Practical management of iron overload

The practical management of iron overload requires reliable estimation of body iron content and distribution as well as an understanding of how iron overload translates into clinical consequences. In this article, it will be seen that our ability to estimate the distribution of excess tissue iron, to predict its consequences and therefore to tailor treatment accordingly is surprisingly imprecise. Our understanding of how iron chelators best prevent these consequences is also limited. It will be seen that the safest and most effective ways of removing excess iron vary depending on the degree and rate of iron loading as well as the underlying condition being treated. Evidence of benefit in survival with chelation treatment takes many years to emerge and currently only exists for thalassaemia major patients treated with desferrioxamine. Long-term survival benefits have not been demonstrated in other iron-overloaded conditions or with other chelation regimens. In such circumstances, practical treatment protocols are necessarily based on inference rather than direct evidence. The rate of iron loading will determine when chelation should start as well as the chelation regimen to be adopted. Excess loading may occur secondary to increased iron absorption or from repeated blood transfusions. Iron loading from blood transfusion This can be estimated from the total number of units of red cells given. A unit processed from 420 ml of donor blood contains approximately 200 mg of iron (0·47 mg/ml of whole donor blood). A more precise estimation of iron loading can be derived from the volume of blood transfused and the mean haematocrit of processed blood obtained from the transfusion centre. For the UK, assuming a mean haematocrit of 0·6 for SAG–M (saline adenine glucose–mannitol) blood, then the iron content is 0·7 mg per ml transfused. Transfusion requirements vary with diagnosis. In thalassaemia major, the rate of iron loading is reasonably well defined provided a mean Hb value of 12 g/dl is achieved. The equivalent of 100–200 ml of pure red cells/kg/year (Modell, 1977) (i.e. 160–330 ml/kg of SAG–M blood/year) are transfused (equivalent to 116–232 mg of iron/kg body weight/year or 0·32–0·64 mg/kg/d) (Thalassaemia International Federation (TIF) Guidelines, 2000). The transfusion requirements and hence iron loading in unsplenectomized thalassaemia major patients are generally higher. However, hypertransfusion can also decrease splenic size (O'Brien et al, 1972) and the early introduction of a hypertransfusion regimen (Modell, 1977) may therefore reduce blood requirement. In other forms of anaemia, such as aplastic anaemia, myelofibrosis and pyruvate kinase (PK) deficiency, the transfusion requirement is highly variable and a record of transfusion requirement is necessary to calculate transfusional iron loading. Iron loading from increased gastrointestinal (GI) absorption In healthy individuals, iron absorption is 1–1·5 mg/d and is balanced by iron loss from skin, gut, menstruation or pregnancy. Iron absorption is increased by anaemia, hypoxia, ineffective erythropoiesis and by the presence of HFE variant genes. In thalassaemia syndromes, iron absorption exceeds iron loss when erythron expansion exceeds five times that of healthy individuals (Cazzola et al, 1997). Thus, iron absorption in thalassaemia intermedia can be up to 5–10 times normal, or 0. 1 mg/kg/d (Pippard et al, 1979; Gordeuk et al, 1987; Pootrakul et al, 1988). Splenectomy appears to increase the rate of GI hyper-absorption in thalassaemia intermedia (Fiorelli et al, 1990) and other conditions such as PK deficiency (Zanella et al, 1993) through mechanisms which remain speculative, such as hypoxia associated with micropulmonary emboli (Chuansumrit et al, 1993). In thalassaemia major, hypertransfusion decreases hypoxia and expansion of the erythron, thereby decreasing GI absorption to 1–4 mg/d (Pippard & Weatherall, 1984). In individuals who are poorly transfused, absorption rises to 3–4 mg/d or more. This represents a supplementary 1–2 g of iron loading per year (equivalent to up to 3 months chelation treatment). In other forms of haemolytic anaemia, excess iron absorption is highly variable. Thus, in sickle cell disease, despite chronic haemolytic anaemia, iron overload is not seen in untransfused patients (O'Brien, 1978; Porter & Huehns, 1987). In contrast, in PK deficiency, iron overload is common in untransfused patients, particularly in those who are splenectomized (Zanella et al, 1993). The co-inheritance of HFE mutations (H63D, C282Y) with conditions predisposing to increased iron absorption such as certain haemolytic anaemias could, in principle, increase iron absorption (Arruda et al, 2000). However, in thalassaemia major, the co-inheritance of HH genes has little impact on net iron loading (Longo et al, 1999). Serum ferritin The association between serum ferritin and levels of body iron is well established (Worwood, 