Iron therapy in chronic kidney disease: current controversies.

Anaemia in chronic kidney disease (CKD) is a complex disease that requires an integrated approach to incorporate both diagnostic and therapeutic interventions and to address the different facets of its aetiology and pathophysiology. The advent of erythropoiesis stimulating agents (ESA) has revolutionised the therapy of anaemia of CKD, and has resulted in a significant decline in the need for blood transfusions in CKD patients. The routine application of ESA has also led to the need for concomitant iron supplementation. ESA and iron therapy now form the cornerstone of anaemia management in CKD. Intravenous iron administration is effective with acceptable safety, and may improve ESA responsiveness. However, less is known about the long-term safety of iron supplementation in CKD patients. Whereas maintenance (weekly to monthly) intravenous iron has been routinely used in maintenance dialysis patients, iron replacement in patients with non-dialysis-dependent CKD is less well studied, in spite of the much larger number of patients affected. This review discusses iron supplementation in CKD with an emphasis toward controversial issues that continue to pose dilemmas in clinical practice. Concerns related to both the optimal amount of iron supplementation and to the safety of various agents available in clinical practice are presented.


INTRODUCTION
Anaemia is a common complication of chronic kidney disease (CKD) and is associated with poor outcomes including higher death risk in both maintenance dialysis patients (Regidor et al. 2006), and in individuals with non-dialysis-dependent CKD stages 3 to 5 . Since the introduction of the first erythropoiesis stimulating agent (ESA), anaemia management has become a core component of everyday nephrology practice. Whereas the application of ESA therapy has significantly decreased the need for blood transfusions in dialysis patients, it has also resulted in the widespread use of intravenous (IV) iron supplements.
The results of recent clinical trials examining the effects of ESA (Singh et al. 2006) have highlighted potential problems stemming from treatments approved for short-term indications (in the above referenced study, the correction of anaemia via high ESA doses) without proper assessment of their long-term safety. Similar concerns exist for IV iron preparations, which are all currently approved for the treatment of iron deficiency in various stages of CKD, but whose long-term safety continues to be debated. Optimal iron management involves the administration of enough iron to assure unhindered erythropoiesis (provided enough erythropoietin is also available), but avoids deleterious consequences that are either the result of too much iron (irrespective of the ther-apeutic agent), or, that are related to reactions that are specific to the particular agent used. We review our current understanding about what an optimal iron level means and discuss currently available IV iron products, together with the various concerns related to potential adverse effects of therapeutic iron administration.

THE ROLE OF IRON IN CKD ANAEMIA
Haemoglobin synthesis is suboptimal without the presence of iron (Fishbane et al. 2004), thus successful haematopoiesis in CKD necessitates the presence of both ESA and iron (Kalantar-Zadeh et al. 2009b). CKD patients may lose iron as a result of gastrointestinal or other bleeding, frequent blood testing or blood losses related to haemodialysis, and ESA use has magnified the absolute or relative iron deficiency caused by such losses (Kalantar-Zadeh et al. 2009a). Together with ineffective gastrointestinal iron absorption, this loss of iron can amount to a negative iron balance of 1.5 to 3 grams per year (Sakiewicz & Paganini 1998). Effective erythropoiesis requires the replacement of, at least, a similar amount of iron.

IRON STORES: DIAGNOSTIC DILEMMAS
Optimising iron management starts with a reliable assessment of iron stores. Unfortunately, the methods that are routinely available to achieve this have significant disadvantages. Table 1 shows laboratory markers of iron status that can be employed

Diagnostic marker Strengths Limitations
Bone marrow iron content, measured from biopsy Gold standard. Invasive test, semi (but not fully) quantitative.
Liver iron content, measured from biopsy Gold standard. Invasive test, semi (but not fully) quantitative.
Liver iron stores via SQUID Noninvasive and safe method. Investigational; limited experience available.

