Anaemia in Surgical Sepsis and Stress: The Roles of Erythropoietin, Iron and Steroids

Background: Inflammation in surgical sepsis and stress frequently causes anaemia, leading to increased rates of blood transfusions. Recent evidence shows that blood transfusions carry a greater risk for shortand longterm complications than previously thought. Objective: To review the role of erythropoietin (EPO), iron and/or steroids as an alternative treatment to blood transfusions in critically ill patients. Methodology: A systematic review was prepared from recent literature on inflammation-induced anaemia, anaemia in the critically ill and/or septic patient and the roles of EPO, iron and corticosteroids in these patients. A meta-analysis was completed for EPO. Results: Inflammatory cytokines alter haematopoietic and biochemical pathways, leading to anaemia. Inflammation decreases circulating EPO and upregulates hepcidin, resulting in decreased free iron. Twelve randomised-controlled trials demonstrate that EPO administration in critically ill patients reduces the need for blood transfusions by 31% (p=0.005) however does not significantly decrease mortality (p=0.15). Intravenous iron also reduces the need for blood transfusions but has not been utilised in sepsis-associated anaemia. No trials focusing on the effects of steroids on sepsis-associated anaemia were identified. Conclusion: Due to the lack of data specific to sepsis-associated anaemia in post-operative patients, the roles of EPO, iron and steroids remain under investigation. More research specific to surgical patients is needed.


Introduction
Post-operative complications are a common concern, of which sepsis has the highest incidence and attracts an overall mortality of 5.4% [1]. Patients who develop sepsis often develop significant anaemia, which leads to an increased rate of blood transfusions [2]. Blood transfusions are not without risk and are associated with morbidity and mortality. The leading cause of mortality following allogeneic blood transfusion includes transfusion-related acute lung injury, febrile reactions, ABO and non-ABO haemolytic transfusion reactions and transfusion associated sepsis [3]. The rate of febrile non-haemolytic reactions is 1% while severe acute haemolytic reactions occur at a rate of 1:76,000. The rate of ABO incompatibility is 1:40,000 [4,5]. Intraoperative transfusions occurred in 3.8% of general surgical patients and were associated with increased risk of 30-day mortality, pneumonia and sepsis [6]. In other surgical patients, blood transfusions have been associated with delayed wound healing, increased long-term mortality risk and increased hospital stay [7,8]. The negative transfusion related immunomodulatory (TRIM) effects have also been associated with increased risk of nosocomial infections and the development of autoimmune disease later in life [9,10,11]. Unfortunately, RBC transfusions have also been found to promote tumour recurrence in colorectal, hepatocellular and possibly other cancers [12,13,14].
In light of this growing body of concerning research, there has been a move to implement strategies to decrease transfusion rates. Recent studies have showed that simply lowering the haemoglobin (Hb) threshold for transfusion is an effective way of reducing transfusion rates in critically ill patients without significantly affecting mortality [15]. The Transfusion Requirements in Septic Shock (TRISS) trial included 998 septic patients who were randomized to have a transfusion threshold of 70 g/L or 90 g/L [15]. The data showed that the lower threshold group received a median of 1 transfusion, while the higher received 4 transfusions [15]. Importantly there was no significant difference in 90-day mortality (p=0.44, 95% CI 0.78-1.09) [15]. As the target is to continue to decrease transfusion rates in critically ill patients, it is important to understand the underlying physiology and explore further strategies to reduce exposure to blood products. This review will discuss the pathophysiology that underpins sepsis-associated anaemia and the possible alternative treatment options.

Methods
A literature search was conducted using CINAHL, MEDLINE, EMBASE and the Cochrane Library databases from the years 1950-2014. Additional articles were identified through manually reviewing references from articles that were retrieved. For pathophysiology of anaemia in septic patients the Mesh search terms and keywords used in this study included: "sepsis", "systemic inflammatory response syndrome", "anemia", "immune response", "cytokines", "inflammatory mediators". For intravenous iron and steroids in post-operative septic patients the search strategy used medical subject heading terms and text words, including: "sepsis", "sepsis syndrome", "septicemia", "systemic inflammatory response syndrome", "SIRS" AND "postoperative", "perioperative" AND "intravenous iron", "ferric compounds", "ferrous compounds", "perioperative" AND "steroids", "corticosteroids", "adrenal cortex hormones", and "glucocorticoids". The search was limited to articles published in English.

