Distinct expression trend of signature antigens of Staphylococcus aureus osteomyelitis correlated with clinical outcomes

2.1 Patient enrollment and microbiological analysis to identify pathogens

This study was the second phase of the 2012 AOCPP Bone Infection in China,¹³ and was approved by the Ethics Committees of the First Affiliated Hospital of ZMU and the Southwest Hospital of Chongqing (SHC) From January 2014 to December 2017, we enrolled 55 patients with S. aureus long bone infection (monomicrobial or polymicrobial) confirmed by clinical culture of the specimen that was collected from during surgical irrigation and debridement at the ZMU and SHC. The inclusion criteria are (1) patients aged 18 years or older; (2) confirmed oxacillin-/methicillin-sensitive Staphylococcus aureus (OSSA/MSSA) or methicillin-resistant Staphylococcus aureus (MRSA) infection involving a long bone (femur, tibia, fibula, humerus, radius, ulna, and clavicle) with one (or a combination) of the osteomyelitis, fracture fixation hardware/prosthetic joint infection, infection around an arthroplasty; (3) be able to understand the content of the informed consent form (ICF) and able to participate in the clinical investigation; (4) signed written ICF. The exclusion criteria are (1) patients who cannot give informed consent; (2) patients who cannot attend the follow-up visits; (3) recent history of substance abuse (i.e., recreational drugs, alcohol) that would preclude reliable assessment; and (4) prisoners. All patients understood the study clearly and joined voluntarily. Relative documents of ICF were signed by each individual before serum specimens were collected. The infection sites, the duration of infection before administration, and the types of infection of the involved long bone were recorded. All infection pathogenic bacteria were determined by clinical laboratory examinations at the administration hospital and confirmed again by the Joint Orthopaedic Research Center of ZMU and the University of Rochester Medical Center (URMC) at ZMU (Tables 1, 2 and Figure 1).^12-14

2.2 Serum collection and standard serum selection

Whole blood was drawn from the patients at 0 month (n = 55), 6 months (n = 39), and 12 months (n = 39) of time points, and stored in the standardized serum separator tube that was supplied by AOCPP, and was set to clot for about 2 h at room temperature. The serum was then collected after centrifugation at 12,000 rpm for 20 min at room temperature, and stored in aliquots at −80°C for later use. To generate a standard curve for further Luminex analysis, sera from four patients who underwent total hip or knee joint replacement without infection disease were combined as a noninfected control. Four of the 55 patients whose baselines showed the highest expression level of seven-candidate antigens were pooled for duplicated serial dilution of: 1:100, 1:500, 1:2500, 1:12,500, 1:62,500, 1:312,500, 1:1,562,500. As part of standard clinical care, 0- 6-, and 12-month white blood count (WBC) (n = 52, 20, 19), CRP (n = 42, 19, 19), and ESR (n = 41, 20, 19) were tested and recorded for each individual (Figure 2).

2.3 The criteria of infection controlled versus adverse outcome

Infection controlled is defined as none of the clinical signs and symptoms of active disease (fever, chills, focal pain, erythema, swelling and draining sinuses, etc.) observed at the end of 12-month follow-up and no images (radiography, computed tomography, magnetic resonance imaging, ultrasound) evidence showing the progression of bone infection. The adverse outcome is defined as failure response to therapy, as evidenced by one or more of (1) persistence of drainage; (2) a sinus tract recurrence or failure to close; (3) clinical systemic signs of infection (chills, fever, and local pain); and (4) imaging evidence of infection progression.^{2, 6, 15}

