Molec. Aspects Med. Vol. 13, pp. 113 - 165, 1992
0098 - 2997/92 $15.00 © 1992 Pergamon Press plc
Printed in Great Britain. All rights reserved.
IRON CHELATORS IN MEDICINE Chaim Hershko Department of Medicine, Shaare Zedek Medical Center, Jerusalem, Israel
Contents
INTRODUCTION
114
CHAPTER 1 The Pharmacology of Desferrioxamine
115
CHAPTER 2 Effects of Long-term Chelating Therapy
121
CHAPTER 3
125
Desferrioxamine Toxicity
CHAPTER 4 Use of DF in Conditions Unrelated to Iron Overload
128
CHAPTER 5 Development of New Oral Iron Chelators
144
ACKNOWLEDGEMENTS
149
REFERENCES
150
113
Introduction
Although the iron chelating drug desferrioxamine (DF) has been available for clinical use for over three decades, it is only within the last 15 years that it has been gradually accepted as an essential drug in the management of transfusional iron overload. Through the pioneering work of British investigators (Barry et al., 1974; Modell et al., 1982) and the admirable devotion of Italian haematologists (Gabutti, 1990; Zurlo et aL, 1989; Borgna-Pignatti et al., 1989) it has been demonstrated beyond any doubt that long-term DF therapy is capable to prevent the severe complications of transfusional iron overload and to improve significantly the life expectancy of thalassaemic patients. However, management of acute and chronic iron toxicity is only one facet of the potential clinical use of iron chelators. Recognition of the pivotal role of iron in the generation or toxic oxygen-derived species through the Haber-Weiss reaction (Halliwell and Gutteridge, 1984, 1986), the ability of DF to limit the damage associated with free radical formation in reperfusion injury, and the possibility of inhibiting cell proliferation by interaction of DF with the iron-dependent enzyme ribonucleotide reductase, has resulted in an ever increasing number of studies exploring the possibilities of new therapeutic applications of iron chelating agents. In the text that follows, we shall review the pharmacokinetics of DF, examine the evidence supporting the clinical usefulness of long-term iron chelating therapy in transfusional iron overload; the toxicity of DF and variables that may increase or reduce the risks of such toxicity; review some of the most significant recent studies on the potential usefulness of iron chelators in conditions unrelated to iron overload and, finally; list some of the most important orally effective iron chelating drugs that are currently being considered as possible alternatives to DF therapy.
114
Chapter 1
The Pharmacology of Desferrioxamine
Desferrioxamine B (DF) is a colorless crystalline substance produced by Streptomyces pilosus. It consists of a chain of three hydroxamic acids terminating in a free amino acid group which, in turn, enables it to form salts with organic and inorganic acids (Keberle, 1964). It is capable of combining with ferric (Fe3+) iron at a 1:1 molar ratio and with a stability constant of 1031. The DF molecule (Fig. 1) is wrapped around the iron nucleus encasing it in an envelope of organic material. The affinity of DF to Fe 2+ and other metal ions such as Zn2+ , Cu 2+ , Mg 2+ and Ca2+ is much lower and ranges from 102 to 1014 . In contrast to the voluminous literature describing the clinical effects of DF therapy, information on the pharmacology of DF is relatively scanty and is largely limited to studies published in the early 1960s (Wohler, 1963; Peters et al., 1966; MeyerBrunot and Keberle, 1967). One of the key observations in these early studies was recognition of the fundamental difference in behaviour of DF as compared to ferrioxamine, the DF-iron complex. As aresult of the change in configuration following interaction with iron, ferrioxamine becomes an extremely stable compound which is resistant to enzymatic degradation (MeyerBrunot and Keberle, 1967). Ferrioxamine is distributed in the extracellular space and is unable to penetrate cells, as evidenced by a plateau in its serum concentrations following nephrectomy (Keberle, 1964). In the intact dog, 98% of labelled ferrioxamine is recovered in the urine within 3 days. In contrast to ferrioxamine, DF is capable of penetrating various tissues and is rapidly catabolized in vivo. Consequently, radioiron-labelled DF which, by definition, represents ferrioxamine cannot be utilized for studying the in vivo behaviour of DF, a consideration which has been unfortunately overlooked in the past in a number of studies.
