Erythropoietin (EPO), a member of the type I cytokine superfamily, was first identified as the hormone that stimulates erythroid progenitors within the bone marrow to mature into erythrocytes. In recent years, however, many other physiologic roles for EPO have been identified. EPO is now known to be a local product of diverse cells that specifically protect cells from potential cytotoxic events (for review, see Erbayraktar and coworkers 1). In this capacity, EPO maintains and protects tissue function, especially during metabolic stress.

The behavior of the classical EPO–erythroid precursor system in serious illness is reasonably well understood. Typically, both the production of EPO and its action in the bone marrow are impaired by multiple factors (e.g. circulating EPO-suppressing proinflammatory cytokines 2), resulting in anemia. An exception to this generalization is observed in acute renal failure, in which the systemic concentrations of EPO are transiently increased, presumably as a result of unregulated release of EPO from injured EPO-producing cells within the renal interstitium 3. However, increases in circulating EPO following renal failure do not usually reach the minimum concentration required for effective paracrine–autocrine signaling in preclinical models (see below).

The results of multiple clinical studies have shown that pharmacologic doses of recombinant human erythropoietin (rhEPO) effectively reactivate the bone marrow in critical illness to produce erythrocytes. Although blood transfusions can be avoided in rhEPO-treated patients, clinical trials to date have shown no differences in patient survival or recovery (e.g. 45). In one small study performed in a multidisciplinary intensive care unit 6, however, the length of stay was a third shorter for those patients who received rhEPO.

Although EPO that is produced in an autocrine–paracrine manner has been implicated in tissue protective effects in the brain, spinal cord, retina, and heart, similar protective roles in severe illnesses have not been directly evaluated. Notably, published clinical trials have focused on erythrocyte production and thus were not designed to assess potential benefits of rhEPO on survival or recovery unrelated to treatment of anemia. A number of preclinical models that mimick aspects of multiple organ dysfunction syndrome (e.g. splanchnic artery occlusion induced shock 7, ischemic renal damage 8, and intestinal injury 9) are ameliorated by rhEPO, suggesting other potential roles for rhEPO in critical illness. In this article, we review probable contributions of the nonclassical EPO system to physiologic conditions associated with severe illness. We conclude by outlining several essential parameters to be considered when designing clinical trials to evaluate potential tissue protection by EPO in critical illness.

EPO is a tissue protective cytokine that mediates local (innate) stress responses 101112. The innate stress response system evolved to counteract invasion by infectious agents. In this biologic adaptation, a nidus of infection is rapidly populated by macrophages that secrete inflammatory cytokines, which in turn both trigger apoptosis and recruit additional macrophages. The net result of this apoptotic feedback loop is an amplification of injury involving 'innocent bystander' cells, sterilizing the region surrounding the pathogen. Although this approach is efficient for microbes, an identical response is activated by other insults (e.g. metabolic stress). In this case, the innate stress response is maladaptive because viable tissue is irreversibly injured.

Multiple organs and tissues express EPO and its receptor (EPOR), implicating both in the local stress response system 1314151617181920. The tissue response to stress is characterized by an increase in EPO and EPOR within the penumbra of injury (i.e. the region at risk for cell death). In cerebral ischemia, for example, a rapid and marked upregulation of EPOR occurs, followed only later by an increase in local EPO production 101112. These two processes prevent the spread of injury by neutralizing the apoptotic program initiated by exposure to proinflammatory cytokines such as tumor necrosis factor-α and interleukin-1, among others 21. Therefore, when using exogenous EPO as a tissue protective cytokine, it is crucial to administer it early in order to activate existing EPORs expressed by viable cells within the penumbra, thus abrogating apoptosis.

Many preclinical data support the concept of early rhEPO administration for tissue protection. First, the powerful mechanism of ischemic preconditioning (increased tissue protection by a brief pre-exposure to nontoxic stressors) depends on EPO upregulation within the affected tissues 11222324. Preconditioning occurs following exposure to a wide variety of stressors in addition to hypoxia and ischemia, including lipopolysaccharide, seizures, and exposure to excitotoxins. Second, many tissues injured by ischemia, mechanical trauma, excitotoxins, and other stressors are significantly improved by administration of rhEPO following injury (reviewed by Erbayraktar and coworkers 25 and by Beumi and coworkers 26) in multiple species, including humans 27. Third, rhEPO has been associated with improved residual tissue function (e.g. following myocardial infarction in rats 28). Notably, a few clinical trials have been conducted using rhEPO in chronic or subacute conditions, and these demonstrated improved clinical status after rhEPO administration. For example, the effects of rhEPO administration to patients experiencing severe congestive heart failure include a significant improvement in exercise tolerance, as well as reduced need for hospitalization and diuretics 29303132. However, these studies were not designed to assess the effects of rhEPO independent of increased hemoglobin concentrations. This distinction is important because preclinical studies conducted in experimental models 17333435 have shown direct (i.e. without increases in serum hemoblogin) beneficial effects of rhEPO on myocardium, including improved remodeling following ischemic injury.