2.4 MagPlex-Avidin magnetic beads conjugation and Luminex assay

The seven immunodominant antigens of S. aureus were provided by GenScript, include (1) cell wall-modifying proteins Amd and Gmd, (2) IsdA, IsdB, and IsdH, and (3) secreted virulence factor of CHIPS, and SCIN (GenScript, Lot# U8848CL180801/P10011801). These biotinylated antigens of S. aureus were diluted to the final concentration of 100 nM following the previous protocol developed at the URMC.¹¹ Briefly, the final concentrations of the antigens were conjugated with magnetic beads. A total of 1,000,000 beads/ml were mixed with 50 µl equal volume per well of patient serum (10,000-fold dilution) and incubated for 2 h under sonication at room temperature. After washing three times with 100 µl of 0.05% PBST, the secondary antibody of the phycoerythrin-conjugated goat anti-human immunoglobulin G (IgG) reagent in phosphate-buffered saline-1% bovine serum albumin was added onto the plate shaker at 800 rpm for 60 min. After this was repeated three times, the plates were ready for Luminex 200^TM analysis. The mean fluorescent intensity (MFI) value was arbitrary as IgG titers for further data analyses if the coefficient of variation (CV) among the replicates was less than 20%. To make sure that the data generated from the multiplex immunoassay was acceptable, the next situation was remeasured. (1) If a particular sample had bead count below 25, the sample was run again on the Luminex machine or we repeated the immunoassay for the sample until the bead count was greater than 25; (2) %CV = 100 × standard deviation/mean, a CV of less than 20% among the replicates is acceptable. (3) %difference = (FI-trimmed mean)/trimmed mean) × 100, to minimize intra-assay variability the %difference was kept below 20%.

2.5 Statistical data analyses

SAS (Version 9.4) and GraphPad Prism Software (Prism7.0) were used to perform statistical analyses. The correlation analysis linear attendance of MFI of three-time points (0-, 6-, and 12-month) of adverse outcome versus infection controlled was analyzed by mixed-effects models of type III tests of fixed effects with GLIMMIX procedure. Quantitative results were shown as means ± standard deviations, as well as the linear relationship between the dilutions ratio of STN and MFI values using Pearson correlation analysis. One-way analysis of variance (ANOVA) was used to compare the MFI of three-time points (0, 6, and 12 months), and with the Kruskal–Wallis test, multiple comparisons were performed for the duration of infection. CRP, WBC, ESR, and two-way ANOVA was used for the seven-candidate antigens expression of the baseline. A p-value of less than 0.05 was defined as statistically significant.

3.1 The lower extremity was the dominant infection site, and the type of osteomyelitis and infection pathogen of MSSA/OSSA significantly affected the duration of treatment

A total of 55 patients with S. aureus infection included 46 males (41.7 ± 13.2 years old) and 9 females (35.5 ± 10.5 years old). The infection sites were dominant in the lower extremities (51, 92.75%), especially the tibia and fibula 45.45% (n = 25) and the femur 40.0% (n = 22) (Table 1). Overall, a monomicrobial infection pathogen was responsible for 89.1% (n = 49) of cases, while a polymicrobial infection was seen in 10.9% (n = 6) of cases while only the tibia and fibula were involved in this study (Table 2). Among polymicrobial infections, Pseudomonas aeruginosa, Corynebacterium, Klebsiella pneumoniae, or Citrobacter freundii were combined with MRSA or MSSA (Table 3). Interestingly, the period in which a patient sought medical assistance during a fracture fixation infection and prosthetic joint infection was significantly shorter than osteomyelitis (Figure 1A), while the patients with MRSA infections received more hospital care than patients who were infected with MSSA/OSSA (Figure 1B). Considering that this group of patients was mainly recruited from the rural area in Southwest China, this may be the main reason for the different regional characteristics of bone infection in this study.

3.2 CRP, WBC, and ESR tests cannot distinguish between the clinical outcomes of a controlled infection and adverse outcome

Longitudinally CRP (Figure 2A), WBC (Figure 2B), and ESR (Figure 2C) data collection showed that the 6-month and 12-month follow-up time points were significantly lower than the baseline, but none of these tests were able to distinguish between a controlled infection or adverse outcome. The clinical results of a controlled infection or adverse outcome (*, ^#p < 0.05) are shown in Figure 2D, may result in clinicians discharging patients before the infection has been controlled. Without intermittent clinical follow-up visits, this may put the patients at risk of recurrence of infection over time.