1986) and the test is easy to perform compared with other tests for iron overload. However, the correlation between serum ferritin and body iron is not sufficiently precise to be of strong prognostic value. Serum ferritin will be disprortionately raised when inflammation or tissue damage is present. Conversely, serum ferritin will be falsely depressed when patients are scorbutic, a frequent occurrence in iron overload owing to rapid oxidation of vitamin C (Chapman et al, 1982). Serum ferritin is also influenced by chelation treatment in a manner which is not a simple relationship with body iron. Thus, particularly for values above 3000 µg/l, serum ferritin levels fall faster with chelation than would be predicted from a diminution of body iron alone (Davis & Porter, 2000). A further problem is that the relationship between serum ferritin and body iron appears to be different for different haematological conditions, tending to be depressed relative to body iron in conditions in which iron overload is not predominantly secondary to blood transfusion. Thus, in thalassaemia intermedia, serum ferritin tends to underestimate the degree of iron overloading (unpublished observations). With these caveats in mind, serum ferritin measured at regular intervals (at least 3 monthly with thalassaemia major) has some therapeutic and prognostic uses. A target ferritin of approximately 1000 µg/l is generally recommended standard practice in thalassaemia major (TIF Guidelines, 2000) and other forms of iron overload resulting from blood transfusion. However, the only study demonstrating a link between serum ferritin and prognosis showed that patients with values below 2500 µg/l on two thirds of occasions had less risk of cardiac complications than patients who failed this achievement (Olivieri et al, 1994). The rate of fall in serum ferritin with chelation treatment is also a useful tool for giving information back to patients about their progress. Serum ferritin values below 3000 µg/l fall at a reasonably consistent rate with intravenous desferrioxamine therapy (130 µg/kg/month on 50 mg/kg/d) and this can be used to encourage patients and clinicians about realistic target values which can be achieved with such a regimen (Davis & Porter, 2000). Finally, serum ferritin measurement can help to reduce the risk of desferrioxanine overdosing, particularly as the level of body iron falls with treatment. In thalassaemia major, if the mean daily desferrioxamine dose (mg/kg) divided by the serum ferritin (µg/l) exceeds 0·025 (this ratio is referred to as the Therapuetic Index) then the dose should be decreased (Porter et al, 1989b, 1998). The relationship between serum ferritin and body iron may differ in other conditions and this ratio is not applicable. This is particularly the case in sickle cell disorders in which the serum ferritin is disproportionately increased for weeks following vaso-occlusion (Brownell et al, 1986; Porter & Huehns, 1987). Liver iron The liver is the major site of iron storage in iron overload, containing 70% or more of body iron stores (Modell & Berdoukas, 1981; Angelucci et al, 2000). Liver iron correlates closely with total body iron in transfusional iron overload and total body iron, and is approximately equivalent to 10·6 times the hepatic iron concentration (in µg/g of liver, dry weight) (Angelucci et al, 2000). The liver is also accessible for biopsy or for non-invasive measurement of iron content by Superconducting Quantum Interface Device (SQUID) or magnetic resonance imaging (MRI) and has therefore been the measurement most used for monitoring treatment and assessing prognosis. Liver iron concentration appears to have prognostic value in iron overload. The major study showing such a relationship was in 59 patients over 7 years of age, who had been treated with desferrioxamine for variable periods of time. It was found that all patients who died with cardiac complications had liver iron concentrations > 15 mg/g dry weight (Brittenham et al, 1994). It was then argued (Olivieri & Brittenham, 1997) that patients with liver iron values in this range should be regarded as ‘high risk’. However, as heart failure was the cause of death in all patients, it is likely that liver iron was not the primary determinant of cardiac-free survival but represented an association of risk. Furthermore, the duration of exposure to excess iron must also be taken into account as a prognostic determinant. Another important prognostic variable is probably the age and the levels of iron loading when this happens. Thalassaemia major patients currently alive in their late thirties and early forties began treatment in late childhood or adolescence at a time when liver iron values were considerably in excess of 15 mg/g dry weight (Barry et al, 1974). This shows that, provided iron levels can be subsequently reduced, a poor outcome does not inevitably result. It has been suggested that realistic target levels of liver iron in treated patients