Introduction

Staphylococcus aureus is a leading cause of serious bacterial infections1,2. S. aureus bacteremia (SAB) may persist in some patients despite appropriate antibiotic therapy. The use of intravascular catheters or prosthetic devices, metastatic infection, methicillin resistance, vancomycin minimal inhibitory concentration (MIC), and accessory gene regulator (agr) dysfunction have been suggested as risk factors for persistent bacteremia (PB)3,4,5,6. Several reports have identified clinical outcomes associated with PB compared with resolving bacteremia (RB). In patients with SAB, PB is typically associated with complications of bacteremia, longer hospitalization, and increased mortality than RB4,6,7,8,9. Factors that predict mortality must be elucidated to determine the optimal strategy for treating S. aureus PB.

The outcome of S. aureus infection could be influenced by the host immune response to S. aureus. The cytokine response to S. aureus infection appears to be related to clinical outcomes. IL-6 is an early inflammatory marker of complicated SAB, and IL-10 is associated with mortality and PB10,11,12,13. Also, previous studies have attempted to establish the contribution of T cell immunity during S. aureus infection, along with efforts to develop a vaccine against S. aureus infection14,15. CD4+ T helper (Th) cell immunity might be important for fighting S. aureus infections in humans16. In particular, Th1 cells produce IFN-γ, which promotes bacterial elimination inside macrophage phagosomes and plays an important role in immunological and inflammatory processes during S. aureus infection14,16,17,18.

Although there have been several reports evaluating the mortality of SAB and risk factors for PB, the factors affecting the mortality of patients with PB remain unclear. In addition, most studies of T cell immunity during S. aureus infection have been performed in murine models, not in humans17,18,19. The goal of this study was to evaluate the clinical, microbiologic, and genotypic risk factors of 30 days mortality in patients with S. aureus PB in a prospective cohort over a 13 years period, and to investigate the cytokine kinetics in the plasma and the T cell immune response in a clinical setting of S. aureus PB.

Patients and methods

Human studies

Plasma and peripheral blood mononuclear cells (PBMCs) were obtained from patients at the Asan Medical Center as part of the S. aureus bacteremia and immune response studies. This study was conducted at the Asan Medical Center, a 2700-bed tertiary-care referral center in South Korea, from July 2008 to December 2020. All adult patients with SAB were prospectively enrolled in a cohort and observed over 12 weeks. Patients were excluded from the analysis if they had polymicrobial bacteremia, had been discharged before obtaining positive blood culture results, or had SAB within the previous three months. Demographic characteristics, underlying diseases or conditions, the severity of underlying diseases, the severity of bacteremia, place of infection, site of infection, presence of a central venous catheter (CVC) or prosthetic device, patient management, and clinical outcomes were recorded. Acute Physiology and Chronic Health Evaluation II (APACH II) and Pitt bacteremia scores were calculated on the day of the first positive blood culture to assess the severity of bacteremia20,21. The Charlson comorbidity index was employed to determine the severity of comorbid conditions22.

Study definitions

PB was defined as bacteremia for ≥ 7 days while receiving appropriate antibiotics therapy. The place of acquisition was categorized as community-acquired, healthcare-associated, or hospital-acquired infection23. Infective endocarditis was defined according to the modified Duke criteria24. Two sets of cultures were obtained to diagnose incident bloodstream infections. The index day was defined as the day on which the first positive blood culture was obtained. Empirical antibiotic treatment was considered appropriate if at least one antibiotic effective against the SAB isolate was started within 24 h after the index blood culture. Prosthetic devices included orthopedic devices, cardiovascular electronic devices, prosthetic valves, and vascular grafts. Septic shock was defined as sepsis with persisting hypotension that required vasopressors to maintain the patient’s blood pressure despite adequate fluid resuscitation25. The concentrations of cytokines and the proportion of IFN-γ secreting CD4 + T cells were compared according to the patient's bacteremia severity scores and 12-week mortality. The patients were divided into groups: Group 1 had Pitt bacteremia score < 4, and Group 2 had Pitt bacteremia score ≥ 4. Group 3 was patients who survived for 12 weeks from the index day, and Group 4 was patients who died within 12 weeks.

Microbiological analysis

All S. aureus isolates were identified by standard methods. The first blood isolate obtained from the patient was used for the microbiological and molecular assessments. The minimum inhibitory concentration (MIC) of vancomycin was determined using a broth microdilution method (BMD) according to the standard protocol. Methicillin resistance was confirmed by detecting the mecA gene via polymerase chain reaction (PCR).