Serum ferritin
Low ferritin levels are highly specific for detection of iron deficiency.
Moderately high ferritin levels could occur in the setting of non-iron-related conditions.
Transferrin saturation ratio (iron saturation ratio) a Higher sensitivity than serum ferritin to detect iron deficiency.
The denominator (TIBC) can be low in malnutrition and/or inflammation.
Serum iron Direct measurement of circulating iron. Diurnal fluctuation; can be low in inflammation.

Reticulocyte haemoglobin content
Measures immediate incorporation of iron into reticulocytes. Limited data, reference levels debatable.
to assess an individual's iron stores. Commonly used iron markers are serum iron, the iron saturation ratio (ISAT, also known as the transferrin saturation ratio [TSAT]), and serum ferritin. The gold standard method for assessment of iron stores is the semiquantitative staining of bone marrow iron, but this is rarely used due to its invasive nature (Fishbane et al. 2001;Kaneko et al. 2003). Similarly, direct liver iron store assessment requires a liver biopsy, although an indirect assessment via the Superconducting Quantum Interference Device (SQUID) may become an option in the future (Asanuma et al. 2005). Other nontraditional but much less frequently utilised iron markers include the reticulocyte haemoglobin content (Fishbane et al. 2001;Kaneko et al. 2003), the percentage of hypochromic red cells (Asanuma et al. 2005), and levels of soluble transferrin receptor (Bovy et al. 1999), erythrocyte zinc protoporphyrin (Chiang et al. 2002) and hepcidin (Malyszko et al. 2005). With the exception of serum iron, TSAT and ferritin, none of these measures are currently routinely applied in the United States, even though the percentage of hypochromic red cells and the reticulocyte haemoglobin content are available if needed, (Wish 2006) and European guidelines recommend their routine use in clinical practice (European Renal Assoc.-European Dialysis and Transplant Assoc. 2004). Furthermore, most of these so-called measures of iron stores are susceptible to being confounded by inflammation or other non-iron-related factors (see Table 1), long-term studies of outcomes using these tests in CKD are missing and they also do not provide information on iron overload (Kalantar-Zadeh et al. 2005a). As a result, serum ferritin and TSAT remain the only routine tools available in clinical practice to assess the iron status in the United States.
A common problem in everyday practice is a discrepancy in the routinely applied measures of iron stores, most often when a low or normal TSAT is coupled with an elevated serum ferritin level. This group most often signals confounding conditions that affect serum ferritin level (Rambod et al. 2008). During the acute phase response, proinflammatory cytokines such as interleukin-1 beta (IL-1␤) and tumour necrosis factor alpha (TNF-␣) increase the synthesis of ferritin through an increased translation of preformed ferritin mRNA (Rogers et al. 1990;Kwak et al. 1995;Harrison and Arosio 1996). The reason for increased ferritin synthesis under inflammatory conditions is unclear. Hypothetically, higher amounts of ferritin may trap more body iron to protect from bacterial infection. This may result in so-called "functional iron deficiency," which may be useful in acute inflammation, but harmful (and leading to refractory anaemia) in chronic disease states that are characterised by inflammation that is unrelated to bacterial infections. IL-1␤ induces ferritin gene expression by translational control of mRNA, but unlike the iron-dependent ferritin gene expression, it also requires the background presence of some cellular iron (Rogers et al. 1994). The permissive role of minimal iron in regulating