H2N (~H2)5
.CONH /CONH (/CH2")2 "(CH2)5 (CH2)2 ~(CH2)5 CH3 " ,N-C \N - C / ~1~1-~ OH0 13H6 OH0
Fig. 1. Desferrioxamine B, 115
116
C. Hershko
The most straightforward method of studying the fate of DF is by means of the tritium-labelled drug. Such studies, performed in dogs, have shown that although 70% of labelled DF is excreted in the urine within 3 days, over half of this excretion is in the form of DF metabolites (Keberle, 1964). DF is excreted by both glomerular filtration and tubular secretion, whereas ferrioxamine is partly reabsorbed following glomerular filtration (Peters et al., 1966). The combination of both of these tubular mechanisms tends to reduce the ability of injected DF to promote the urinary excretion of chelated iron. In contrast to ferrioxamine, the concentration in plasma of DF in nephrectomized dogs does not reach a plateau. Its disappearance curve is complex, reflecting a combination of metabolic breakdown, and penetration into various tissues. Bearing these limitations in mind, estimates such as the volume distribution of DF in 62% of the body, and a metabolic half life of about 1 hr (Peters et al., 1966) may only be regarded as rough approximations. Tissue concentrations of labelled DF in dogs following intravenous injection are highest in the bile and brain, intermediate in the spleen, kidney and plasma, and lowest in the heart, lungs and fatty tissues. Although the great efficiency of the liver in clearing DF from the circulation is clearly evidenced by the rapid concentration of labelled DF in the bile, DF is not retained in the liver to a significant extent. Most of the biliary, and a great part of the urinary excretion of the tritiated compound is in the form of DF metabolites. The nature of these metabolites in the bile has not been determined. In the urine, the major compound is metabolite C, in which the original amino group of DF has been replaced by a carboxyl group. By far the most active tissue in the enzymatic breakdown of DF is the plasma, followed by various parenchymatous organs (MeyerBmnot and Keberle 1967). Rat plasma is about 4 times more efficient in DF catabolism than plasma obtained from the dog, or man. These studies employing tritiated DF yielded valuable information on animals, but the amount of radioactivity required precluded the use of similar techniques in man. A useful substitute is the measurement of filtrable (chelated) plasma iron following DF administration. Using a saturating solution of iron in a manner similar to the measurement of latent iron binding capacity, both ferrioxamine and unbound DF can be measured simultaneously. Such studies have shown that following the intravenous bolus injection of DF, plasma levels of DF drop to half of the initial concentration within 5 to 10 min (Summers et al., 1979). Conversely, slow subcutaneous infusion of DF at a dose of 100 mg/kg/24hr results in a gradual buildup of plasma DF concentrations reaching a plateau in about 12 h. This is accompanied by a simultaneous increase in plasma ferrioxamine concentrations. There is a direct correlation between the total 48 hr urinary excretion of iron and peak plasma concentrations of ferrioxamine during DF infusion. Studies with 59Felabelled ferrioxamine showed a volume distribution of about 91 and a fractional clearance rate of 0.3/hr. Both of these measurements are very similar to those obtained in the early canine studies indicating a dilution of ferrioxamine in the extraceUular compartment and subsequent renal clearance. In an other study employing ftltrable plasma iron measurements in thalassaemic patients, a direct linear correlation was found between plasma ferrioxamine levels 2 hr after a bolus injection of DF, and the logarithm of 24 urinary excretion of chelated iron (Hershko and Rachmilewitz, 1979). The gastrointestinal absorption of both DF and ferrioxamine is poor (Keberle, 1964). Oral administration of DF is effective in blocking the intestinal absorption of inorganic iron, but has only a marginal effect on urinary iron excretion. Although a slight, but consistent increase in urinary iron excretion has been shown following oral DF treatment at doses ranging from 3 to 9 g/d (Callender and Weatherall, 1980), the cost effectiveness of such an approach is very low.