The staphylococcal cassette chromosome mec (SCCmec) type and agr genotype were identified as previously described26,27. The agr function was determined by the δ-hemolysin activity as described elsewhere28. Multi-locus sequence typing (MLST) was performed for all strains as described previously29. MLST allele names and STs were derived from the MLST database (http://www.mlst.net). Clonal complexes (CCs) were assigned to groups of isolates that shared six of seven alleles via eBURST (http://eburst.mlst.net).

Cytokine assays

The concentrations of IL-6, IL-10, IL-17A, IL-12p70, IL-13, IL-4, TNF-α, and IFN-γ in plasma were measured with a ProcartapPlex Multiplex Immunoassay (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. If the level for a particular cytokine was below the detectable limit, the concentration was recorded as 0.

Peripheral blood mononuclear cell isolation

Venous blood was collected in heparinized vacutainer tubes (BD Biosciences Pharmingen, San Jose, CA, USA) and diluted with 1 × PBS (HyClone Laboratories, Inc., South Logan, UT, USA). The PBMCs were isolated by Ficoll-Hypaque (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) gradient separation. The PBMCs were used directly or washed in RPMI-1640 and diluted in freezing medium containing 40% RPMI-1640, 50% fetal calf serum (Gibco by Invitrogen, Carlsbad, CA, USA), and 10% dimethyl sulfoxide (Sigma Aldrich, St. Louis, MO, USA), gradually frozen in a freezing container (Mr. Frosty, Nalgene Cryo 1 °C; Nalgene Co., Rochester, NY, USA), and stored in liquid nitrogen until analyzed.

In vitro activation, intracellular staining, and flow cytometry analysis of PBMCs

PBMCs were thawed and washed before counting and the exclusion of non-viable cells by trypan blue staining. Cells were resuspended to a final density of 5 × 105–1 × 106/well and stimulated with 5 × 107–1 × 108 CFU/mL of heat-killed S. aureus (the multiplicity of infection was 1:100) for 4 days. The cell viability was confirmed by Near-IR fluorescent reactive dye (Invitrogen), and then the cells were stained for CD3 (BD horizon, clone UCHT1), CD4 (eBioscience, clone RPA-T4), and CD8 (eBioscience, clone RPA-T8). The cells were fixed and permeabilized using the BD Cytofix/Cytoperm Fixation and Permeabilization Kit, followed by intracellular staining with fluorochrome-conjugated antibodies against IFN-γ (eBioscience, clone4S.B3). Flow cytometric data were acquired with a BD FACSCanto II and analyzed using FlowJo software version 10.8.1 (Becton, Dickinson & Company; www.flowjo.com)30.

Statistical analysis

We compared surviving and deceased patients with PB. Categorical variables were compared using the Chi-square test or Fisher’s exact test, as appropriate. Continuous variables were compared using Student’s t test and the Mann–Whitney U-test. Logistic regression was used to assess risk factors associated with PB mortality. Risk factors for 30 days mortality from PB were assessed using univariate analysis, and statistically significant variables were included in the multivariate analysis. Two-way ANOVA with Bonferroni post-tests was used to analyze the cytokine kinetics. A two-tailed p value < 0.05 was considered significant. All statistical analyses were performed using IBM SPSS Statistics for Windows, version 25 (IBM Corp., Armonk, NY, USA) and GraphPad Prism version 9.4.1 (GraphPad Software, Inc., San Diego, CA, USA; www.graphpad.com).

Ethics declarations

The study was approved by the Institutional Review Board (IRB) of Asan Medical Center (protocol No. 2013-1002). Informed consent was obtained from all individual participants included in the research. The study was performed in accordance with the Helsinki Declaration.

Results

Clinical characteristics and outcomes of patients with PB

From July 2008 to December 2020, 1760 episodes of SAB were reported. Among these 1760 episodes, 242 (13.8%) developed PB with a median duration of 11 days (interquartile range [IQR], 8–16 days). The following types of PB infections were identified: CVC-related infections (33.5% [81/242]), bone and joint infections (20.2% [49/242]), endocarditis (9.9% [24/242]), skin and soft tissue infections (7.0% [17/242]), pneumonia (2.1% [5/242]), and primary bacteremia (7.0% [17/242]). The rate of metastatic infections was 48.3% [117/242].