inflammatory synthesis of ferritin is consistent with recent clinical and epidemiologic findings in CKD patients, in whom, under absolute iron deficiency, the serum ferritin is almost always low, even in the presence of inflammation. But once minimally required iron is available, ferritin regulation becomes a function of non-iron-dependent factors such as inflammation (Kalantar-Zadeh et al. 2004b). This may explain why most studies have found a superior specificity for a low serum ferritin in diagnosing iron deficiency, whereas its sensitivity is poor (Kalantar-Zadeh et al. 1995;Fishbane et al. 2004). Therefore, a low ferritin level appears to be a reliable indicator of iron deficiency, while a normal-to-moderately high serum ferritin (which is a common finding in dialysis patient, Figure 1) is a poor diagnostic test as it does not rule out iron deficiency, nor does it indicate adequate or elevated iron stores (Kalantar-Zadeh et al. 2004b). Extremely elevated serum ferritin levels (usually above 2,000 ng/ml) are usually indicative of true iron overload, a condition also known as haemosiderosis (Powell et al. 1998;Kalantar-Zadeh et al. 2004b). Moderately high levels of serum ferritin, i.e. up to 2,000 ng/ml, have not been found to be associated with excessive tissue iron stores in postmortem autopsies (Gokal et al. 1979). Even though a recent study has indicated a mathematically significant correlation between serum ferritin and liver iron stores using the indirect imaging known as SQUID (Canavese et al. 2004), it appears unlikely that serum ferritin levels below 2,000 ng/ml are associated with clinically harmful increases in body iron stores.
The increase in serum ferritin during inflammation, infection, liver disease, malignancies and other non-iron-related conditions may hinder the ability to assess iron stores under the concurrent presence of these conditions. Serum ferritin is increased with malignancies such as in neuroblastoma (Blatt et al. 1990;Silber et al. 1991), renal cell carcinoma (Kirkali et al. 1999;Miyata et al. 2001) or Hodgkin lymphoma (Hann et al. 1990). Hyperferritinaemia is also associated with liver dysfunction, probably because the liver is the main organ to clear circulating ferritin molecules (Isa et al. 1988). High ferritin levels have been reported in CKD patients with glomerular disease and proteinuria (Branten et al. 2004). Chronic inflammation is quite common in chronic disease states such as CKD or rheumatoid arthritis (Smith et al. 1977), where inflammation may contribute to high ferritin levels more strongly than the actual iron content, and may even mask iron deficiency (Smith et al. 1977;Kalantar-Zadeh et al. 2001;Kalantar-Zadeh et al. 2003;Kalantar-Zadeh et al. 2004b).
In CKD, hyperferritinaemia is paradoxically associated with hyporesponsiveness to ESA and a more severe anaemia (Bailie et al. 1999;Gunnell et al. 1999). In a recent study, 34% of haemodialysis patients with elevated serum ferritin, had inflammation as the probable cause of the increased ferritin level (Kirchbaum 2001;Kirschbaum 2002). In another recent observational study in 82 haemodialysis patients, those with a serum ferritin greater than 800 ng/ml had a significantly higher CRP level and a worse Malnutrition-Inflammation Score (MIS) compared to patients with a serum ferritin below 800 ng/ml (Kalantar-Zadeh et al. 2004b). Multivariable models showed that both CRP and TSAT, independent of each other, correlated significantly with serum ferritin. These findings imply that a moderately high serum ferritin is not just a mere marker of iron stores but also an indicator of inflammation and/or malnutrition as well as other non-iron-related conditions in CKD patients (Kalantar-Zadeh et al. 2004b).

