, pregnant patients, patients <18 years old, patients with a culture negative for S. aureus, and those unwilling or unable to consent were excluded.

Clinical Outcomes

Although a diverse array of clinical, patient-reported, and functional outcome data were collected in the registry⁴⁰, the present study focused only on the 1-year clinical outcome of the baseline surgical procedure, which was categorized as either “infection control,” “adverse outcome,” or “inconclusive.” “Infection control” was determined according to a note from the treating surgeon at the 1-year postoperative visit, which specifically cited “infection control” and/or the lack of any signs or symptoms of infection at that time. “Adverse outcome” was defined prior to study initiation as documentation of refracture or infection related to the initial surgical procedure at 1 year, septic nonunion, amputation, or septic death. Outcomes were recorded by the treating surgeons, who were blinded to antibody levels. Patients who had a complete data set for the 1-year outcome but could not be categorized as “infection control” or “adverse outcome” were defined as “inconclusive.” Patients who did not complete the study or had insufficient documentation of outcomes at 1 year postoperatively were excluded (103 patients; 34.7%).

Statistics

Exact binomial tests were utilized to compare relative frequencies of anti-Gmd antibody titer groups. Fisher exact tests and Cochran-Armitage exact trend tests were utilized to compare event rates and categorical characteristics across these groups. Continuous variables were compared across groups with use of Wilcoxon rank-sum tests. The odds of adverse events and controlled infections was modeled with use of univariate logistic regression, with goodness-of-fit assessed with use of the Hosmer-Lemeshow test. Odds ratios (ORs), 95% confidence intervals (CIs), and p values were calculated for each risk factor. When modeled as a continuous predictor, a log-transformation was applied to anti-Gmd antibody titer values to reduce skewness and improve model fit. Predicted probabilities of adverse events and controlled infections were calculated with use of these logistic models. Analyses were conducted with use of SAS (version 9.4; SAS Institute). Significance was set at p = 0.05.

Twelve patients were missing Charlson Comorbidity Index data, 1 was missing BMI data, and 28 were missing data regarding diabetes status. The n value was reduced accordingly when performing statistical analyses.

Results

The anti-Gmd antibody titers in baseline sera from 297 patients with osteomyelitis are presented in Figure 1-A and Appendix Supplemental Table 1. Consistent with the asymmetric graphical illustration of anti-Gmd titers in this cohort, an exact binomial test confirmed that the number of patients with high levels of anti-Gmd antibodies in their serum at the time of the baseline surgical procedure was significantly low (20 of 297; 6.7%; p < 0.0001). This result was not the result of immunosuppression, as all patients had detectable titers against at least 1 of the S. aureus antigens tested (Fig. 1-B).

Fig. 1.

Figs. 1-A and 1-B Anti-S. aureus antibody IgG titers in patients with orthopaedic infections. The data in Figure 1-A are of all 297 patients. Antibody titers are represented as the MFI. Fig. 1-A The anti-Gmd IgG titers are presented in rank, ordered from lowest to highest, and are stratified into 3 groups: undetectable or low anti-Gmd (MFI <1,550, cyan), intermediate anti-Gmd (MFI 1,550 to 9,000, yellow), and high anti-Gmd (MFI >9,000, red). Note that 29% of osteomyelitis patients have undetectable levels of anti-Gmd titers. Fig. 1-B Heat maps of serum IgG levels of antibodies against the 8 antigens are presented for the 194 patients who completed the study out to 1 year postoperatively. Baseline and available 6-month and 12-month measurements illustrate that detectable anti-S. aureus antibodies against at least 1 antigen were evident over the course of therapy. Also of note is that IsdB is by far the most immunogenic antigen among these patients, whereas Gmd is one of the lowest (IsdB > Amd > SCIN > IsdA > Hla > IsdH = CHIPS = Gmd).

A subset of 194 patients with 1 year of follow-up (65.3% of the registry) was divided into the 3 anti-Gmd antibody-level groups defined in Figure 1-A: low (undetectable, 54 patients; 27.8%), intermediate (<10 ng/mL, 122 patients; 62.9%), and high (>10 ng/mL, 18 patients; 9.3%) (Table I). Although no specific outcomes were significantly associated with anti-Gmd titers, the incidences of adverse outcomes were 33.3%, 16.4%, and 11.1%, respectively (trend p = 0.010). Consistently, the observed infection control rates were 40.7%, 50.0%, and 66.7%, respectively. A logistic regression analysis comparing the high-antibody and low-antibody groups showed that high levels of anti-Gmd antibodies were associated with a nearly 3-fold increase in infection-control odds at 1 year postoperatively (see Appendix Supplemental Figure 1; OR, 2.91; 95% CI, 0.95 to 8.92; p = 0.06). Additionally, by assessing anti-Gmd levels as a continuous variable on a logarithmic scale, we found that for every tenfold increase in Gmd antibody concentration, there was a significant 60% reduction in adverse event odds (see Appendix Supplemental Figure 2; OR, 0.40; 95% CI, 0.17 to 0.96; p = 0.04).