Iron Chelators in Medicine
117
Conversely, the intestinal absorption of food iron is inhibited by the parenteral administration of DF (Pippard et al., 1977). In view of these observations, a number of factors should be considered in designing strategies for the optimal utilization of DF. Its rapid clearance from plasma, effective catabolism, and active tubular secretion severely limit the effectiveness of single bolus injections given at large intervals. On the other hand, its apparent distribution in over 60% of the total body volume indicates easy access to intracellular compartments and availability in disease conditions where interaction with chelatable intracellular iron is believed to be beneficial. In normal subjects most of the body iron is unavailable for chelation and urinary excretion following DF treatment represents less than 0.03% of the total iron in the body. Iron in haemoglobin, representing over 2/3 of all iron is unavailable for DF chelation in vivo or in vitro (Keberle, 1964). Likewise, although the stability constant of DF with Fe 3+ exceeds that of transferrin, transferrin-bound iron is a very poor source of iron for in vivo chelation by DF (Hershko et al., 1973). By exclusion then, the most likely source of chelatable iron is that stored in tissues in the form of ferritin or haemosiderin, or a labile iron compartment in dynamic equilibrium with the former. The assumption that storage iron is the direct source of DF-induced iron excretion is attractive, since (a) ferritin is able to donate iron to DF upon in vitro incubation (Keberle, 1964); (b) 59Felabelled ferritin associated with hepatic parenchymal cells is the single most effective source of iron for in vivo chelation in animal studies (Hershko et al., 1973) and; (c) because of a direct linear correlation between the magnitude of iron stores and the amount of iron excreted in the urine following DF treatment, documented over a wide range of values by quantitative phlebotomy or by liver biopsies (Olsson, 1972; HaUberg et al., 1966; Barry, 1974). However, a number of observations indicate that a labile iron pool in equilibrium with storage iron is a more likely direct source of chelatable iron: In the rat model of hepatocellular storage iron labelled with exogenous 59Fe-ferritin, optimal chelation occurs at 2-6 hr, prior to the incorporation of 59Fe into endogenous ferritin (Pippard et al., 1982a). Likewise, hypertransfusion in thalassaemic patients results in diminished plasma iron turnover and a simultaneous suppression of DF-induced urinary iron excretion (Pippard et al., 1982b), whereas haemolysis, induced by phenylhydrazine increases urinary DF-iron excretion (Cumming et al., 1967). In untreated megaloblastic anaemia urinary DF-iron is out of proportion to the magnitude of iron stores, but is rapidly reduced following resumption of normal erythropoiesis by specific treatment (Karabus and Fielding, 1967). A commmon feature of all of these examples is increased availability of iron for chelation by increased catabolism of ferritin- or haemoglobin-bound iron. These considerations lend support to the concept of a chelatable intermediate iron pool (Lynch et aI., 1974) from which iron may either be released to the plasma to combine with transferrin, or diverted into ferritin stores. It is assumed, that iron released from ferritin en route to plasma will fu'st enter the same intermediate iron pool. DF is able to interact with a number of cell types. Early studies with tritiated DF have shown significant uptake by liver, spleen, kidney and brain (Wohler, 1963; Keberle, 1964). More recent studies have shown that DF is able to interact in vitro with iron located in cultured liver cells (Laub etal., 1985; White and Jacobs, 1978; Octave etal., 1983), heart cells (Link etal., 1985; Sciortino et al., 1980) and monocytes (Esparza and Brock, 1981), although direct evidence for its cellular
118
C. Hershko
uptake is available only in hepatocytes (Laub et al., 1985). In iron overload, excess iron may be deposited in almost all tissues, but the bulk of iron is found in association with two cell types: reticuloendothelial (RE) cells found in the spleen, liver and bone marrow, and parenchymal tissues represented mainly by hepatocytes. In contrast to RE cells in which iron accumulation is relatively harmless, parenchymal siderosis may result in significant organ damage. It is quite important therefore to determine whether DF, or any other chelating drug may or may not interact preferentially with one of these two cellular storage compartments. The source of iron, and the proportion of iron retained in ferritin stores or recycled into the circulation from the 2 cell types is quite different. R.E. cells are unable to assimilate transferrin iron and they derive iron from the catabolism of haemoglobin in non-viable erythrocytes (Hershko, 1977). Most of this catabolic iron is recycled within a few hours. In contrast, hepatic parenchymal cells maintain a dynamic equilibrium with plasma transferrin, with iron uptake predominating when transferrin saturation is high, and release when serum iron and transferrin saturation are low (Cook et al., 1970). In contrast to RE cells, the turnover of parenchymal iron stores is extremely low. In general, iron overload associated with increased intestinal absorption such as hereditary haemochromatosis results in predominant parenchymal siderosis, whereas in conditions wherein iron overload is caused by multiple blood transfusions the primary site of siderosis is the RE cells. Considerable redistribution of iron may take place subsequently. It is possible to examine the selective interaction of an iron chelator with either parenchymal or RE iron stores by using radioiron-labelled probes targeted into one of these compartments. Quantitative radioautographic studies have shown that 59Fe-labelled ferritin, haemoglobinhaptoglobin complex and transferrin-iron are assimilated by hepatic parechymalcells exclusively. In contrast, radioiron in non-viable heat-damaged