Through improvements in supportive care measures over the last few decades, such as those guidelines based on the ‘Surviving Sepsis Campaign’ and early goal-directed therapy trials, we have been able to help most of our patients survive through the initial systemic inflammatory response period [1, 2]. Unfortunately, a significant portion of patients still have protracted hospital courses and eventually succumb to late nosocomial infections [3]. These late sepsis-associated morbidities are thought to be related to global host immunosuppression.

Though this is not a newly described phenomenon, there is still considerable controversy as to how this immunosuppression occurs [4, 5]. Many investigators, including ourselves, have demonstrated extensive lymphocyte and antigen-presenting cell apoptosis in secondary lymphoid organs [6–8]. In addition, the presence of T cell exhaustion, as well as the expansion of regulatory T cell populations, has been described in both septic humans and mice [9–11]. Also, we and others have demonstrated that the expansion of an immature myeloid population, known as myeloid-derived suppressor cells (MDSCs), occurs following sepsis [12, 13]. However, the question remains, which of these phenomena, if any, contribute to or are critical for persistent inflammation and the late global immunosuppression that leads to protein wasting and the increased prevalence of nosocomial infections in our septic patients?

Despite the large body of data that exists looking at each one of these cell populations, there is no consensus, and the answer is more than likely multifactorial. In addition, though the presence of these cell populations may be associated with late mortality and morbidity in sepsis, the role they play in the early response to sepsis may be critical to mitigate the deleterious effects of the systemic inflammatory response syndrome. This is more than likely true of the MDSC, which despite its well-publicized deleterious effects on tumor immune surveillance, may actually be a critically important component of the innate immune response following septic insult.

Originally described in the cancer literature, it is now clear that MDSCs, roughly characterized as CD11b+Gr-1+ cells, arise in many pathologic processes including autoimmunity and burn/trauma injury [12]. MDSCs are purported to suppress immunosurveillance during tumorigenesis, through the production of arginase, reactive oxygen species (ROS), and/or inducible nitric oxide, though their role in sepsis is less clear. Delano et al. [12] originally described a dramatic expansion of MDSCs following murine polymicrobial sepsis and demonstrated that these cells may indeed suppress adaptive immune responses. However, since then, several groups, including ourselves, have demonstrated that this expansion of MDSCs during sepsis may actually be protective [13, 15]. The current report in Intensive Care Medicine by Derive et al. [14] explores this subject further.

Though many investigators have ascribed one standard phenotype to MDSCs, Derive and colleagues demonstrate a phenotypic change in the CD11b+Gr-1+ cells isolated at day 3 following sepsis compared to the CD11b+Gr-1+ isolated 10 days after polymicrobial sepsis. We have also shown such a phenotypic change in neutrophils (CD11b+Gr-1+ cells) from the bronchiolar lavages of septic mice and shown that neutrophils isolated at day 3 following sepsis produced significantly less ROS compared to neutrophils isolated 7 days following sepsis and were less mature [15]. Derive et al. demonstrate a similar decrease in ROS in response to IL-6 at day 3 but an increase in ROS production in MDSCs isolated from mice 10 days post sepsis. They also show that at day 3 MDSCs produce significantly less cytokines in response to LPS stimulation than MDSCs isolated from mice at day 10 post sepsis. These data together with gene expression array data from MDSCs isolated from mice either at day 3 versus day 10 following sepsis suggest that CD11b+Gr-1+ cells isolated 3 days following sepsis expand but are not capable of being activated until days 7–10.

In addition, the authors also demonstrate that MDSCs isolated from mice at day 10 following sepsis are able to attenuate inflammatory cytokines locally, facilitate bacterial clearance, and overall improve survival to polymicrobial sepsis. These data are similar to the findings of Sander et al. [13], who showed that the MDSCs that expand following septic insult suppress inflammatory responses and may indeed be protective. However, taken together with the data that demonstrate that MDSCs suppress T cell responses, it is easy to see how the robust expansion of these cells that occur following sepsis may also contribute to adaptive immunosuppression.

Unfortunately, these data also exemplify the difficulty in defining a discrete novel population in the MDSC from traditional innate immune effector cells, such as the neutrophil, using only the cell surface markers, CD11b+Gr-1+. In the adult, CD11b+Gr-1+ can encompass neutrophil, monocyte, and MDSC populations, which are clearly heterogenous, comprised of both granulocytic and monocytic subsets [16]. The authors do demonstrate that these MDSCs can suppress adaptive immune responses, which is one of the defining characteristics of the MDSC; but again, are the authors looking at the MDSC or the more mature neutrophil? Is there a difference between these two cell populations or do the neutrophils that expand following the initial ‘left’ shift that typically follows severe infection represent a compartment of these MDSCs? More than likely, as the data the authors and other investigators have suggested, the MDSC and the phenotypic changes that occur following sepsis are part of the primary myeloid compartment reconstitution that occurs following infection, otherwise known as emergency myelopoiesis [17]. As ‘MDSCs’ emerge from the bone marrow in an attempt to localize and clear the infectious insult, they represent a continuum from immature to mature activated innate immune effector cells that are, at least in the mouse, protective to the initial septic hit.

However, is the absence of activated MDSCs in some patients following sepsis the reason why our patients subsequently succumb to nosocomial infection? Is it the expansion of MDSCs in the host following sepsis that leads to global adaptive immune suppression and secondary infection in the sepsis patient? The answer is unknown, as the corollary for the MDSC in human sepsis has yet to be identified. Once this has been established, more work will be needed to understand clearly what their role is in our critically ill patient populations and whether or not the MDSC is truly the friend or the foe. More than likely, it will be both.