should be less than 7 mg/g dry weight as heterozygotes with HH can reach such levels without significant pathology resulting (Olivieri & Brittenham, 1997). This analogy assumes that the distribution of iron between heart and liver is similar in heterozygotes with HH to patients with transfusional iron overload treated with chelators. This necessarily cannot take into account that such levels take a lifetime to accumulate in heterozygotes with HH but are reached in early childhood in transfusion-dependent patients. Furthermore, recent evidence suggests that levels of liver iron correlate poorly with cardiac iron in patients on chelation. Despite these caveats, a liver iron value above 7 mg/g dry weight, particularly in a young individual, is probably the best indicator currently available for initiating chelation treatment. Liver iron concentrations are usually measured by chemical determination on a liver biopsy sample. Biopsy is an invasive procedure, but in experienced hands has a very low complication rate (Angelucci et al, 1995). Inadequate sample size (< l mg dry weight) or uneven distribution of iron in the presence of cirrhosis (Villeneuve et al, 1996) may give misleading results. The measurement of liver iron can be performed on wet or dried samples and the conversion of one to the other varies according to the drying and measuring techniques used in the laboratory. An agreed international standardization of measurement and reporting would be helpful to practising clinicians. Liver iron can also be measured non-invasively by magnetic biosusceptiometry (SQUID) (Brittenham et al, 1994), although these are very expensive to purchase and maintain and there are no such machines in the UK. MRI using T2 relaxation has hitherto not given sufficiently precise estimation of liver iron because liver iron in overload conditions lies above the linear range. Recent developments using a modified form of acquisition called T2* gives a somewhat better approximation to liver iron as measured by biopsy (Anderson et al, 2000a). Heart iron and function Although heart failure is the main cause of death in iron overload (Zurlo et al, 1989; Brittenham et al, 1994), little is known about the relationship of heart iron content to mortality. This is partly because endomyocardial biopsy is impractical in routine clinical practice and is not a useful indicator of heart iron as a whole owing to very uneven iron distribution in the heart (Barosi et al, 1989). Post-mortem measurement of iron concentration in many organs (Modell & Berdoukas, 1981) showed than, even in patients dying of heart failure, heart iron was only a fraction of that in the liver. Furthermore, a simple relationship between heart iron concentration and heart failure is not consistent with the marked improvement in clinical heart failure left ventricular ejection fraction and cardiac dysrhythmias observed soon after commencing intravenous desferrioxamine (Davis & Porter, 2000). Recent development of cardiac T2* MRI measurement has provided an opportunity to study the relationship between this indirect parameter of heart iron and cardiac function as well as liver iron (Anderson et al, 2000a, b). Preliminary results suggest that patients with right or left ventricular ejection fraction below control values tend to have the lowest T2* values and, by implication, the highest myocardial iron. Interestingly, there was no correlation between liver iron as measured by biopsy or MRI and cardiac T2*. A cardiac MRI can measure right and left ventricular ejection fractions at the same time as T2* and, in principle, could be measured in any hospital with suitable MRI facilities. How this parameter can best be used in the monitoring and treatment of patients requires careful prospective study. Urinary iron estimation Different iron chelators induce different proportions of urinary iron relative to faecal excretion. Deferiprone excretes iron almost entirely in the urine (Collins et al, 1994), desferrioxamine excretes an approximately equal amount in the faeces (Pippard et al, 1982a), and the new oral chelator developed by Novartis, ICL670A (Sergejew et al, 2000), appears to excrete iron exclusively in the faeces in preclinical studies. Urine iron excretion with desferrioxamine increases in proportion to iron loading with transfusions (Modell & Beck, 1974), dose of chelation used (Sephton-Smith, 1962; Pippard et al, 1978a), ascorbate status (Pippard et al, 1982a) and point in the transfusion cycle (Pippard et al, 1982a), being proportionally higher in the urine as the Hb level falls. The proportion of urinary and faecal excretion iron excretion increases with the dose of desferrioxamine given (Pippard et al, 1982a). There is very considerable day to day variation in urinary iron excretion with the same dose of chelation, even in fully controlled trial conditions. Because of this day to day variability in urinary iron excretion and the variable proportion of faecal excretion, 24 h