Comparisons of the demographic and clinical characteristics of the deceased and surviving patients are shown in Table 1. A total of 49 (20.2%) patients died within 30 days after the index day, and 193 (79.8%) patients survived. The deceased patients were more likely to have liver cirrhosis (30.6% [15/49] vs. 11.9% [23/193], p < 0.01), a high Charlson comorbidity score (median score 4 vs. 3, p = 0.03), and a high APACHE II score (median score 19 vs. 16, p = 0.01) than the surviving patients. The surviving patients, on the other hand, were more commonly male (69.9% [135/193] vs. 51.0% [25/49], p = 0.01), had a community-acquired infection (21.2% [41/193] vs. 4.1% [2/49], p = 0.01), diabetes mellitus (39.4% [76/193] vs. 20.4% [10/49], p = 0.01), metastatic bone and joint infections (15.0% [29/193] vs. 4.1% [2/49], p = 0.04), and it took longer to remove an eradicable focus (2 days vs. 1 day, p = 0.04).

Table 1 Demographic and clinical characteristics of patients with persistent S. aureus bacteremia.
Full size table

Risk factors for 30 days mortality of patients with PB

A multivariate logistic regression analysis was performed to identify independent risk factors associated with 30 days mortality from PB (Table 2). The multivariate analysis indicated that the APACHE II score (adjusted odds ratio [aOR], 1.10; 95% confidence interval [CI] 1.03–1.17) and female sex (aOR 3.84; 95% CI 1.40–10.54) were independent risk factors for 30 days mortality from PB.

Table 2 Multivariate analysis of the risk factors for 30 days mortality in patients with persistent S. aureus bacteremia.
Full size table

Microbiological characteristics and genotypic studies in MRSA PB

The microbiological characteristics and genotypic studies of the methicillin-resistant Staphylococcus aureus (MRSA) PB are shown in Supplementary Tables S1 and S2, respectively. Of the 242 episodes of PB, 185 (76.4%) were MRSA. Of the 185 patients with MRSA bacteremia, 36 (19.5%) died within 30 days from the index day, and 149 (80.5%) survived. A total of 145 (78.4%) patients were initially treated with vancomycin. Most MRSA isolates (71.4%) had vancomycin MIC of ≤ 1.0 mg/L. Surviving patients were more likely to have a vancomycin MIC of ≤ 1.0 mg/L than deceased patients (p = 0.02). Common genotypes of MRSA included ST5-SCCmec II (58.9% [109/185]) and ST72-SCCmecIV (28.6% [53/185]), and common agr genotypes were group I (37.8% [70/185]) and group II (58.4% [108/185]). A total of 120 isolates (64.9%) showed agr dysfunction. There were no significant differences in MLST type, SCCmec type, agr genotype, or agr dysfunction between the deceased and surviving groups.

Risk factors for 30 days mortality from MRSA PB

A multivariate logistic regression analysis was employed to identify independent risk factors associated with 30 days mortality from MRSA PB (Supplementary Table S3). The multivariate analysis indicated that the APACHE II score (aOR 1.07; 95% CI 1.01–1.13), liver cirrhosis (aOR 3.77; 95% CI 1.43–9.95), and a vancomycin MIC of ≥ 1.5 mg/L (aOR 3.17; 95% CI 1.39–7.25) were independent risk factors for 30 days mortality in MRSA PB.

Kinetics of IL-6 and IL-10 in patients with PB

Among 242 patients with PB, 22 patients had blood samples collected periodically. Clinical backgrounds and outcomes of the 22 patients are listed in Supplementary Table S4.Their IL-6, IL-10, IL-17A, IL-12p70, IL-13, IL-4, and TNF-α plasma concentrations were analyzed. One patient died within 30 days, and three patients died within 12 weeks of the index day. IL-12p70, IL-13, and IL-4 levels were undetectable in most patients. IL-17A and TNF-α levels were measured in only eight and two patients, respectively. Therefore, IL-17A, IL-12p70, IL-13, IL-4, and TNF-α were excluded from the analysis. The IL-10 level could not be detected in three patients, and thus only 19 patient samples were available for the analysis of IL-10. Changes in the plasma levels of IL-10 and IL-6 during bacteremia are presented in Fig. 1. The cytokine kinetics were analyzed according to the last positive culture day. Plasma concentrations of IL-10 and IL-6 generally decreased during the study.