Systematic Review and Meta-analysis on
Use of EPO in Critically Ill Patients

Search Strategy
A systematic search was carried out to include studies between 1950 and April 2014, using CINAHL, MEDLINE, EMBASE and the Cochrane Library databases. The medical subject heading terms and text words included: "sepsis", "sepsis syndrome", "septicemia", "systemic inflammatory response syndrome", "SIRS" AND "postoperative", "perioperative" AND "erythropoietin". The complete search strategy is presented in Appendix 1. Additional papers were discovered by manually searching through the reference lists of relevant publications.

Study Selection and Search Results
The following protocol was used to identify studies for inclusion in the systematic review: 1. Patients admitted to the intensive care unit 2. Patients > 1 year of age 3. Patients receiving scheduled doses of an erythropoietin receptor agonist 4. Randomised clinical trial 5. Any comparator: none, placebo or other medications The following protocol was used to identify studies for exclusion in the systematic review: 1. Neonates and infants 2. Uncontrolled trials Of the 174 records identified, 162 were excluded using the above inclusion and exclusion criteria. This resulted in 12 remaining studies which were included for meta-analysis.

Data Extraction
Data were extracted by two reviewers and cross-checked to reach consensus. Disagreements were settled by discussion and consensus. Study characteristics, baseline characteristics of the study and control groups and the characteristics of intervention were extracted. Outcome measures included: mortality, ICU and hospital length of stay, incidence of bloodstream infection, rate and total amount of red cell transfusion. Complications extracted included thrombosis, hypertension, stroke and myocardial infarction.

Data Synthesis
Discrete and continuous data were analysed using the Cochrane Review Manager (version 5.3.3). The Mantel-Haenszel test was applied to our fixed-effects model to develop summary measures of effect. Intention-to-treat analysis was performed. For dichotomous data, we expressed summary measures of effect across studies as odds ratios with 95% confidence intervals. For continuous data such as units of red blood cells received, measure of effect was expressed as weighted mean differences with 95% confidence intervals. Statistical heterogeneity was assessed using the I 2 statistic.

Pathophysiology of Anaemia in Surgical Stress and Sepsis
Critically ill patients, including those with sepsis, develop anaemia in up to 70% of cases, of which 44-50% will require a blood transfusion [16,17,18]. In septic patients, blood transfusion are required in up to 50% of cases due to a normocytic, normochromic anaemia with low iron, low transferrin but high ferritin, which usually develops through altered iron metabolism and erythropoiesis [17][18][19][20][21][22]. This is due to the inflammatory response stimulated by surgical sepsis and stress leading to elevated inflammatory markers, which inversely correlates with Hb levels [21,23,24].

Altered Erythropoiesis
Rogiers et al. showed that non-septic critically ill patients had a normal correlation between haematocrit and EPO levels, while septic patients developed more severe anaemia and had lower circulating EPO levels [20]. The inflammatory cytokines IFN-gamma, IL-1 and TNF-alpha have been implicated in the alteration of EPO production [19,25,26]. These inflammatory markers inhibit hypoxiainduced EPO production [25]. High TNF-α levels inhibit erythropoiesis by suppressing haematopoietic stem cells [27]. Conversely, IL-6 showed a stimulatory effect on EPO production and was able to reverse the inhibitory effects of other inflammatory cytokines [25]. These inflammatory cytokines, as well as oxidative stress pathways, have also been implicated in impaired red cell longevity in septic patients [28].

Altered Iron Metabolism
In septic patients a functional iron deficiency can lead to ineffective erythropoiesis [2,18]. This phenomenon may be partly explained by the hormone, hepcidin [29]. After iron is absorbed by enterocytes, the haem is catabolised and pumped through a basal transporter (ferroportin-1), which is inhibited by hepcidin [30,31]. Therefore, hepcidin causes ferritin to accumulation within these cells [30].
There is a significant increase of IL-6 in hypoxic conditions, which induces EPO and hepcidin [25,32]. IL-6 has also been implicated as the main inducer of hepcidin in septic patients, causing a subsequent drop in serum iron levels and therefore impairing Hb production [33,34]. Furthermore, patients who needed transfusions had significantly higher hepcidin levels than patients who did not require transfusions [34].