urinary iron measurement with desferrioxamine must be interpreted with caution. The decision of when to treat iron overload depends on linking a given level of iron overload to risk. As seen above, risk from iron overload is likely to depend not just on body or liver iron levels but also the duration of exposure to excess iron. Furthermore, not all tissues are equally susceptible to damage from a given level of excess iron; heart failure occurs at significantly lower levels of tissue iron than cirrhosis (Modell & Berdoukas, 1981). The most important reason for initiating treatment is to prevent heart failure, which is the major cause of death in transfusional iron overload (Zurlo et al, 1989). However, considerable other morbidity results from iron overload, particularly if it occurs in childhood, and the prevention of complications such as diabetes, hypogonadotrophic hypogonadism, poor growth, hypothyroidism and hypoparathyroidism are achievable goals with optimal chelation (De Sanctis et al, 1989; Zurlo et al, 1989). The point at which blood transfusions have deposited enough iron to cause irreversible tissue damage has not been assessed in prospective trials. Current practice is based on balancing the known risks of excessive early treatment with the perceived risks of delayed treatment. Current practice is to start desferrioxamine when ferritin values reach 1000 µg/l or when 10–20 transfusions have been given. It is clear that the risk of overdosing is greatest in young children, particularly from effects on growth (Piga et al, 1988; Olivieri et al, 1992a) and audiometric disturbances (Olivieri et al, 1986). If treatment is commenced before the age of 3 years then this should be given with great caution. Liver iron concentrations above 7 mg/d dry weight should be regarded as an indication for treatment, although arguably lower levels of liver iron (e.g. above 3·5 mg/g dry weight) also require treatment (Olivieri & Brittenham, 1997). As already discussed, the rate of iron loading is highly variable and the relationship between serum ferritin and body iron can be different from thalassaemia major (Fiorelli et al, 1990). Treatment typically needs to begin later in life than thalassaemia major. An estimation of liver iron is advisable before starting treatment to see whether this has exceeded 7 mg/g dry weight, as serum ferritin tends to underestimate parenchymal iron loading in this condition. In general, the rates of iron loading will be less than in thalassaemia major and the regime must be modified accordingly. Many patients with thalassaemia intermedia begin on a hypertransfusion regimen similar to thalassaemia major at some point in their lives and a chelation regime similar to that given in thalassaemia major will need to commence at this point. As with thalassaemia intermedia, the rate of iron overloading is highly variable and difficult to predict. Ineffective erythropoiesis and the co-inheritance of HFE mutations may increase the risk of iron loading (Cotter et al, 1999) as, unlike thalassaemia intermedia, conditions such as pyruvate kinase deficiency and inherited sideroblastic anaemia commonly affect Northern Europeans and co-inheritance of a HFE mutations will occur in about 1 in 10 such patients. Liver biopsy for measurement of tissue iron levels and for assessment of fibrosis is advisable before starting treatment. It is probable that the measurement of cardiac iron is these conditions using MRI will be an additionally useful variable. In congenital dyserythropoietic anaemia, iron overload is generally a function of age and the degree of ineffective erythropoiesis, although the rate of loading is variable (variability independent of HFE) and some patients never require chelation therapy (Wickramasinghe et al, 1999). Iron overload is not seen in sickle cell disease without blood transfusions. Although there are no large studies of the effects of transfusional iron loading in HbSS, it is likely to carry similar risks to other disorders. Typically, transfusion begins later in life than in transfusion-dependent anaemias and is often given sporadically, except in the prevention of recurrent stroke. Manual exchange transfusion decreases the risk of iron loading and can even reverse iron loading (Porter & Huehns, 1987) and should be encouraged. However, when exchange is not possible on a regular basis because of poor venous access or because sufficient numbers of trained staff are not available, then top-up transfusion is frequently given, inevitably leading to iron overload. In steady state, there is a reasonable correlation between serum ferritin and the number of units of blood transfused (Porter & Huehns, 1987) or liver iron (Brittenham et al, 1993). However, the measurement of iron overload in sickle cell disease using serum ferritin is seriously flawed, as the serum ferritin remains disproportionately raised for weeks after a vaso-occlusive episode (Brownell et al, 1986; Porter & Huehns, 1987). Liver