erythrocytes, ferritin-immunoprecipitate, or colloidal iron are incorporated by RE cells only (Hershko et al., 1973). Several studies employing such selective storage iron labels in rats have shown that DF interacts preferentially with hepatocellular iron stores and that the contribution of RE cells to DF-induced iron excretion is limited (Pippard et al., 1982a; Kim et al., 1985). However, there are a number of caveats in such studies which should be carefully observed: Comparison of chelating efficacy by radioiron measurements assumes similar times through which the labelled compartments are available for chelation. However, radioiron is retained in RE cells for a much shorter time than in parenchymal cells. Secondly, information on the distribution of iron excretion between urine and bile (faeces) is critical for correct interpretation of data. As hepatocytes are unable to incorporate circulating ferrioxamine (59Fe-DF), the biliary excretion of 59Fe-DF is clear evidence of intrahepatic chelation of biliary radioiron. In contrast, the interpretation of urinary radioiron excretion is more complicated. Such excretion is derived entirely from circulating ferrioxamine which in turn may originate from the chelation of iron in RE stores, or iron-in-transit to or from circulating transfen'in (Pippard et al., 1982a), which may be contributed by either RE or parenchymal cells. Considering the enormous volume of iron recycled by RE cells through haemolysis and ineffective erythropoiesis in thalassaemia and other iron loading anaemias, the bulk of iron-intransit for urinary iron excretion in such patients is most probably derived from RE cells. A number of experimental and clinical observations support the assumption that the urinary excretion of chelated iron is derived mainly from RE cells. Studies in hypertransfused rats using continuous DF infusion to capture all chelatable iron have shown that in contrast to hepatocellular
Iron Chelators in Medicine
119
radioiron excretion which is confined entirely to the bile, most of the radioiron excretion derived from the RE label is recovered in the urine (Hershko, 1975, 1978; Hershko et al., 1978). Moreover, when DTPA, a water-soluble synthetic chelator which does not enter cells, is employed in the same experimental model, there is no enhancement at all of hepatocellular iron excretion, but the enhancement of urinary RE radioiron excretion is very similar to that observed previously with DF. According to the "alternative pathways" hypothesis derived from these observations (Hershko and Weatherall, 1988) DF obtains iron for chelation by one of two alternative mechanisms; (a) in situ interaction with hepatocellular iron and subsequent biliary excretion, and; (b) chelation of iron derived from RBC catabolism in the RE system with subsequent urinary excretion. The observations described do not permit a firm conclusion as to whether RE-derived iron is chelated by DF within the R.E. cell or following its release into the plasma or upon the hepatocyte surface (Pippard et al., 1982a). Clinical observations in patients with primary and secondary haemochromatosis lend strong support to the "alternative pathways" hypothesis. In thalassaemic patients in whom the specific activity of storage iron in the spleen, representing RE stores, and the liver has been studied during splenectomy 10 to 12 days after ferrokinetic studies, the specific activity of splenic non-heme iron was identical with that of chelated urinary iron and both were 15 times higher than the specific activity of hepatic iron. These measurements indicate that chelated urinary iron is most probably derived from splenic and not from hepatic iron stores (Hershko and Rachmilewitz, 1979). In another study of thalassaemic patients (Pippard et al., 1982b), a reciprocal relation between urinary and faecal DF-induced iron excretion has been demonstrated. Hypertransfusion resulted in the suppression of plasma iron turnover, indicating reduced ineffective erythropoiesis and reduced RE breakdown of nonviable erythrocytes. This was associated with reduced urinary, and increased faecal excretion of chelated iron. The same study has also shown that increasing doses of DF administered to thalassaemic patients in whom the greatly increased plasma iron turnover is attributed to increased haemoglobin iron catabolism in RE cells, result in a predominant urinary iron excretion. In contrast, in hereditary haemochromatosis where plasma iron turnover is normal but hepatocellular iron stores are increased, the same treatment results in a predominant increase in faecal iron excretion. All of these studies indicate that urinary iron excretion is closely related to RE iron metabolism and is enhanced by increased RE haemoglobin catabolism. In contrast, faecal iron excretion depends mainly on hepatocellular iron concentrations and is actually increased by reduced plasma iron turnover. The chemical nature of the chelatable iron pool has been the subject of a number of studies. In cultured hepatocytes, radioiron supplied as ferric citrate is rapidly incorporated into cytosol ferritin. Such radioiron is readily available for chelation by DF (Octave et al., 1983). In subsequent studies employing tritiated DF-analogues, these chelators were shown to accumulate within plasma-membrane related structures and in lysosomes (Laub et al., 1985). The authors proposed that autophagy of cytosolic ferritin may greatly facilitate the chelation of ferritin iron by DF due to the acidic pH and hydrolytic enzymes in lysosomes, and that this may represent the chelatable intracellular iron pool. This assumption is supported by the observation that the size of haemosiderin iron cores within lysosomes is smaller than that of ferritin (Weir et al., 1984) which, in addition to the loss of their protective apoferritin coating makes lysosomal haemosiderin iron even more accessible for chelation. Further support for the identification of lysosomal ferritin as the target of DF in hepatocytes is the observation that the phase of hepatocellular lysosomal
