Németh et al. reply:

A typical bench-to-bedside path in medicine is to first discover specific targets using basic science and then develop therapeutic agents that are tested clinically. In the field of bone marrow stromal cells (BMSCs; also referred to as mesenchymal stem cells or MSCs), this course was reversed—running from bedside to bench. BMSCs, which are known to have immunoregulatory properties, have been used successfully in humans to combat graft-versus-host disease (GVHD)1. Our studies were based upon the clinical observations of Ringden and co-workers that BMSCs do not have any adverse effect, and might even be beneficial, when given to humans suffering from peritonitis2. They treated two patients suffering from severe, antibiotic-resistant peritonitis with mismatched allogeneic BMSCs. The peritonitis disappeared after the BMSC infusions; one patient died 4 months later of a fungal infection, but the other patient recovered. The mechanism(s) of action of BMSCs were unknown, however. Thus, our goal was to learn how BMSCs act using a mouse model of sepsis: cecal ligation and puncture (CLP). As we recently reported in these pages3, BMSCs, when given at the time of surgery or soon afterward, rescue mice from the lethal effects of CLP surgery. After coming in contact with them, BMSCs release prostaglandins to reprogram macrophages to induce their synthesis and secretion of interleukin-10 (IL-10). Monneret is absolutely right about the need for caution in making predictions about the efficacy of human treatments based on experiments done in mice; whether the BMSCs act in the same way in humans as they do in the mice remains to be determined, and better animal models of sepsis are surely needed (see, for example, ref. 4). What should already be clear from our work, however, is that the cells are 'smarter' than drugs. In the CLP model, their actions could not possibly be mimicked by systemic administration of prostaglandins, which would have a myriad of side effects. Monneret stresses the fact that “everything is thus a matter of timing.” He is right; this is why cellular therapy might be superior to drug therapy. The BMSCs appear to “think globally, but act locally,” providing assistance to only those cells that need it, where and when they need it—and most likely, this depends on the signals they receive at any given time. Furthermore, it is conceivable that BMSCs may help different cells in different ways, secreting agents in response to a variety of environmental cues and rebalancing the innate immune response to maximize its utility to the host. Ultimately, we will learn from clinical studies whether these hypotheses are correct.