iron (and ideally heart iron with MRI) should be measured before treatment and at regular intervals thereafter. Decisions about when to start treatment will depend on practical as well as theoretical considerations. Regular use of desferrioxamine infusion can be very difficult for sickle patients, particularly in adolescence and beyond, and Portacaths have a high thrombosis and infection rate in sickle cell disorders (personal observations). The age at which regular transfusion begins is critical to the iron removal strategy to be adopted. Inherited anaemias which may be transfusion dependent and require chelation treatment include Fanconi anaemia and Diamond–Blackfan anaemias (Ambruso et al, 1982). The decision to start chelation in such conditions will be based on similar principles to thalassaemia major. In transfusion-dependent anaemias of adult onset, such as acquired aplastic anaemia, myelofibrosis, osteopetrosis or iron overload resulting from transfusion with repeated courses of high-dose chemotherapy, the decision to remove accumulated iron must be weighed against the prognosis of the underlying condition on a case by case basis. Until the recent identification of the genes responsible for genetic haemochromatosis, the most effective way of identifying this condition was using a transferrin saturation of > 50% on two samples, the second of which being a fasting value (Worwood, 1999). Serum ferritin is usually elevated but has been shown to be less reliable as a screen than fasting transferrin saturation. Historically, confirmation of the diagnosis has been achieved by liver biopsy and measurement of liver iron with an ‘hepatic iron index’ > 2 (hepatic iron index = mmol iron per g dry wt divided by the patients age). However, it is now possible to make an early genetic diagnosis based on homozygosity for C262Y or compound heterozygotes for C282Y/H63D. This then raises the question as to the role of liver biopsy in deciding when to commence phlebotomy, particularly in younger patients. Detection of fibrosis is often given as a reason for biopsy but it has been shown that, in absence of hepatomegaly, a raised alanine transaminase (ALT) or a serum ferritin > 1000 µg/l, fibrosis is very unlikely and biopsy may not be necessary (Guyader et al, 1998). Therefore, in younger patients, identified using genetic screening and without the above risk factors, liver biopsy may not be mandatory. Treatment is generally indicated in men with serum ferritin levels of > 300 µg/l and in women with serum ferritin levels of > 200 µg/l, regardless of the presence or absence of symptoms. In cases in which transferrin saturation or serum ferritin are raised but genetic screening for HFE mutations does not show homozygous C282Y or C282Y/H63D, liver biopsy can be of value in determining the extent of iron loading or of other liver pathology. Phlebotomy is the most efficient and least toxic way to remove excess iron, as 1 pint of blood contains 200 mg of iron and can be performed 1–2 times a week in patients who do not have compromised erythropoiesis. Genetic haemochromatosis This should be treated by phlebotomy (Barton et al, 1998), as a unit of blood can generally be removed once or twice per week, resulting in iron removal of between 1 and 2 g of iron per month, considerably in excess of that achieved with iron chelation therapy. Phlebotomy is continued until the serum ferritin and transferrin saturation indicate that iron stores have just reached the iron deficient range (ferritin < 20 µg/l and transferrin saturation < 16%). Thereafter 1 unit of blood is venesected every 3–4 months to prevent the re-accumulation of iron, generally maintaining serum ferritin < 50 µg/l. Post bone marrow transplant (BMT) Patients who have received successful BMT for β-thalassaemia major must be venesected repeatedly until normal (or low normal) levels of iron stores are obtained (Angelucci et al, 1998). This can be achieved with phlebotomy, the frequency of which will be determined empirically by what can be tolerated without clinically relevant falls in Hb values. Baseline Hb values are affected by the genotype of the donor (i.e. whether thalassaemia trait) and it is unrealistic to expect to achieve the same Hb levels as those seen with patients venesected for genetic haemochromatosis. In other groups of patients, such as those who are in sustained remission following BMT or chemotherapy for haematological malignancy but have received multiple blood transfusions, phlebotomy may be advisable to normalize iron loading. Assessment of iron overload using serum ferritin alone can be unreliable, so that independent confirmation of excess iron loading should be sought. This may simply be a reliable documentation of the transfusion of many units of blood. It is not unusual for a patient undergoing induction, consolidation and BMT to have received in excess of 60–100 units of blood (Bradley et al, 1997). Transferrin saturation in excess of 50% supports a