Blood as a target for adjuvant therapies, such as 'purification methods', for the treatment of acute inflammation related to severe infection has been proposed for many years, even with relatively low levels of proof [1–4]. Arguments for such an approach can be summarized as follows: almost all markers and cell phenotypes have been studied in blood; ex vivo studies using plasma from severe septic patients induce healthy cell phenotype changes mimicking those observed in septic cells [5]; plasma ultra-filtrate coming from septic patients induces ex vivo modification of cell functions [6]; and plasmapheresis therapy has been shown to be efficient at controlling auto-immune disease [7]. If these blood purification techniques are technically feasible, the mechanism(s) by which they can improve outcome remains unclear, and the goals of these techniques have to be clarified: reduction of activated circulating cells and/or removal of mediators, or a combination of the two, need to be clearly demonstrated [8]. This key question guides development of the technical strategy to achieve this goal: cell reduction requires supports for trapping cells while mediator removal requires membranes that adsorb mediators or allow them to cross [9]. Since sepsis mechanisms are still incompletely understood, particularly the impact of changing the levels of mediators on outcome, it is difficult to define the adequate support and methods. In addition, inflammatory patterns during sepsis are not stable over time, moving rapidly from hyper-inflammation to immune-suppression [10].

The article from Rimmelé and colleagues [11] reports the consequences on cytokine profiles and cell-cell interactions of an ex vivo device to capture blood leukocytes and cytokines in blood from septic patients compared to healthy volunteers. This experiment reveals new information using a whole blood heparinized circulation device, with two different hemoadsorption devices (cytosorb with large polystyrene beads, cytosorb with small polystyrene beads) compared to a hemofilter (Oxiris®, Gambro-Hospal) as a control, flowing at 0.75 ml/minute. The surface polarity was modified to obtain a positively charged membrane to catch negatively charged endotoxins, when contact area was small for large beads. Using electron microscopy and immunofluorescence staining with confocal imaging, leukocyte capture can be characterized. Plasma cytokine levels were determined by the luminex technique, allowing a large panel of mediators to be measured. A large panel of leukocyte surface markers was studied by flow cytometry and appropriate antibodies. It was then possible to assess the effects of leukocyte removal on inflammation and immune function. The authors concluded that monocyte and leukocyte capture results in upregulation of interleukin-8 modulation of cell-mediated immunity. Both aspects, cell capture and mediator removal or release, appear feasible, but have to deal with time-dependent changes in the immune profile and also with the characteristics of circulating cells. Recent studies on functional genomic [12] and blood and tissue infiltrating immune cells [13] strongly suggest that the initial phase of sepsis or acute inflammation has a short duration, rapidly followed by quasi-constant immune depression during acute inflammation. Such induced immunodepression concerns all aspects of the inflammatory response, especially involving neutrophils, monocytes, and lymphocytes. The cell number, their function and life span become largely modified [9], with increasing life span of neutrophils, reduced function of monocytes and reduced life span and function of lymphocytes. The latter effect concerns the entire lymphocyte repertoire, except CD4+CD25+foxP3+ cells [14, 15]. These observations question the timing for using hemoadsorption in severe sepsis, since the delay between the onset of sepsis and a therapeutic decision might be highly variable. Looking at the results obtained, such an extra-corporeal device seems to perform well and is a promising method. For clinical use, it will then be essential to determine which patients, when, and for how long this strategy has to be applied. Such goals necessitate easy and repetitive immune monitoring, a major step to clarify. In addition, such a device has to work enough to overcome whole body production of mediators and cell phenotype changes. Considering the more or less selective removal of cells and mediators, the question might be: can plasmapheresis be a good alternative, especially if albumin is used as a replacement fluid? This hypothesis might be supported by the recently reported results of an Italian randomized clinical trial on albumin (Albios). The first report at the SMART meeting in May showed an outcome benefit in septic shock patients treated with albumin at a dose to obtain a plasma level close to 30 g/ml.