3.3 We successfully established a Bio-Luminex standard curve of testing antigens locally but failed to create a repeatable gold standard of Bio-Luminex of testing antigens for horizontal comparisons between three centers

A pooled standard serum was developed from the four patients that were screened from 55 patients who expressed the highest mean fluorescent intensity (MFI) of antigens of IsdA, IsdB, IsdH, Amd, Gmd, CHIPS, and SCIN. Then, a highly repeatable standard curve was established locally at ZMU (Figure 3A). Meanwhile, the China Food and Drug Administration approved intravenous immune globulin (CIVIG) which was then employed to establish a comparable standard curve for all centers at ZMU, URMC, and Virginia Commonwealth University (VCU) due to the Chinese regulation on the exportation of human tissue. Unfortunately, the results disproved the consistency of the expression levels of each candidate antigen (Figure 3B), URMC (Figure 3C), and VCU (data not shown). Thus, we failed to generate an exact match MFI of candidate antigens by using CIVIG. Therefore, the ZMU, URMC, and VCU were forced to analyze the data separately (Figure 3B,C).

3.4 Significant overexpression of IsdA, IsdB, IsdH, Amd, Gmd, CHIPS, and SCIN in the infected patient baseline that compared with noninfected patient and CIVIG

By using the customized standard serum that was pooled from the four patients, the dilution of 1:20,000 was applied for all three groups of infected patient baseline, noninfected patient, and CIVIG to perform all seven antigens of IsdA, IsdB, IsdH, Amd, Gmd, CHIPS, and SCIN testing. The results indicated that when compared with NIP, there was significantly overexpression of IsdA (2.3-fold), IsdB (1.6-folds), IsdH (2.3-fold), Amd (2.5-fold), Gmd (2.2-fold), CHIPS (2.3-fold), and SCIN (2.4-fold). When we compared with CIVIG, the results were significantly higher: IsdA (5.6-fold), IsdB (2.7-fold), IsdH (8.5-fold, Amd (3.2-fold), Gmd (3.5-fold), CHIPS (4.9-fold), and SCIN (4.1-fold) (two-way ANOVA, *p < 0.0001) (Figure 4). These findings indicated that all seven-candidate antigens are potential immunological signatures for clinical outcomes of therapy.

3.5 Distinct longitudinal expression trend of IsdB, Amd, Gmd, and CHIPS were observed between infection-controlled patients versus adverse outcome patients

On the basis of clinical criteria of chronic osteomyelitis therapy,^{2, 6} the clinical results of the current study were divided into an infection-controlled group and an adverse outcome group. In follow-up time points, the longitudinal expression trend of each individual antigen IsdB, Amd, Gmd, and CHIPS showed a significantly distinct increase in the -adverse outcome group, while showing a significant decreasing trend in the infection-controlled group (**p < 0.05). However, this was not observed in the IsdA, IsdH, and SCIN groups (Figure 5A–G). We further found that distinct MFI of IsdB, cell wall-modifying protein Amd and Gmd, and secreted virulence factor CHIPS were confirmed by using the difference of the 12 months follow up from the baseline between infection controlled patients versus adverse outcome patients (Data not shown).