120
C. Hershko
degradation of injected 59Fe-labelled exogenous ferritin coincides with the timing of maximal availability of 59Fe for in vivo chelation by DF (Pippard et al., 1982a; Unger and Hershko, 1974). In contrast to hepatocytes, there is no available information on the chemical nature of chelatable RE iron although the ability of DF to mobilize iron directly from peritoneal macrophages has been clearly demonstrated in vitro (Esparza et al., 1981). It is possible that, similar to parenchymal cells, DF may enter secondary lysosomes in RE cells to interact directly with iron derived from haeme catabolism. However, it is also possible that RE iron may be chelated following its release from these cells. The DTPA studies mentioned earlier (Hershko, 1975) have shown that RE iron may be chelated by a drug which is unable to enter these cells with an efficiency which is equal to that of DF both in experimental animals and in man (Hershko etal., 1978b; Pippard etal., 1986). The existance of a chelatable, low molecular weight plasma iron fraction (Hershko and Peto, 1987) has been documented in patients with severe iron overload by a number of investigators (Hershko et al., 1978a; Batey et al., 1978; Anuwatanakulchal et al., 1984; Wang et al., 1986; Wagstaff et al., 1985; Gutteridge et al., 1985; Singh et al., 1990). Such non-transferrin plasma iron (NTPI) is only found after complete saturation of circulating transferrin. NTPI was shown to promote the formation of free hydroxylradicals and to accelerate the peroxidation of membrane lipids in vitro (Gutteridge etal., 1985). The rate of low molecular weight iron uptake by cultured rat heart cells is over 300-times greater than that of transferrin iron (Link et al., 1985). Such uptake was shown to result in increased myocardial lipid peroxidation and abnormal contractility, and these effects were reversed by in vitro treatment with DF. Recognition of NTPI as a potentially toxic component of plasma iron in haemoch romatosis may be useful in designing better strategies for the effective administration of DF and other iron chelating drugs.
Chapter2
Effects of Long-term Chelating Therapy
There are no prospective randomized studies available to indicate the ability of iron chelating therapy to prevent or reverse the complications of chronic iron overload. In the absence of better information, one is limited to the comparison of morbidity and mortality figures in DF-treated patients with similar data in non-compliant patients or in historical controls. Such comparison, however, is inaccurate as historical controls are different from patients receiving DF treatment by present-day technology in many respects, the most important of which are the intensity and effectiveness of transfusion programs. It is also difficult to compare the results of DF therapy reported from various centers as the manner of DF administration, dosage, use of supplementary vitamin C and tocopherol, age, duration of treatment and methods of documenting response to therapy vary from group to group, and even from report to report within the same group. For the sake of comparison, it may be useful first to review information on a large group of thalassaemic children of the pre-DF era reported by Engle eta/. in 1964. The most impressive data are the very low proportion of patients surviving beyond 20 y of age, the onset of heart failure between 10 and 15 y, the fact that only 1 patient survived long enough to receive more than 100 g iron by cumulative blood transfusions, and the low levels of haemoglobin maintained by infrequent transfusions. Chronic anaemia is an important factor contributing to the early onset and severity of congestive heart failure, making comparison with present-day non-chelated thalassaemics difficult. Survival in thalassaemic patients in recent years is definitely better than in those reported by Engle et al. 27 y ago. ModeU et al. have found in 1982 that over 50% of patients lived longer than 20 y (1982). Similarly, Giardina et al. have shown recently that in thalassemic patients treated at the Comell Medical Center receiving a hypertransfusion program combined with regular s.c. DF chelation the median survival has increased to 28 years compared to 18 years in patients treated formerly by a low-transfusion regimen and no chelation (1990). However the most extensive and detailed information on survival in thalassemia comes from Italy, where over 5000 patients with thalassaemia major are treated and monitored by a remarkably well coordinated national study of the National Association for Pediatric Ontology and Hematology (AIEOP). In Italy continuous subcutaneous DF treatment has been introduced and used systematically since 1978 (Gabutti 1990). In a report by Borgna-Pignatti et al. (1989) on 942 thalassaemic patients from Italy, a steady improvement in survival has been shown for cohorts born in 1960-1964, 1965-1969 and 1970-1974 (Table 1). The total mortality was 15.7% and heart disease was the cause of death in 2/3 of cases. Overall survival from birth for patients 121
122
C. Hershko
born in 1970-1974 and therefore subjected to chelation therapy from an early age, was 97% at 10 years and 94% at 15 years (Zurlo et al., 1989). In another recent report on 211 thalassaemic patients from Australia and England receiving regular iron chelating therapy since 1977, the predicted survival at age 36 was 85% (Hoffbrand and Wonke, 1989). These impressive figures illustrate the significant improvement that has taken place in recent years in the survival of patients with thalassaemia major receiving adequate therapy. However, it is difficult to determine how much of this improvement in life expectancy may be attributed to iron chelating therapy, hyportransfusion programs avoiding the additional strain of chronic anaemia and ineffective erythropoiesis, or better management of infection. Table 1 Years of birth
Years after 10th birthday
5
10
15
78.3
55.4
40.8
(69.4-87.2)
(44.7-66.1)
29.8-51.8
1965-69
83.6 (78.5-88.6)
69.7 (62.9-76.5)
n.e.