WHAT IS THE OPTIMAL IRON LEVEL?
Optimal iron level is one that allows unhindered erythropoiesis without inducing harmful effects related to iron overload. Therapeutic administration of IV iron bypasses the body's regulatory mechanisms of iron level, thus reliable monitoring of such levels is essential in order to avoid iron overload. Unfortunately, the determination of such optimal levels is made difficult by the above mentioned problems with available diagnostic tests. Short of better monitoring tools, we need to address these issues of efficacy and safety as they relate to the levels of our commonly applied diagnostic tests (TSAT and serum ferritin). Observational studies and clinical trials have indicated that optimal erythropoiesis in patients on Maintenance Haemodialysis (MHD) occurs at a TSAT between 30-50% and a serum ferritin level as high as 1200 ng/ml (Besarab et al. 2000;Kalantar-Zadeh et al. 2007). Addressing the scenario of low TSAT accompanied by high serum ferritin, a randomised controlled trial showed that administration of IV iron to anaemic patients with a TSAT of <25% and a serum ferritin between 500-1200 ng/ml resulted in better erythropoiesis compared to no iron administration, with no shortterm adverse effects (Coyne et al. 2007).
In spite of the better erythropoiesis seen with higher ferritin levels, such levels may appear unsafe on the long run in light of studies showing an association between dialysis morbidity, including risk of infection and iron overload reflected by a high serum ferritin (Eschbach and Adamson 1999). Observational studies have reported an association between higher ferritin and mortality in MHD patients (Kalantar-Zadeh et al. 2001), but subsequently adjusting for markers of malnutrition and inflammation showed that serum ferritin levels between 200 and 1,200 ng/ml and TSAT values between 30-50% were associated with the lowest all-cause and cardiovascular death risk (Kalantar-Zadeh et al. 2005b). Hence, the previously observed associations between moderately high levels of serum ferritin and mortality in unadjusted and case-mix adjusted models appear to be mostly due to the above detailed confounding effects of malnutrition and inflammation. Regarding associations of other markers of iron stores with long-term outcomes, a recent epidemiologic study in 1,283 haemodialysis patients showed that a low, rather than a high, serum iron was significantly associated with higher mortality and hospitalisation (Kalantar-Zadeh et al. 2004a), and low total iron binding capacity was shown to be associated both with iron deficiency and with increased mortality in maintenance haemodialysis patients (Bross et al. 2009). Similar findings were reported in a study examining outcomes associated with TSAT and ferritin levels in male U.S. veterans with moderate and advanced nondialysis-dependent CKD, in that TSAT <25% was associated with a significant increase in all-cause mortality ( Figure 2) (Kovesdy et al. 2009). Higher serum ferritin levels were associated with a trend towards increased mortality (Figure 3), but when taking into account concomitant TSAT and ferritin levels, low TSAT was associated with higher mortality even in patients with elevated serum ferritin (Kovesdy et al. 2009). The same study also examined associations between iron stores and the progression of CKD, and found a significant association between higher TSAT and steeper slopes of estimated glomerular filtration rate, indicating the potential for deleterious renal effects of elevated iron levels (Kovesdy et al. 2009). These latter associations were, however, limited to patients with markers of an activated inflammatory system, suggesting that the deleterious renal effects may have been the result of chronic inflammation (which has been linked to more significant progression of CKD in earlier studies (Rao et al. 2007;Yilmaz et al. 2007), instead of a direct affect of higher tissue iron stores (Kovesdy et al. 2009).
How can we than reconcile the apparent discrepancy between data suggesting a detriment from higher iron levels on the one hand, and studies indicating beneficial outcomes in patients with substantially higher iron levels on the other? One possible explanation is the imperfect nature of our diagnostic tests (especially ferritin) as markers of iron excess, since the higher morbidity and mortality seen with elevated iron indices may be due to non-iron-related factors. Thus, considering high ferritin levels as the primary cause of increased mortality in the setting of inflammation or infection and preventing optimal anaemia management with IV iron for serum ferritin levels greater than 500 ng/ml (Locatelli et al. 2004) or 800 ng/ml (National Kidney Federation 2001) may be irrational (Dukkipati & Kalantar-Zadeh K. 2007). Another potential reason for the discrepancy is a difference between the various studied patient populations. Studies examining the detrimental effects of iron on cardiovascular outcomes have indicated a significant interaction with the age of the study population, with a benefit from lower iron being more significant in younger patients (Sullivan 1981;Haidari et al. 2001;Ramakrishna et al. 2003;Zacharski et al. 2007). It has been suggested that excess iron may be more important in the early stages of the atherosclerotic process, but less so in later stages or in patients with established disease (Zacharski et al. 2007). Since the vast majority of MHD patients already suffer from advanced cardiovascular disease (US Renal Data Systems 2006b), excess iron may be less relevant to this process, and it may thus be beneficial overall through allowing optimal erythropoiesis.