suspicion of iron overload. Non-invasive estimation of liver iron stores using SQUID or MRI can be helpful if available. Sideroblastic anaemia Iron overload often complicates sideroblastic anaemia. Co-inheritance of an HFE variant such as C282Y of H63D may increase the risk of iron loading. Phlebotomy should be considered as a therapeutic option as some patients have surprisingly shown an improvement in Hb levels following normalization of tissue iron with phlebotomy (Cotter et al, 1999). The rate of phlebotomy that can be tolerated is highly variable and needs to be determined empirically. Porphyria cutanea tarda This disorder, characterized by a photosensitive dermatosis and hepatic siderosis, is associated with a number of risk factors including co-inheritance of an HFE mutation, high alcohol ingestion, hepatitis C infection (Bulaj et al, 2000) and thalassaemia trait (Adjarov et al, 1984). Treatment is effective with phlebotomy or desferrioxamine (Rocchi et al, 1991). Provided there is adequate erythropoietic reserve, phlebotomy is generally simpler to perform. The goals of chelation can be summarized firstly as the achievement of safe tissue iron concentrations by promoting negative iron balance and secondly as the detoxification of iron until this goal has been achieved. Only a small proportion of body iron is available for chelation at any point in time and some of this iron is required for a variety of essential metabolic functions. Chelatable iron is that which becomes available from the continuous catabolism of red cell haemoglobin amounting to 20–30 mg/d, as well as iron that is made available within cells from the continuous breakdown of ferritin within lysosomes. It follows that large intermittent doses of chelator will be less efficient and potentially more toxic than lower doses given more continuously. Because iron is needed for essential metabolic functions, the dose schedule used for any chelator involves a balance between the risks of excess iron and those of excess chelator (Porter, 1997). The efficiency of currently used chelator regimens is remarkably low; 90% of administered subcutaneous desferrioxamine is excreted without binding iron (Porter et al, 2000) with 96% of deferiprone having the same fate (Al-Refaie et al, 1995b). As discussed, a ‘safe’ level of tissue iron varies with age diagnosis and rate of iron loading and probably also on iron distribution within the body. An assessment of organ function in addition to tissue iron levels will influence chelation management (TIF Guidelines, 2000). Detoxification of iron requires stable binding of the six co-ordination sites of iron by a chelator, with the prevention of redox cycling and oxidative damage. In general, hexadentate chelators, having six binding sites, have a more stable co-ordination chemistry than bidentate chelators, which have only two active co-ordination sites and therefore require three chelator molecules for each iron atom. This is particularly important at low concentrations of chelators when bidentate compounds tend to dissociate. Rapid improvement in cardiac function with continuous intravenous desferrioxamine is probably a result of chelation of relatively small toxic iron pools within the plasma compartment, such as plasma non-transferrin-bound iron (Porter et al, 1996), or within the heart, such as the chelatable labile intercellular pool (Zanninelli et al, 1997). The reader is referred to other publications (Porter, 1996) for a detailed discussion of the principles of chelation therapy. Mechanism of action and pharmacology Desferrioxamine is a siderophore (a naturally occurring iron-carrier) and is produced and purified from the microbe Streptomyces pilosus. One molecule of chelator binds one atom of iron forming the highly stable hexadentate iron complex ferrioxamine at physiological pH values. Owing to its size, desferrioxamine is poorly absorbed from the gut. Desferrioxamine is metabolized in the liver into iron-binding metabolites which are more plentiful when the therapeutic index is high and there is greater excess of iron-free chelator (Porter et al, 1998). Urine iron is derived from that released after the breakdown of red cells in macrophages, whereas faecal iron is derived from iron chelated within the liver (Hershko & Rachmilewitz, 1979; Pippard et al, 1982b). Desferrioxamine has a short plasma half-life (initial half-life, 0·3 h) (Lee et al, 1993), being eliminated rapidly in urine and bile. Once an infusion of desferrioxamine stops, iron chelation will cease soon thereafter. Because at any moment only a small proportion of body iron is available for chelation, the longer the duration of infusion, the more efficient the chelation process. This may be particularly important in patients with severe iron overload (Davis & Porter, 2000). Evidence of efficacy There is now overwhelming evidence for the efficacy of desferrioxamine on long-term survival and in preventing complications of iron overload and, hence