1970-74
96.6 (94.4-98.8) (p
0098 - 2997/92 $15.00 © 1992 Pergamon Press plc
Printed in Great Britain. All rights reserved.
IRON CHELATORS IN MEDICINE Chaim Hershko Department of Medicine, Shaare Zedek Medical Center, Jerusalem, Israel
Contents
INTRODUCTION
114
CHAPTER 1 The Pharmacology of Desferrioxamine
115
CHAPTER 2 Effects of Long-term Chelating Therapy
121
CHAPTER 3
125
Desferrioxamine Toxicity
CHAPTER 4 Use of DF in Conditions Unrelated to Iron Overload
128
CHAPTER 5 Development of New Oral Iron Chelators
144
ACKNOWLEDGEMENTS
149
REFERENCES
150
113
Introduction
Although the iron chelating drug desferrioxamine (DF) has been available for clinical use for over three decades, it is only within the last 15 years that it has been gradually accepted as an essential drug in the management of transfusional iron overload. Through the pioneering work of British investigators (Barry et al., 1974; Modell et al., 1982) and the admirable devotion of Italian haematologists (Gabutti, 1990; Zurlo et aL, 1989; Borgna-Pignatti et al., 1989) it has been demonstrated beyond any doubt that long-term DF therapy is capable to prevent the severe complications of transfusional iron overload and to improve significantly the life expectancy of thalassaemic patients. However, management of acute and chronic iron toxicity is only one facet of the potential clinical use of iron chelators. Recognition of the pivotal role of iron in the generation or toxic oxygen-derived species through the Haber-Weiss reaction (Halliwell and Gutteridge, 1984, 1986), the ability of DF to limit the damage associated with free radical formation in reperfusion injury, and the possibility of inhibiting cell proliferation by interaction of DF with the iron-dependent enzyme ribonucleotide reductase, has resulted in an ever increasing number of studies exploring the possibilities of new therapeutic applications of iron chelating agents. In the text that follows, we shall review the pharmacokinetics of DF, examine the evidence supporting the clinical usefulness of long-term iron chelating therapy in transfusional iron overload; the toxicity of DF and variables that may increase or reduce the risks of such toxicity; review some of the most significant recent studies on the potential usefulness of iron chelators in conditions unrelated to iron overload and, finally; list some of the most important orally effective iron chelating drugs that are currently being considered as possible alternatives to DF therapy.
114
Chapter 1
The Pharmacology of Desferrioxamine
Desferrioxamine B (DF) is a colorless crystalline substance produced by Streptomyces pilosus. It consists of a chain of three hydroxamic acids terminating in a free amino acid group which, in turn, enables it to form salts with organic and inorganic acids (Keberle, 1964). It is capable of combining with ferric (Fe3+) iron at a 1:1 molar ratio and with a stability constant of 1031. The DF molecule (Fig. 1) is wrapped around the iron nucleus encasing it in an envelope of organic material. The affinity of DF to Fe 2+ and other metal ions such as Zn2+ , Cu 2+ , Mg 2+ and Ca2+ is much lower and ranges from 102 to 1014 . In contrast to the voluminous literature describing the clinical effects of DF therapy, information on the pharmacology of DF is relatively scanty and is largely limited to studies published in the early 1960s (Wohler, 1963; Peters et al., 1966; MeyerBrunot and Keberle, 1967). One of the key observations in these early studies was recognition of the fundamental difference in behaviour of DF as compared to ferrioxamine, the DF-iron complex. As aresult of the change in configuration following interaction with iron, ferrioxamine becomes an extremely stable compound which is resistant to enzymatic degradation (MeyerBrunot and Keberle, 1967). Ferrioxamine is distributed in the extracellular space and is unable to penetrate cells, as evidenced by a plateau in its serum concentrations following nephrectomy (Keberle, 1964). In the intact dog, 98% of labelled ferrioxamine is recovered in the urine within 3 days. In contrast to ferrioxamine, DF is capable of penetrating various tissues and is rapidly catabolized in vivo. Consequently, radioiron-labelled DF which, by definition, represents ferrioxamine cannot be utilized for studying the in vivo behaviour of DF, a consideration which has been unfortunately overlooked in the past in a number of studies.
