Introduction

The recent issue of Critical Care includes a report by the group ofHotchkiss on the improved survival of primary and secondary fungal sepsis bytargeting the negative co-stimulatory molecules programmed death-1 receptor (PD-1)and chronic T-lymphocyte activation antigen 4 (CTLA-4) in mice [1].

What is the rationale behind and how far we are from treating patients using thisapproach?

PD-1 and CTLA-4 are negative co-stimulatory molecules on immune cells. Their ligandsare PD-1 L and CD80/86, respectively, which are expressed on activated immunecells but also on many nonimmune tissues. In contrast to positive co-stimulatorymolecules (such as the CD80/86 ↔ CD28 pathway), cross-linking of negativemolecules on immune cells, especially T cells, by their respective ligandssuppresses immune cell activation or even induces apoptosis resulting inlymphopenia. Genetically, deficiency of negative regulatory molecules results insevere hyperinflammatory and autoimmune diseases, demonstrating the key function ofregulatory pathways for maintaining immunological homeostasis under differentchallenging conditions [2, 3]. In fact, a successful pregnancy is not feasible without activation ofthose negative regulatory pathways to allow the acceptance (tolerance) of thesemi-allogeneic foetus. Similarly, dominance of regulation is the major aim oftransplant immunology to induce transplant tolerance and long-term drug-freeallograft survival.

Too much of a beautiful thing, however, limits the protective immune responsiveness.First, it was reported that tumours downregulate the adaptive immune response viaactivating the negative co-stimulatory pathway that helps the tumour to escape fromimmune surveillance or even to gain growth support by immunologically deactivatedstroma. Overexpression of the ligands of negative co-stimulatory molecules, likePD-1 L, by tumour cells or tumour-surrounding stroma cells mediatesanergy/tolerance of tumour-specific immune cells, particularly T cells. Similarly,chronic virus infections such as HIV are associated with deactivated T cellsoverexpressing negative regulatory molecules. In addition, regulatory T cells thatcontrol immune responsiveness are using those molecules for executing theirregulatory activity.

Targeting negative regulating molecules (CTLA4, PD-1/PD-1 L, CD25) is thereforea novel therapeutic option to reverse undesired immune silencing. In fact, biologicstargeting those negative molecules result in promising data improving anti-tumourimmune response and improving outcome in both experimental models and tumourpatients [4, 5].

There is still a high unmet medical need to improve the outcome of severesepsis/septic shock from both the medical and the health-economical points of view.All efforts to target hyperinflammation by broad or specific anti-inflammatory drugsfailed in phase II/III clinical trials. This concept of anti-inflammatory therapy ofsepsis was developed on the basis of the preclinical models of endotoxin-inducedseptic shock – models that are obviously not predictive for theimmunopathology of sepsis in the majority of critically ill patients.

We could show almost 25 years ago that poor outcome of sepsis is associatedrather with immune dysfunction in both immunosuppressed and nonimmunosuppressedpatients, particularly if sepsis is established for several days [6, 7]. The severest form of immune dysfunction, so-called immunoparalysis, isdefined as diminished monocytic HLA-DR expression of <30% (or <8,000molecules/cell by the new Quantibrite method) and ex vivolipopolysaccharide-induced TNF secretion of <300 pg/ml. Monocyticdysfunction reflects the failure of the adaptive immune system, particularly of Tcells. Mediators of T cells (IFNγ and granulocyte–monocytecolony-stimulating factor from T-helper type 1/T-helper type 22 cells were able torestore immunoparalysis both in vitro and in vivo[7]. Moderate post-traumatic/surgical dysfunction predicts risk of infectiouscomplications and sepsis. Adaptive transfer of IFNγ-secreting T cells couldreverse immune dysfunction following experimental stroke and preventstroke-associated pneumonia [8]. These and similar preclinical and clinical data from our group and othergroups support the concept of immune monitoring-guided rebuilding immunocompetenceas a new concept of sepsis prevention/treatment [6–9]. This view is further supported by the recent excellent postmortem studyof patients who died from sepsis and multiorgan failure demonstrating strong T-cellexhaustion [10].

Data from several groups demonstrated an overexpression of PD-1/PD-1 L as wellas CTLA4 on immune tissue but also on some nonimmune tissues in septic animal modelsand patients [10–12]. Consequently, targeting negative molecules in sepsis can reverseimmunoparalysis and improve survival in experimental sepsis as also shown by thegroup of Hotchkiss and others [13, 14]. An editorial by Goyert and Silver discusses PD-1 as a new putativetarget for sepsis treatment [15]. The positive effects of PD-1 targeting in different models strengthenthis idea. In this direction, the very recent report by Chang and colleagues inCritical Care nicely demonstrates the beneficial effect of PD-1targeting in clinically relevant preclinical models of primary and secondary(post-caecal ligation and puncture) candida sepsis [1]. The results are striking and pushing forward the concept of immunereconstitution as a new option.

What might be the next steps?

Proof-of-concept trials in patients are still missing. The tools are availablebecause different pharmaceutical companies developed them for oncology. However,encouraging big pharma companies to perform any studies in sepsis has been verydifficult because of the failed clinical trials in the past. The data are sopromising that clinical trials are a must. We can only recommend that enrolledpatients are defined very well by applying standardised immune monitoring tostratify patients into those suffering from immunoparalysis – it makes nosense or can even be harmful to push adaptive immunity in immunocompetent patientsduring the dominant hyperinflammatory phase. Nobody would give insulin withoutglucose monitoring to prevent useless treatment of normoglycaemic or hypoglycaemicpatients.