Induction of Stress Erythropoiesis
In response to the physiological stress in hypoxia and sepsis, the rate of erythropoiesis can increase by up to 10-fold. This process is known as stress erythropoiesis and the regulators include erythropoietin and glucocorticoids [35]. It has been demonstrated in vitro that glucocorticoid availability, as well as effective binding of glucocorticoids to the glucocorticoid receptor, is required for stress erythropoiesis [36].
Glucocorticoid levels are often suppressed in patients with sepsis due to adrenal insufficiency that occurs in 30-70% of septic patients [37]. This likely occurs due to poor venous drainage and increased arterial blood flow in the glands, decreased cortisol-binding globulin and glucocorticoid receptors [37,38,39].

Erythropoietin
As EPO deficiency has been detected in critically ill patients, it has been hypothesised that administering EPO may improve Hb levels. Although there have been no studies specific to post-operative patients, an array of randomised controlled trials have been performed to assess the clinical benefits and possible harms associated with EPO in burn, critically-ill, trauma and cardiothoracic patients (Table 1). These 12 RCTs, containing 3,716 patients in total, have been systematically reviewed and a meta-analysis completed for both transfusion independence and mortality (Figure 1 and Figure 2). Blood transfusion independence is defined as the ability of EPO to prevent the need for at least 1 red blood cell transfusion. This was assessed by comparing the percentage of patients who received any RBC transfusion within the study period. The study periods used to determine mortality and transfusion differences were between 21 -42 days, except for Luchette et al who collected data at 12 weeks.
Two studies showed that EPO was able to significantly increase reticulocyte counts, while three reported significantly increased Hb levels [40,41,42,43,44]. Corwin et al. showed a significant reduction in total units transfused (p<0.002) [45]. However, the total number of patients that received blood transfusions was not significantly different between the two groups [45]. Their second study also showed significant reduction in the total number of units transfused and total number of patients requiring transfusion (50.5% vs. 60.4%; p<0.001) [46]. Georgopoulos and colleagues showed that both EPO treatment groups had lower transfusion rates (37% and 27% vs. 58%) [42]. Silver et al. also showed a significant reduction in the percentage of patients requiring a transfusion in a long-term acute care setting [43]. Corwin et al. demonstrated a significantly lower 29-day mortality (8.5% vs. 11.4%, p=0.04) [44]. Five of the 12 studies, including the two RCTs which only utilized one dose of EPO, showed no benefits across all parameters [47][48][49][50][51]. Thromboembolic disease was the most commonly recorded adverse event and was only significantly increased in one study (8.7% vs. 5.8%, p=0.04) [44].  The data in Figure 2 shows no significant difference in mortality between the groups (p=0.15). In the trials reviewed, the EPO group showed an overall reduction in mortality of 13%, however, the confidence interval could not exclude and increase in mortality up to 5% (95% CI 0.72-1.05) ( Figure 2). Interestingly, Figure 1 illustrates that the EPO group showed a 31% increase in transfusion independence (p=0.005, 95% CI 0.53-0.89). However, due to the lack of studies specific to surgical sepsis and stress, variability in EPO dosing and study endpoints the heterogeneity of these studies must be noted.

Iron
A recent systematic review of 72 studies, involving 10,605 participants, demonstrated IV iron was associated with a significant increase in mean Hb concentrations compared with oral iron or no iron supplementation [52]. Furthermore, they showed that IV iron therapy was associated with a significant reduction in transfusion requirements, particularly when combined with EPO (p=0.04) [52]. Unfortunately, IV iron was associated with a significant increase in the risk of infection (1.33, 95% CI) [52].
There is limited clinical data available on the use of iron therapy in post-operative septic patients and critically-ill patients. This likely stems from the concept of iron having potentially toxic effects through the formation of reactive oxygen species and increasing iron availability, which may facilitate infection [53].
It has also been suggested that IV iron administration can increase hepcidin levels, therefore exacerbating the anaemia [54]. Recently, there has been a wave of research into novel therapies that target the hepcidin-ferroportin axis [30]. Although the majority are still within Phase 1 and 2 clinical trials, there has been some success in animal models with direct hepcidin antagonists, hepcidin production inhibitors and ferroportin stabilizers [55].