Figure 1
figure 1

Cytokine kinetics during the bacteremia period. Plasma concentrations of IL-10 and IL-6 were analyzed in serial blood samples from 19 and 22 patients, respectively. (a) and (b) are the mean concentrations of IL-10 and IL-6 during the bacteremia period. Day 0 is the last positive culture day.

Full size image

We compared groups according to the Pitt bacteremia score and 12-week mortality (Figs. 2 and 3). IL-10 concentrations were higher in patients with high bacteremia scores than those with low bacteremia scores 4–9 days prior to the last positive culture day (Fig. 2, p < 0.05). The concentrations of IL-10 in the 4–9 days before the last positive culture day were higher in patients who died within 12 weeks from the index day than those who survived (Fig. 3, p < 0.05). IL-6 concentrations were generally higher in patients with higher Pitt bacteremia scores and those who died within 12 weeks, but the differences between groups did not reach statistical significance.

Figure 2
figure 2

Comparison of cytokine concentrations in groups 1 and 2 in serial blood samples collected during the study period. (A) IL-10 concentration in plasma. Group 1 comprises 15 patients, and group 2 comprises four patients. (B) IL-6 concentration in plasma. Group 1 comprises 18 patients, and group 2 comprises 4 patients. Day 0 was the last day of positive culture. *p < 0.05. Group 1 comprises patients with a Pitt bacteremia score of < 4. Group 2 comprises patients with a Pitt bacteremia score of ≥ 4.

Full size image
Figure 3
figure 3

Comparison of cytokine concentrations in group 3 and 4 in serial blood samples collected during the study period. (A) IL-10 concentration in plasma. Group 3 comprises 15 patients, and group 4 comprises four patients. (B) IL-6 concentration in plasma. Group 3 comprises 18 patients, and group 4 comprises 4 patients. Day 0 was the last day of positive culture. *p < 0.05. Group 3 comprises patients who survived for 12 weeks from the index day. Group 4 comprises patients who died within 12 weeks from the index day.

Full size image

Kinetics of IFN-γ and CD4+ T cells secreting IFN-γ in patients with PB

IFN-γ was undetectable in the plasma of 22 patients from whom blood samples were collected periodically (Supplementary Table S4). We collected PBMCs from these 22 patients to analyze IFN-γ further. After stimulation of the PBMCs with heat-killed S. aureus, the IFN-γ in the cell culture medium was detected for PBMCs isolated from 19 patients. IFN-γ concentrations increased until just before the blood cultures underwent a negative conversion and decreased after the last positive blood culture day for 15 patients with low bacteremia scores (Fig. 4a). IFN-γ concentrations were very low for all time points for four patients with high bacteremia scores.

Figure 4
figure 4

Comparison between groups of IFNγ concentrations and the proportion of IFNγ-secreting cells among the CD4+ T cells in patients with serial blood samples collected during the study period. PBMCs were collected from 22 patients and stimulated with heat-killed S. aurerus, and the concentration of IFNγ was measurable in 19 patients. In addition, the proportion of IFNγ-secreting cells among the CD4+ T cells was assessed using flow cytometry for 16 patients. (A) IFNγ concentration in Group 1 and Group 2. (B) The proportion of IFNγ-secreting cells among CD4+ T cells in Group 1 and Group 2. (C) IFNγ concentration in Group 3 and Group 4. (D) the proportion of IFNγ-secreting cells among CD4+ T cells in Group 3 and Group 4. Day 0 was the last day of positive culture. *p < 0.05. Group 1 comprises patients with Pitt bacteremia score < 4. Group 2 comprises patients with Pitt bacteremia ≥ 4. Group 3 comprises patients who survived for 12 weeks from the index day. Group 4 comprises patients who died within 12 weeks from the index day.

Full size image

To determine whether CD4+ T cells produce IFN-γ, intracellular staining was performed, and IFN-γ secreting cells among the CD4+ T cells of 16 patients could be analyzed by using flow cytometry. The proportion of IFN-γ secreting CD4+ T cells tended to be correlated with the IFN-γ concentration and was significantly high 1–3 days before the last positive culture day for 15 patients with low bacteremia scores (Fig. 4b, group 1). IFN-γ secreting CD4+ T cells were identified for only one patient with a high bacteremia score (Fig. 4b, group 2). When comparing groups of patients according to the 12-week mortality, the concentration of IFN-γ increased until the blood culture became negative and decreased after the last positive culture day in both groups (Fig. 4c). In addition, the proportion of IFN-γ secreting cells among the CD4+ T cells was the highest 1–3 days before the blood culture became negative for 13 patients who survived for 12 weeks from the index day (Fig. 4d, group 3). Figure 5 shows the proportion of IFN-γ secreting cells among the CD4+ T cells analyzed using flow cytometry 1–3 days before the last positive culture day.