TABLE I.

Clinical Outcome Versus Anti-Gmd Antibody Titer

Outcome	Anti-Gmd Antibody Levels			P Value*
	Low (MFI <1,550) (N = 54)	Intermediate (MFI 1,550-9,000) (N = 122)	High (MFI >9,000) (N = 18)
				High Vs. Low	Trend
Adverse outcome	18 (33.3%)	20 (16.4%)	2 (11.1%)	0.078	0.010
Fracture present	2 (3.8%)	2 (1.6%)	0 (0%)
Infection present	5 (9.4%)	5 (4.1%)	1 (5.6%)
Definitive procedure†	2 (3.8%)	7 (5.7%)	1 (5.6%)
Amputation	6 (11.3%)	4 (3.3%)	0 (0%)
Septic death	3 (5.7%)	2 (1.6%)	0 (0%)
Infection controlled	22 (40.7%)	61 (50.0%)	12 (66.7%)	0.101	0.064
Inconclusive	14 (26.4%)	41 (33.3%)	4 (22.2%)	0.999	0.893

^*Based on Fisher exact test for comparing high versus low groups and the exact Cochran-Armitage test for trend for comparing the ordinal low, intermediate, and high groups.

^†Outcomes cited as “pseudarthrosis,” “arthrodesis,” “retained spacer,” and “fusion” in the doctor note at 1 year postoperatively.

Given these data suggesting that the absence of circulating anti-Gmd antibodies at the time of debridement for osteomyelitis is a risk factor for an adverse event within 1 year, we performed logistic regression analyses to determine relative odds compared with other known risk factors for surgical site infection and/or periprosthetic joint infection¹. These risk factors included BMI of >40 kg/m², diabetes, age of >70 years, male sex, and Charlson Comorbidity Index of >1. Table II shows the results of this analysis, in which the absence of detectable circulating anti-Gmd antibodies was the only significant risk factor associated with increased odds of an adverse outcome in this cohort (2.68-fold increase; p = 0.008). Of these 194 patients, the AO Trauma CPP Bone Infection Registry only contained sufficient risk factor data on patients with BMI of >40 kg/m², diabetes, age of >70 years, sex (male), and Charlson Comorbidity Index of >1.

TABLE II.

Relative Odds of an Adverse Event from Lack of Anti-Gmd Monoclonal Antibodies and Known Risk Factors^*

Risk Factors	Risk of Adverse Events
	No.	Incidence	Odds Ratio	95% CI	P Value
Anti-Gmd <1 ng/mL	194	27.80%	2.68	1.30, 5.54	0.008*
Charlson Comorbidity Index >1	182	21.40%	1.63	0.70, 3.76	0.255
BMI >40 kg/m2	193	7.30%	1.05	0.28, 3.95	0.946
Diabetes	166	17.50%	1.71	0.68, 4.29	0.256
Age >70 yr	194	17.00%	2.28	0.99, 5.21	0.052
Female	194	32.50%	1.52	0.74, 3.12	0.256

^*Odds ratios, 95% CIs, and p values were calculated based on univariate logistic regression models.

Discussion

Recent studies have shown no changes in the rates of periprosthetic joint infection, the primary pathogen, treatment algorithm, or prevalence of poor outcomes since the original revision surgical standards of care were established half a century ago^1,44,45. These experts concluded that development of effective immunotherapy against S. aureus is among the highest priorities in orthopaedics¹. All active and passive vaccine trials to date have failed⁴⁶. Of note, these trials were based on protective efficacy in small animals, and antibody opsonophagocytic activity was the primary biomarker of immunity in human volunteers and patients. Retrospectively, failure of opsonophagocytic antibodies to protect humans is not surprising when considering that patients with agammaglobulinemia show no increase in the incidence of S. aureus infection⁴⁶. Additionally, traditional rodent models have not been predictive of human responses to staphylococcal infections for either protective efficacy^47,48 or human inflammatory responses to sepsis⁴⁹.