Steroids
The use of steroids as an adjunctive therapy for patients with severe sepsis and septic shock has fluctuated over the last few decades. However, a recent systematic review by Annane et al. demonstrated a reduction in patient mortality (p<0.02) and increase in shock reversal (p<0.02), following low-dose corticosteroid treatment [56].
Unfortunately there is limited data available on the effect of steroid intervention on septic patients' Hb levels, although the RCT by Annane et al. demonstrated no significant improvement in patients receiving steroids (100g/L vs. 102g/L) [57]. Glucocorticoid dependent stress erythropoiesis is a major physiological factor in maintaining haemoglobin levels in septic patients [35,36].
When up to 70% of these patients may in fact be deficient in glucocorticoids due to adrenal insufficiency, it will be important to assess whether steroid replacement therapy could benefit through maintenance of stress erythropoiesis [37].

Discussion
Negative transfusion outcomes may be related to TRIM effects which have been linked to higher risks of nosocomial infections, development of autoimmune disease and recurrence and progression of solid malignancies. The cellular mechanisms involved seem to provide an overall immune inhibitory effect which potentiates these damaging pathological processes. Unfortunately, these detrimental effects are imposed on critically ill patients if their Hb falls to a transfusable level.
The anaemia of critical illness is a complex process that has three main contributing pathways. Firstly, expression of inflammatory cytokines cause decreases in EPO transcription, mRNA and haematopoietic stem cell potential as well as blunting erythroid maturation. Secondly, inflammatory cytokine IL-6 has been demonstrated to be a strong inducer of hepcidin, which subsequently promotes the development of a functional iron deficiency. Finally, specifically in septic patients, adrenal insufficiency can lead to the lack of endogenous glucocorticoids and consequently hinder the process of stress erythropoiesis. In light of these mechanisms, EPO, iron and steroids could be considered as alternative treatments.
EPO has been validated as effective treatment for anaemia associated with multiple illnesses, however, its efficacy in critically ill patients has been a topic for debate. This review shows a benefit for the use of EPO to reduce transfusion requirements and therefore reduce the need for harmful PRBC transfusion. Importantly, there was no significant increase in mortality in the EPO group. However, lack of efficacious sepsis-specific data has also led to the marked heterogeneity of studies illustrated in this review. Currently, guidelines are not recommending the use of erythropoietin in sepsis-associated anaemia due to this lack of specific data [58,59]. It may also be hypothesised that exogenous EPO is of even lesser value in septic patients due to inflammatory cytokine suppression of other steps of the erythropoietic pathway [20].
IV iron is effective in raising Hb levels in many clinical settings, however, the trials are lacking in septic patients. This is likely due to the assumption that increasing levels of iron could potentiate ROS (reactive oxygen species) formation and bacterial infections [53]. However, this risk has been thoroughly investigated in animal models and more recently in haemodialysis patients to show no significant correlation with bacteraemia [60,61]. Unfortunately, it has been demonstrated that IV iron administration may increase hepcidin levels and therefore exacerbate the iron withholding mechanisms which are activated during anaemia of inflammation [54]. Hepcidin targeting therapies are currently under evaluation and may provide an exciting and effective treatment that targets one of the root causes of inflammation-associated anaemia.
The role of steroids in the treatment of anaemia in surgical sepsis and stress has not been explored. Recently the use of low-dose hydrocortisone has become recommended in severe sepsis and septic shock, which may provide excellent opportunities to investigate whether steroids have an effect on Hb levels and transfusion rates.

Conclusion
This meta-analysis shows that the use of EPO in critically ill patients can significantly decrease transfusion requirements. However, further studies are needed to understand the pathophysiological mechanisms underlying the development of inflammation induced anaemia in the surgical patient. To date there have been limited clinical trials in these patients, therefore, further clinical trials are required to assess the effectiveness and safety of EPO, IV iron and steroids.