IRON REPLACEMENT: TO DO, OR NOT TO DO?
Iron replacement has become an integral part of the chronic anaemia management of patients with MHD, but there still appears to be a fear to administer higher amounts of iron, Kovesdy et al. 2009. Kovesdy et al. 2009. even when this could result in superior therapeutic efficacy. There are multiple reasons for this so-called "iron apprehension," which is due mostly to concerns about the safety of the specific agent employed (shown in Table 2), or the amount of iron administered.

SHORT-TERM ADVERSE REACTIONS
Short-term adverse reactions associated with IV iron administration can be in the form of hypotension, nausea, vomiting or diarrhoea. The most severe reactions are anaphylactic or anaphylactoid reactions, which could present with rash, urticaria, laryngeal oedema or hypotension. Such reactions appear to be more common with iron dextran (~0.6%), due to the antigenicity of the dextran component (Fishbane 2003). In order to minimise severe outcomes from the administration of iron dextran, the approved US Food and Drug Administration (FDA) label requires a test dose to be administered prior to the full therapeutic dose and patients need to be carefully monitored during the infusion. The absence of a dextran component in other IV iron products has resulted in a significantly lower incidence of life-threatening events related to these products, and a test dose does not need to be administrated when using such agents in the United States (Michael et al. 2002;Chertow et al. 2006).

IRON OVERLOAD
In the pre-ESA era, there were frequent reports about the risk and consequences of iron overload as a result of blood transfusion or iron administration to anaemic dialysis patients (Bregman & Gelfand 1981;Waterlot et al. 1985). Due to ESA  Licensing varies considerably in countries and localities, in terms of dosing, administration and indication by market. 1b http://www.pharmacosmos.com/ for the description of Cosmofer ® . 2c http://www.ferrlecit.com/ for the description of Ferrlecit ® . Prescribers should always consult the product data sheet relevant to the market prior to use. The description of Venofer ® and Ferinject ® comes from the product monographs and all dosing is taken from the relevant data sheets.
administration and possibly to ongoing blood loss in patients with CKD this phenomenon is nowadays extremely rare. Correlating serum markers with tissue iron stores can be difficult, but it is likely that true tissue iron overload occurs at serum ferritin levels above 2000 ng/ml (Gokal et al. 1979). When adhered to, our present practice recommendations make it unlikely that such elevated levels are reached as a result of iron replacement therapy (European Renal Assoc. /European Dialysis and Transplant Assoc. 2006a).

INFECTIONS
A clinical trial performed almost 3 decades ago amongst 137 iron deficient individuals in Somalia showed that the risk of infection in those who received iron supplementation was almost 5 times higher than those who received placebo (Murray et al. 1978), and a limited number of observational studies have indicated an association between high serum ferritin and infection (Waterlot et al. 1985;Seifert et al. 1987). The IV replacement route delivers iron directly to the circulation, thus bypassing the natural route of iron ingestion (Fishbane 2003). Release of iron into the circulation can support the growth of bacteria, which can increase the risk of infection and potentially worsen its outcomes. It is important to note that to date, no randomised controlled study has substantiated the risk of increased infection as a result of IV iron administration in dialysis patients (Hoen et al. 2002;Sirken et al. 2006). Nevertheless, it is recommended to exercise caution when using IV iron therapy in patients with established infections, and it is probably prudent to withhold such therapy in patients with severe bacterial infections.