As stated above, the principle mechanism whereby EPO confers tissue protection involves the modulation of cellular apoptosis within the penumbra (region at risk). Because apoptosis is an active genetic expression program, a significant time window exists within which it can be terminated. Briefly, agents that can prevent apoptosis can be effective long after the injury has occurred. This phenomenon was corroborated by EPO tissue protective studies. One impressive example is the spinal cord, in which waves of apoptosis occur for days after a mechanical injury has been sustained 36. Notably, rhEPO administered even 24 hours after injury is very effective in ameliorating injury (Fig. 1). In contrast, ischemic experimental brain injury is condensed in its response, and so the window of opportunity is only about 3–4 hours 37. In addition to modulating apoptosis, EPO maintains the integrity of capillary function (e.g. the blood–brain barrier 38). Therefore, the potential contributions of the size of the therapeutic window must be considered based on available preclinical data.

The presence of a therapeutic window dictates specific time constraints for efficacious administration of exogenous EPO as a tissue protectant. It is noteworthy that, in the larger clinical studies of critical illness conducted to date, administration of rhEPO was not initiated until 3 days after admission to the intensive care unit 45. This delay in rhEPO administration, which is even longer considering the time of onset of illness, cannot reasonably be expected to provide elevated rhEPO levels within the therapeutic window specific to an organ or tissue.

Because critical illness is so heterogeneous, selection of patients and illness is of utmost importance. As summarized above, rhEPO has been shown to be particularly effective in conditions in which apoptosis plays a major etiologic role. Ischemic injury to nervous tissue, the heart, and the kidney is attenuated following administration of rhEPO within hours, but not days, of the insult. For example, available preclinical data of experimental stroke suggest that EPO is much less effective if it is administered later than 6 hours after the insult 37. Therefore, studies concerning rhEPO administration in the setting of critical illness should begin before or immediately after admission to the intensive care unit, with an enrolment cutoff time of perhaps 5 hours.

The serum concentrations of EPO required for tissue protection are higher than those required for erythropoiesis. One reason for this is that the receptor for tissue protection exhibits a lower affinity (approximately 1000-fold) as compared with erythroid progenitors 39. Another reason may be the presence of blood–tissue barriers such as for the brain and spinal cord. Preclinical data suggest that the minimum therapeutic level needed for protection against tissue injury appears to be in the order of 300–500 mIU/kg body weight (intravenously or intraperitoneally) for the organs that have thus far been adequately investigated 3740. The successful phase II clinical trial in stroke 27 employed a dose within this range. This requirement means that, from a practical perspective, only intravenous dosing routes should be contemplated for clinical studies.

A number of documented and theoretical problems have been associated with administration of high dose rhEPO. Most notably, rhEPO interacts with thrombocyte production and activates endothelial cells to augment platelet aggregation, increasing the likelihood of microinfarctions and macroinfarctions 4142. In confirmation of this, several recent clinical trials in which rhEPO was administered to cancer patients were terminated following increased symptomatic or fatal thrombosis in patients receiving the drug 43444546. Additionally, cancer patients who receive chemotherapy administered via catheters may have an especially markedly increased risk for thrombosis 47. The relevance of this phenomenon to critically ill patients who may receive limited dosing has not been determined, but it may be important for certain high risk patients. However, limited available data suggest that dosing in acute settings is not likely to be detrimental 27.

Additionally, many tumors express EPOR 4849, and several large clinical trials have clearly shown adverse outcomes following administration of EPO to cancer patients 4446. The potential hazard of acute dosing of EPO to critically ill patients not known to have a tumor burden is probably exceedingly minor, considering the potential benefits of treating a life-threatening disease.

Preclinical experience strongly suggests that rhEPO will confer significant tissue protection in the setting of critical illness. Appropriate clinical trials will require administration of these agents within a reasonable period of time, coincident with the therapeutic window that is characteristic of each organ and tissue considered in the study, and at doses high enough to engage the paracrine–autocrine receptors. The biologic substrate on which EPO acts is a conserved evolutionary adaptation designed to preserve the organism at the expense of the local tissue beds, which EPO antagonizes. It is important to realize that the vast therapeutic armamentarium available to modern medicine reduces significantly the protective role of the innate stress response. In this setting, salvage of tissue that is otherwise normal but has been injured by the host specific innate stress response can probably be achieved by administering tissue protective cytokines such as rhEPO. Only properly structured clinical trials can answer these questions. However, based on much preclinical data and limited human studies, it is clear that tissue protective cytokines are highly promising agents that may provide new therapeutic options in many forms of tissue injury.

None declared.

EPO = erythropoietin; EPOR = erythropoietin receptor; rhEPO = recombinant human erythropoietin.