4.1 The regional differences in bacterial infections may be related to the geographical location of patients, regional climate characteristics, and medical conditions, where it may be necessary to test the genetic characteristics of pathogens for confirmation

The patients admitted in this study were mainly from rural areas of Southwest China. These areas are located at an altitude of 1000–1600 m, with an average annual temperature of 12.6–13.1°C. It is often raining while being hot and humid at the same time. As a result, the relative humidity is high, and the climate types are three-dimensional and diverse. It belongs to the humid monsoon region of the central subtropical plateau, and is the low-value area of perennial solar radiation and is an area susceptible to bacteria.¹⁶ Before 2011, comprehensive healthcare insurance had not yet been fully implemented in China, and some patients had not received proper medical care due to economic reasons, which played a decisive role in the prolongation of the bone infection.¹⁷ On the basis of data collected in our studies, the bone infections caused by S. aureus have obvious regional characteristics.^{13, 14} Most of these patients had lived in mountainous, far from the cities and had not been exposed to hospital-derived S. aureus. This may be the main reason why the strains that infected patients are different from other regions.^{18, 19} All specimens were collected from bone biopsy and nasal swab have undergone genetic testing and with pending data analysis.

4.2 A distinct overexpression trend of IsdB, Amd, Gmd, and CHIPS was observed in adverse outcome patients

Existing clinical indicators for monitoring infection cannot distinguish whether the treatment is effective, and there is an urgent need to find sensitive and effective biochemical signature markers. As shown in the results, to monitor treatment outcome of bone infection, CRP, WBC, and ESR testing are the current clinical standard tests and tend to decrease at the 6- and 12-month time points. Unfortunately, all these changes cannot distinguish clinical outcomes of infection-controlled or adverse outcomes which represent a significant limitation and demands sensitive signature markers to surveil treatment outcomes. The S. aureus genome can encode about 2700 proteins, including cytoplasm, cell membrane, cell surface, and secreted proteins.^20-22 Among the many proteins, cell surface and secreted proteins have a high chance of being recognized by the host immune system with resultant antibodies against S. aureus.^23-25 Utilizing a multiantigen Luminex immunoassay with specific antibodies against S. aureus, the humoral immune response can be measured in chronic orthopedic infections and have shown promising value.^26-28 Nishitani et al.¹¹ screened 14 proteins of S. aureus and concluded that measurement of the host antibody response is a predictor of ongoing infection that may prove to have prognostic value. Our results show that all seven antigens were overexpressed in infected patients compared that those of noninfected patients: IsdA (2.3-fold), IsdB (1.6-fold), IsdH (2.3-fold), Amd (2.5-fold), Gmd (2.2-fold), CHIPS (2.3-fold), and SCIN (2.4-fold). However, only IsdB, Amd, Gmd, and CHIPS showed significant overexpression longitudinally in the adverse outcomes group versus infection-controlled group by using mixed-effects models of type III tests of fixed effects with GLIMMIX procedure. We noticed that the expression level of Gmd in this study was higher in the adverse outcomes group, which is slightly different from those patients' serum collected from North American and Europe.¹⁰ Further study to illustrate the mechanism is warranted.

In summary, the current clinical standard of care tests is unable to distinguish therapeutic outcomes of S. aureus osteomyelitis. Significantly, overexpression of antigens IsdB, cell wall-modifying protein Amd and Gmd, and secreted virulence factor CHIPS at 0-, 6-, 12-month time points and their distinct overexpression trend could be the signature markers of adverse outcome of S. aureus osteomyelitis.

4.3 Limitations

(1) The most of patients who enrolled in the study were from rural areas, and current data did not include the majority of patients with a traumatic bone infection in China's urban area. (2) Some patients missed follow-up time points of 6th and 12th months. Some patients who delayed 6-months follow-up to 8th months were classified as 6th-month specimens. Several patients who were late for the 12th-month visit by about 1 month were considered 12 months. (3) The intravenous immunoglobulin for clinical use is prepared from the serum of between 1000 and 15000 donors per batch,²⁹ which represents an average level of candidate antigens (IsdA, B, and H, SCIN, CHIPS) in a population. Because the sample size of four noninfection patients is too small to represent a broad population as the CIVIG does, this could result in a higher expiration level of signature antigens mentioned above. (4) Also, the sample size calculation by an a priori power analysis still lacks because randomization is not the same.