Knowing that these failed attempts to develop vaccines neglected translational research with in vivo models containing face and construct validity of surgical site infection^48,50, we identified Gmd as a validated target for immunotherapy and developed an anti-Gmd monoclonal antibody (1C11) via an unbiased antigen discovery screen^{14,24,26,37,38}. Recently, we showed that (1) 1C11 synergizes with the standard-of-care antibiotic therapy (vancomycin) in the 1-stage-exchange murine model of methicillin-resistant S. aureus implant-associated osteomyelitis¹⁵; (2) 1C11 passive immunization of a clinically relevant sheep model is feasible, is safe, and has favorable pharmacokinetics⁴³; and (3) humans who recover from methicillin-resistant S. aureus osteomyelitis have high titers of circulating anti-Gmd antibodies^38,43.

Another historical weakness of staphylococcal vaccine development has been a rush to clinical trials prior to clinical-validation studies that assessed the relationship between host immune response against the target antigen and patient outcomes^5,23. Therefore, we utilized the AO Trauma CPP Bone Infection Registry to assess the relationship between circulating anti-Gmd antibody levels in patients with S. aureus osteomyelitis and clinical outcomes. Consistent with our theory of specific anti-Gmd antibody deficiency in the rare ~1% of elective surgical patients who develop a surgical site infection⁸, we found that all osteomyelitis patients studied had detectable humoral immunity against some S. aureus antigens, but only 20 patients (6.7%; p < 0.0001) had high levels of anti-Gmd antibodies (>10 ng/mL) in their serum prior to surgical treatment of the infection (Fig. 1). Moreover, patients with undetectable anti-Gmd titers demonstrated a trend of increased adverse outcomes compared with those with high anti-Gmd titers (Table I), as well as having significantly (2.68-fold) higher odds of adverse outcomes overall, which was greater than all of the other known host risk factors assessed in the present study (Table II). In terms of protection, we found a trend of higher “infection control” rates in patients with high versus low anti-Gmd titers (Table I), and for every tenfold increase in concentration of endogenous anti-Gmd antibody in serum, a 60% reduction in adverse event odds was observed (see Appendix Supplemental Figure 2; OR, 0.40; p = 0.04). These results imply that endogenous anti-Gmd antibody deficiency is common in osteomyelitis patients and that those with undetectable levels prior to a surgical procedure might benefit from passive immunization with an immunomodulatory neutralizing anti-Gmd monoclonal antibody.

Although these findings represent the first clinical evidence directly associating the humoral immune response of a patient with osteomyelitis against a specific S. aureus bacterial surface antigen with their clinical outcome, several important limitations need to be considered. Registry data analyzed retrospectively cannot formally establish a cause-and-effect relationship between the presence or absence of an antibody response and clinical outcome at 1 year postoperatively. We are unable to correlate 3 types of surgical procedures (fracture repair, arthroplasty, or osteomyelitis debridement) and outcomes (194 patients in 27 subgroups) because the subgroup n values were too small to be meaningful. Additionally, we had missing data, possible reporting errors, potential selection bias of included patients, and patients lost to follow-up. Thus, the present data merely establish significant associations between anti-Gmd antibody levels and S. aureus bone infection outcomes and require randomized controlled trials to prove that anti-Gmd antibodies are efficacious in S. aureus osteomyelitis patients. Second, no data were collected on other known risk factors for infection such as alcohol consumption, drug use, or smoking, and the number of patients with the other known demographic risk factors for surgical site infection and/or periprosthetic joint infection were too low to study. Although the very low numbers of patients with high levels of anti-Gmd titers established a significant association between antibody deficiency and S. aureus osteomyelitis (p < 0.0001), this high anti-Gmd group with validated clinical outcomes proved to be too small (n = 18) for statistical analyses (Table I). Thus, the trends for reduced adverse events and increased infection control in the high anti-Gmd versus low anti-Gmd groups need to be tested in a prospective study with 2.4 times more osteomyelitis patients in each group for 80% power, and 3 times more patients for 90% power.

Conclusions

Deficiency in circulating anti-Gmd antibodies was associated with S. aureus osteomyelitis incidence in type-A hosts and serious adverse clinical outcomes. At 1 year postoperatively, high levels of anti-Gmd antibodies were associated with a nearly 3-fold increase in infection-control odds. Prospective studies to validate these findings in order to enable an anti-Gmd passive immunization therapy for osteomyelitis are needed.