H2N (~H2)5
.CONH /CONH (/CH2")2 "(CH2)5 (CH2)2 ~(CH2)5 CH3 " ,N-C \N - C / ~1~1-~ OH0 13H6 OH0
Fig. 1. Desferrioxamine B, 115
116
C. Hershko
The most straightforward method of studying the fate of DF is by means of the tritium-labelled drug. Such studies, performed in dogs, have shown that although 70% of labelled DF is excreted in the urine within 3 days, over half of this excretion is in the form of DF metabolites (Keberle, 1964). DF is excreted by both glomerular filtration and tubular secretion, whereas ferrioxamine is partly reabsorbed following glomerular filtration (Peters et al., 1966). The combination of both of these tubular mechanisms tends to reduce the ability of injected DF to promote the urinary excretion of chelated iron. In contrast to ferrioxamine, the concentration in plasma of DF in nephrectomized dogs does not reach a plateau. Its disappearance curve is complex, reflecting a combination of metabolic breakdown, and penetration into various tissues. Bearing these limitations in mind, estimates such as the volume distribution of DF in 62% of the body, and a metabolic half life of about 1 hr (Peters et al., 1966) may only be regarded as rough approximations. Tissue concentrations of labelled DF in dogs following intravenous injection are highest in the bile and brain, intermediate in the spleen, kidney and plasma, and lowest in the heart, lungs and fatty tissues. Although the great efficiency of the liver in clearing DF from the circulation is clearly evidenced by the rapid concentration of labelled DF in the bile, DF is not retained in the liver to a significant extent. Most of the biliary, and a great part of the urinary excretion of the tritiated compound is in the form of DF metabolites. The nature of these metabolites in the bile has not been determined. In the urine, the major compound is metabolite C, in which the original amino group of DF has been replaced by a carboxyl group. By far the most active tissue in the enzymatic breakdown of DF is the plasma, followed by various parenchymatous organs (MeyerBmnot and Keberle 1967). Rat plasma is about 4 times more efficient in DF catabolism than plasma obtained from the dog, or man. These studies employing tritiated DF yielded valuable information on animals, but the amount of radioactivity required precluded the use of similar techniques in man. A useful substitute is the measurement of filtrable (chelated) plasma iron following DF administration. Using a saturating solution of iron in a manner similar to the measurement of latent iron binding capacity, both ferrioxamine and unbound DF can be measured simultaneously. Such studies have shown that following the intravenous bolus injection of DF, plasma levels of DF drop to half of the initial concentration within 5 to 10 min (Summers et al., 1979). Conversely, slow subcutaneous infusion of DF at a dose of 100 mg/kg/24hr results in a gradual buildup of plasma DF concentrations reaching a plateau in about 12 h. This is accompanied by a simultaneous increase in plasma ferrioxamine concentrations. There is a direct correlation between the total 48 hr urinary excretion of iron and peak plasma concentrations of ferrioxamine during DF infusion. Studies with 59Felabelled ferrioxamine showed a volume distribution of about 91 and a fractional clearance rate of 0.3/hr. Both of these measurements are very similar to those obtained in the early canine studies indicating a dilution of ferrioxamine in the extraceUular compartment and subsequent renal clearance. In an other study employing ftltrable plasma iron measurements in thalassaemic patients, a direct linear correlation was found between plasma ferrioxamine levels 2 hr after a bolus injection of DF, and the logarithm of 24 urinary excretion of chelated iron (Hershko and Rachmilewitz, 1979). The gastrointestinal absorption of both DF and ferrioxamine is poor (Keberle, 1964). Oral administration of DF is effective in blocking the intestinal absorption of inorganic iron, but has only a marginal effect on urinary iron excretion. Although a slight, but consistent increase in urinary iron excretion has been shown following oral DF treatment at doses ranging from 3 to 9 g/d (Callender and Weatherall, 1980), the cost effectiveness of such an approach is very low.