Figure 5
figure 5

Proportion of IFNγ-secreting cells among CD4+ T cells analyzed using flow cytometry 1–3 days before the last positive culture day. Group 1 comprises 15 patients, and group 2 comprises one patient. Group 3 comprises 13 patients, and group 4 comprises 3 patients. Group 1 comprises patients with Pitt bacteremia score < 4. Group 2 comprises patients with Pitt bacteremia ≥ 4. Group 3 comprises patients who survived for 12 weeks from the index day. Group 4 comprises patients who died within 12 weeks from the index day.

Full size image

Discussion

In this study, we evaluated the risk factors associated with mortality in patients with SAB PB and confirmed a T cell immune response during S. aureus infection. In particular, our study is meaningful in that it sequentially observed changes in cytokines and T cell responses over time in human SAB. The APACHE II score and female sex were independent risk factors of 30 days mortality. The subgroup analysis according to MRSA isolates showed that the APACHE II score, liver cirrhosis, and a vancomycin MIC of ≥ 1.5 mg/L were independent risk factors for 30 days mortality. The IL-10 concentration was high in the early phase of bacteremia in patients with high Pitt bacteremia scores and those who died within 12 weeks from the index day. The proportion of IFN-γ-secreting CD4+ T cells was highest just before the blood cultures turned negative for patients with low Pitt bacteremia scores and those who survived for 12 weeks.

Previously, there have been several studies showing the risk factors of SAB PB. Patients with liver cirrhosis have impaired immunity and increased susceptibility to S. aureus infections. S. aureus is an important pathogen in liver cirrhosis patients31,32. Liver cirrhosis is a risk factor for PB compared with RB; moreover, liver cirrhosis patients with SAB had a higher mortality rate than non-liver cirrhosis patients in previous studies8,33. A high vancomycin MIC is also associated with worse clinical outcomes and treatment failure but there has been a controversy about whether a high vancomycin MIC is associated with persistent bacteremia4,8,34,35. Sheng-Hsiang Lin et al.36 showed that a decreased susceptibility to vancomycin was associated with mortality in MRSA PB. In this study, we evaluated the risk factors associated with 30 days mortality in patients with SAB PB. Liver cirrhosis was identified as a risk factor for 30 days mortality in patients with MRSA PB. In addition, the surviving patients were more likely to have a lower vancomycin MIC, and a vancomycin MIC of ≥ 1.5 mg/L was an independent risk factor for 30 days mortality from MRSA PB. Decreased susceptibility to vancomycin appears to affect the mortality of patients with MRSA PB treated with glycopeptides including vancomycin.

The agr locus of S. aureus globally controls the coordinated production of virulence factors27. Mutations in the agr locus cause agr dysfunction, alter the expression of autolysins and hemolysins, and have a global effect on bacterial pathogenicity37,38. agr dysfunction has been associated with PB, vancomycin bactericidal activity, and the mortality of SAB patients6,28,39. Loss of agr function appears to decrease the susceptibility to glycopeptides40. Moise-broder et al.41 reported that agr group II was a predictor of vancomycin treatment failure of MRSA bacteremia and Park et al.42 showed that agr group II was more pronounced in MRSA isolates with a vancomycin MIC of 2 mg/L. In the present study, agr dysfunction was noted in 64.9% of MRSA PB cases and there was no difference between the two groups. agr group II was the most common genotype of MRSA PB, and although statistically insignificant, agr group II was more frequent in patients who died within 30 days from the index day than in surviving patients. Additional studies are necessary to investigate further the association of mortality in PB patients with the agr group and agr dysfunction.