OXIDATIVE STRESS
Free elemental iron is a very potent oxidising agent, and it can generate toxic free radicals. Several in-vitro studies have indicated the association between iron supplementation and oxidative stress in cell cultures (Zager et al. 2002;Zager et al. 2004). Short-term studies in dialysis patients have also showed an increase in markers of oxidative stress after IV iron administration (Michelis et al. 2003;Agarwal et al. 2004), but it is unclear what the longer term consequences of such chemical alterations are. Release of labile-free iron appears to be more likely with the smaller molecular-weight nondextran iron formulations, thus these products (iron sucrose and iron gluconate) need to be administered in smaller doses and as a slow IV infusion (Van Wyck et al. 2004). There may also be differences between the latter two products in their propensity to induce tissue damage, as shown in a recent study that examined renal injury (proteinuria and enzymuria) occurring with the administration of a single dose of the two different products in patients with CKD stages 3 and 4 (Agarwal et al. 2007). The administration of iron sucrose was associated with a higher protein/creatinine ratio compared to iron gluconate, but there was no difference between the two agents in N-acetyl-beta-Dglucosaminidase/creatinine ratio (Agarwal et al. 2007). The relevance of these results to longer term administration of either agent is unclear.

LONG-TERM SAFETY OF IV IRON ADMINISTRATION
While in vitro studies are important in clarifying the mechanisms of various adverse effects related to IV iron preparations, and short-term clinical studies are essential to prove their efficacy and safety from a regulatory standpoint, the ultimate question remains the long-term safety of these treatments as measured by their effects on survival and other relevant clinical end points. No randomised controlled study has been conducted to assess the risk of increased infection or death as a result of IV iron administration in dialysis patients. When examining a surrogate end point of adverse outcomes, a recent clinical trial showed that in those dialysis patients who received IV iron, the level of the inflammatory cytokine TNF-␣ was in fact decreased when compared to those who did not receive any IV iron (Weiss et al. 2003). The association between iron administration and long-term outcomes has been examined in observational studies. In a study of 10,169 haemodialysis patients in the United States, a tendency towards higher death risk was described in those patients who received higher doses of IV iron (Feldman et al. 2002). The same group of authors revised their initial conclusions when they used time-varying marginal structural models (which adjusted for bias by indication) to reanalyse the risk of death associated with iron administration in 32,566 haemodialysis patients (Feldman et al. 2004). In this latter study, there was no association between mortality and the amount of iron administration. A more recent epidemiological study in 58,058 haemodialysis patients found that, compared to those who did not receive any IV iron, administered IV iron up to 400 mg/month was associated with improved survival (Kalantar-Zadeh et al. 2005b). Although patients who received IV iron had significantly different demographic, clinical and laboratory features at baseline, the survival benefits of IV iron was relatively consistent in different subgroups of haemodialysis patients, including those who had high ferritin but low TSAT values (Kalantar-Zadeh et al. 2005b). These findings appear to be inconsistent with the notion that the administration of IV iron is deleterious. Furthermore, a recent epidemiologic study suggested that the negative outcomes associated with higher targeted haemoglobin levels in several randomised controlled trials of ESA administration may have been the result of the relative thrombocytosis caused by the iron deficiency induced by higher doses of ESA (Streja et al. 2008). The merit of this hypothesis needs to be tested in clinical trials examining the concomitant administration of ESA and iron.

CONCLUSIONS
Treatment of anaemia of CKD involves an integrated approach that addresses the complex aetiology of this entity. Iron replacement is an important component of this treatment strategy and involves routine monitoring of iron stores and replacement of iron in order to assure unhindered erythropoiesis. Flaws inherent of our routine clinical monitoring tools make it difficult to accurately assess iron stores; hence the safe and effective application of iron replacement therapy necessitates a detailed understanding of the caveats in our diagnostic methods. Optimal erythropoiesis appears to occur with an ISAT of 30-50% and a serum ferritin of up to 1200 ng/ml, and iron overload is unlikely within these ranges. Safety concerns related to the various products used for IV iron replacement include allergic type reactions and the generation of free reactive iron in blood with consequent oxidative stress, worsened infections and tissues damage. These problems may mar the various commercially available products to a different extent. There is currently, no evidence to indicate that the adverse effects noted in-vitro and in short-term studies of IV replacement therapy are associated with deleterious long-term clinical outcomes in dialysis patients.