Staphylococcus aureus bloodstream infection (SABSI) is one of the most severe manifestations of S. aureus with an estimated mortality rate of 20% [1]. However, the outcome of SABSI is variable. Many patients develop uncomplicated resolving infections that respond to appropriate antimicrobial therapy while others develop persistent or complicated infection with metastatic foci (e.g. infective endocarditis (IE)), which may be fatal. Factors that contribute to the clinical outcome of SABSI, include the virulence and antibiotic susceptibility of the infecting isolate [2],[3], host innate and humoral immune responses [4],[5] and the underlying condition of the patient [6]. S. aureus molecules such as peptidoglycan and lipoteichoic acid (LTA) are potential stimulators of cytokine production (e.g. TNF-α, IL-1β, IL-6, IL-4, IL-8, IFN-γ and IL-12), in response to infection [7] but un-regulated cytokine production may contribute to S. aureus pathogenesis. Differential cytokine responses to S. aureus may contribute to varying clinical outcomes of SABSI. This prospective study aimed to; identify cytokines or chemokines that are important in the immune response to SABSI, determine the levels of these molecules in sequential plasma samples taken over the course of SABSI and investigate these cytokine responses in SABSI with a complicated clinical course versus an uncomplicated course and those with MRSA infection compared to those with MSSA infection.

Pro-inflammatory cytokines released in response to S. aureus infection may contribute to the cascade of events that lead to severe complications in SABSI [9]. Cytokines including IL-6, GRO-γ and RANTES are up-regulated in response to S. aureus clinical isolates in epithelial cell infection models [10],[11]. We observed changes in the circulating levels of these cytokines in patients over the course of SABSI. In addition, greater circulating IL-6 levels and reduced leptin levels were observed in complicated SABSI compared to uncomplicated SABSI. Complicated infection was defined specifically for this study as described above and IE was the predominant complication. The surprisingly high rate of IE (8/61, 13%) may be partly explained by our policy of echocardiograms for all patients with SABSI regardless of the response to therapy and the presence of a national centre for cardiac and lung transplantation in one of the study hospitals. In addition, an increased circulating level of GRO-γ was noted in response to MRSA Vs MSSA bloodstream infection.

The cytokine response to sepsis has been well documented, with IL-6 suggested as an early marker and a marker of severity [12]-[15]. The cytokine response in SABSI also appears to involve IL-6 elevation early in the course of infection although the levels were lower than those reported in Gram-positive sepsis [16]. An earlier study suggested that IL-6 levels decay more rapidly in sepsis survivors than in non-survivors over a 96 h period supporting its prognostic value [17]. However, in the present study on SABSI, where cytokines were assessed over a longer period, IL-6 levels decreased to similar levels in complicated and uncomplicated SABSI by day seven following laboratory diagnosis. Nonetheless, a significantly elevated IL-6 level on the day of diagnosis of SABSI appears to correlate with a poorer prognosis.

In-vitro studies have shown increased RANTES levels in epithelial cells exposed to S. aureus clinical isolates [11]. However, a role for RANTES in the immune response to SABSI has not been previously reported. RANTES’ main function is as a chemotactic agent for leucocytes facilitating their recruitment to infection sites [18]. RANTES production is thought to be induced by TNF-α and IL-1α, both of which are produced in response to S. aureus components (e.g. peptidoglycan, LTA) [19],[20]. Our observation of a significant rise in RANTES levels in all patient groups by day seven, suggests that RANTES production is stimulated by S. aureus in vivo and it may have a functional role in the immune response to this pathogen. It appears that once produced, circulating RANTES levels remain high. Furthermore, mean levels remained elevated in those with complicated SABSI up to 14 days following diagnosis. The purpose of this persistence is not clear but it may reflect the ongoing response to persistent SABSI.

Leptin levels were significantly higher in uncomplicated versus complicated SABSI on the day of laboratory diagnosis (p ≤ 0.05) and a similar trend was identified on day seven following diagnosis (p = 0.071), indicating that patients with more severe infection had a sustained attenuated leptin response. Leptin is reported to have both beneficial and detrimental effects for the host in the setting of sepsis. It stimulates proliferation of lymphohaematopoetic cells and increases the phagocytic activity of macrophages which may enhance host eradication of S. aureus but it can also activate endothelial cells resulting in a severe sepsis phenotype [21],[22]. Both negative and positive correlations between leptin levels and clinical outcome in patients with sepsis have been reported [22],[23]. The present study supports the suggestion that higher leptin levels are protective. Decreased leptin levels in patients with complicated infection may contribute to poor clinical outcome by reducing the host's ability to eradicate the infection. These findings support the possibility that administration of leptin may benefit patients with severe SABSI as one animal study reported that exogenous administration of leptin reduced the severity of S. aureus-induced septic arthritis in two mouse models [24].

We did not find evidence that GRO-γ levels were altered in patients with a complicated course of infection. However, our data suggest that this cytokine may play a role in the response to MRSA infection specifically. The cause and clinical implications of this finding are unclear. Bacterial components, such as peptidoglycan and LTA that stimulate GRO-γ production via IL-1 and TNF-α [20] are common to MRSA and MSSA. Furthermore, the chemoattractant properties of GRO-γ are equally important in MRSA and MSSA (BSI). It has been shown that mecA augments the expression of virulence genes such as fnb[25] and increased expression of α-toxin and phenol soluble modulin has been shown in community-acquired MRSA [26]. It is therefore possible that differential host immune responses such as increased GRO-γ production may result indirectly from increased expression of virulence genes in MRSA.

There were limitations to this study. Despite extension to two centres, only 61 patients were recruited over 28 months, providing a relatively small study size. There were logistical issues related to obtaining samples from very ill patients or the movement or discharge of patients before samples could be taken. Samples at 14 days after diagnosis were only taken from inpatients with complicated SABSI and not those discharged by this time. Moreover, a true baseline cytokine level was not obtained as this would require screening large numbers of patients before the development and laboratory diagnosis of SABSI. Complicated SABSI was under-represented because of the difficulty obtaining patient consent or assent from relatives of very ill or rapidly deteriorating patients. The contribution of co-morbid conditions associated with SABSI (e.g. concurrent infection, autoimmune conditions, recent surgery etc.) which was not part of this study, to the cytokine response cannot be excluded. Finally, the first blood sample was taken on the day of laboratory diagnosis of SABSI and not when SABSI was first suspected on clinical grounds. This time difference was variable but may have been at least two days, over which cytokine levels may have changed.

In conclusion, differential expression of certain cytokines, such as IL-6 and leptin occurs in patients with different clinical courses of SABSI. While acknowledging that further studies on larger patient groups are required, these cytokines may be considered as potential prognostic markers or potentially be modified to minimise the adverse effects of S. aureus infection in vulnerable patients. Additionally, the increased expression of RANTES in all groups over the course of SABSI suggests that it may play a novel role in response to this pathogen.

SMN participated in the conduct of the experiments described and produced the first draft. AT and JO’G participated in patient recruitment and sample collection, MH and ML facilitated the study, guided the ethics application at MMUH and provided critical appraisal of the manuscript, CG participated in data analysis and interpretation of the data and provided critical appraisal of the manuscript. HH and DFH conceived of the study, DH directed the experimental work and redrafted the manuscript. All authors approved the final manuscript.