Pro-inflammatory cytokines such as IL-6, tissue necrosis factor (TNF), and IL-17A are necessary for initiating an effective inflammatory process against infection and they are associated with multiple-system organ failure and mortality43. IL-10 is an anti-inflammatory cytokine that regulates the immune response to pathogens44. Previous studies have tried to show a correlation between cytokines and SAB. McNicholas et al.12 reported that IL-6, a pro-inflammatory cytokine, might be an early inflammatory marker of complicated SAB. IL-10 was found to be associated with mortality and persistent bacteremia10,11,13. It is significant that, unlike previous studies, we measured IL-6 and IL-10 concentrations serially throughout the bacteremia period. Although there was no significant difference, the plasma concentration of IL-6 generally tended to be high in patients with high Pitt bacteremia scores and those who died within 12 weeks, before the blood culture became negative. The plasma concentration of IL-10 4–8 days before the last positive blood culture day was significantly higher in patients with high bacteremia scores and patients who died within 12 weeks from the index day. The IL-10 concentration in the early stage of bacteremia can be a predictive factor of poor clinical outcomes of persistent bacteremia.

There have been several lines of evidence showing that T cell immunity plays an important role in S. aureus infection. T lymphocyte-deficient mice were more susceptible to infection with S. aureus, and T cell-derived IFN-γ might be a pivotal regulator of neutrophil recruitment during S. aureus infection in a mouse model45,46. Human immunodeficiency virus infected patients with impaired Th1 cell immunity and low CD4+ counts are highly susceptible to S. aureus and have more invasive infections and a higher recurrence rate than patients with intact T cell immunity47,48,49,50,51. The first study showing the activation of a T cell immune response in human S. aureus infection was published in 2015. Brown et al.30 showed that human S. aureus blood stream infection induces S. aureus antigen-specific IFN- γ-producing CD4+ (Th1) cells.

In our study, we attempted to ascertain the T cell immune response within a real world clinical setting in patients with SAB PB. After stimulation of PBMCs from PB patients with heat-killed S. aureus, we were able to analyze the concentrations of secreted IFN-γ in the cell culture medium. The concentration of IFN-γ was highest 1–3 days before the blood culture became negative, and it decreased after the last positive blood culture day in patients with low disease severity compared to those with severe disease. In addition, the proportion of IFN-γ-producing CD4+ Th1 cells showed similar kinetics to the IFN-γ concentration during infection. Translating these findings, we found that CD4+ T cells are an important source of IFN-γ production during bacteremia and they play an important role in bacterial clearance and contribute to disease severity in SAB PB. While previous studies have shown a potential for T cell immunity to play a protective role against SAB infection, our study further clarifies the role of T cell immunity in S. aureus PB by showing the kinetics of CD4+ T cells during S. aureus bacteremia.

Our study has several limitations. First, the patients with a long duration of bacteremia were likely to be managed in great detail by infectious disease specialists, which included receiving additive diagnostic tests and antibiotic treatment. These actions might have resulted in a bias against the outcome of PB. Second, because the definition of PB is seven days, the possibility of an immortal time bias cannot be overlooked. Further analysis is needed to reduce immortal time bias and determine the presence of differences in outcomes depending on the duration of bacteremia. Third, only a small number of PB patients had serial blood samples collected. Additionally, the cytokines of some patients were unable to be analyzed. It might be necessary to check whether these cytokines did not exist or their concentration was too low to be detected. We might be able to use methods such as amplification in the near future. In the case of IFN-γ, we could not detect IFN-γ directly in plasma and could analyze IFN-γ only after stimulation of PBMC with heat-killed S. aureus. It is conceivable that memory Th1 cells that were previously exposed to S. aureus were induced to produce IFN-γ. Additionally, in our other experiments, IFN-γ was more detectable in plasma collected early in bacteremia. This means that IFN-γ levels in plasma can be affected by the bacterial concentration and antibiotics treatment. Additional studies with a larger sample size might be required to clarify the role of cytokines and T cell immunity during S. aureus bacteremia. Despite these limitations, this study is meaningful in that it is the first study to demonstrate the kinetics of T cell responses and describe the role of IFN-γ-producing CD4+ Th1 cells in a specific clinical setting of human S. aureus PB.

In summary, this study evaluated the risk factors for mortality of persistent SAB and the T cell immune response during SAB PB in a large cohort over 13 years. Clinical factors such as APACH II score in PB, liver cirrhosis, and vancomycin MIC in MRSA PB were associated with mortality. An elevated IL-10 concentration was predictive of mortality from PB. The most significant finding is that IFN-γ producing CD4+ T cells play an important role in bacterial clearance during bacteremia and affect disease severity. The current study will help understand the T cell immune response to S. aureus infection in humans and will guide further studies on SAB.