Iron Chelators in Medicine
117
Conversely, the intestinal absorption of food iron is inhibited by the parenteral administration of DF (Pippard et al., 1977). In view of these observations, a number of factors should be considered in designing strategies for the optimal utilization of DF. Its rapid clearance from plasma, effective catabolism, and active tubular secretion severely limit the effectiveness of single bolus injections given at large intervals. On the other hand, its apparent distribution in over 60% of the total body volume indicates easy access to intracellular compartments and availability in disease conditions where interaction with chelatable intracellular iron is believed to be beneficial. In normal subjects most of the body iron is unavailable for chelation and urinary excretion following DF treatment represents less than 0.03% of the total iron in the body. Iron in haemoglobin, representing over 2/3 of all iron is unavailable for DF chelation in vivo or in vitro (Keberle, 1964). Likewise, although the stability constant of DF with Fe 3+ exceeds that of transferrin, transferrin-bound iron is a very poor source of iron for in vivo chelation by DF (Hershko et al., 1973). By exclusion then, the most likely source of chelatable iron is that stored in tissues in the form of ferritin or haemosiderin, or a labile iron compartment in dynamic equilibrium with the former. The assumption that storage iron is the direct source of DF-induced iron excretion is attractive, since (a) ferritin is able to donate iron to DF upon in vitro incubation (Keberle, 1964); (b) 59Felabelled ferritin associated with hepatic parenchymal cells is the single most effective source of iron for in vivo chelation in animal studies (Hershko et al., 1973) and; (c) because of a direct linear correlation between the magnitude of iron stores and the amount of iron excreted in the urine following DF treatment, documented over a wide range of values by quantitative phlebotomy or by liver biopsies (Olsson, 1972; HaUberg et al., 1966; Barry, 1974). However, a number of observations indicate that a labile iron pool in equilibrium with storage iron is a more likely direct source of chelatable iron: In the rat model of hepatocellular storage iron labelled with exogenous 59Fe-ferritin, optimal chelation occurs at 2-6 hr, prior to the incorporation of 59Fe into endogenous ferritin (Pippard et al., 1982a). Likewise, hypertransfusion in thalassaemic patients results in diminished plasma iron turnover and a simultaneous suppression of DF-induced urinary iron excretion (Pippard et al., 1982b), whereas haemolysis, induced by phenylhydrazine increases urinary DF-iron excretion (Cumming et al., 1967). In untreated megaloblastic anaemia urinary DF-iron is out of proportion to the magnitude of iron stores, but is rapidly reduced following resumption of normal erythropoiesis by specific treatment (Karabus and Fielding, 1967). A commmon feature of all of these examples is increased availability of iron for chelation by increased catabolism of ferritin- or haemoglobin-bound iron. These considerations lend support to the concept of a chelatable intermediate iron pool (Lynch et aI., 1974) from which iron may either be released to the plasma to combine with transferrin, or diverted into ferritin stores. It is assumed, that iron released from ferritin en route to plasma will fu'st enter the same intermediate iron pool. DF is able to interact with a number of cell types. Early studies with tritiated DF have shown significant uptake by liver, spleen, kidney and brain (Wohler, 1963; Keberle, 1964). More recent studies have shown that DF is able to interact in vitro with iron located in cultured liver cells (Laub etal., 1985; White and Jacobs, 1978; Octave etal., 1983), heart cells (Link etal., 1985; Sciortino et al., 1980) and monocytes (Esparza and Brock, 1981), although direct evidence for its cellular
118
C. Hershko
uptake is available only in hepatocytes (Laub et al., 1985). In iron overload, excess iron may be deposited in almost all tissues, but the bulk of iron is found in association with two cell types: reticuloendothelial (RE) cells found in the spleen, liver and bone marrow, and parenchymal tissues represented mainly by hepatocytes. In contrast to RE cells in which iron accumulation is relatively harmless, parenchymal siderosis may result in significant organ damage. It is quite important therefore to determine whether DF, or any other chelating drug may or may not interact preferentially with one of these two cellular storage compartments. The source of iron, and the proportion of iron retained in ferritin stores or recycled into the circulation from the 2 cell types is quite different. R.E. cells are unable to assimilate transferrin iron and they derive iron from the catabolism of haemoglobin in non-viable erythrocytes (Hershko, 1977). Most of this catabolic iron is recycled within a few hours. In contrast, hepatic parenchymal cells maintain a dynamic equilibrium with plasma transferrin, with iron uptake predominating when transferrin saturation is high, and release when serum iron and transferrin saturation are low (Cook et al., 1970). In contrast to RE cells, the turnover of parenchymal iron stores is extremely low. In general, iron overload associated with increased intestinal absorption such as hereditary haemochromatosis results in predominant parenchymal siderosis, whereas in conditions wherein iron overload is caused by multiple blood transfusions the primary site of siderosis is the RE cells. Considerable redistribution of iron may take place subsequently. It is possible to examine the selective interaction of an iron chelator with either parenchymal or RE iron stores by using radioiron-labelled probes targeted into one of these compartments. Quantitative radioautographic studies have shown that 59Fe-labelled ferritin, haemoglobinhaptoglobin complex and transferrin-iron are assimilated by hepatic parechymalcells exclusively. In contrast, radioiron in non-viable heat-damaged erythrocytes, ferritin-immunoprecipitate, or colloidal iron are incorporated by RE cells only (Hershko et al., 1973). Several studies employing such selective storage iron labels in rats have shown that DF interacts preferentially with hepatocellular iron stores and that the contribution of RE cells to DF-induced iron excretion is limited (Pippard et al., 1982a; Kim et al., 1985). However, there are a number of caveats in such studies which should be carefully observed: Comparison of chelating efficacy by radioiron measurements assumes similar times through which the labelled compartments are available for chelation. However, radioiron is retained in RE cells for a much shorter time than in parenchymal cells. Secondly, information on the distribution of iron excretion between urine and bile (faeces) is critical for correct interpretation of data. As hepatocytes are unable to incorporate circulating ferrioxamine (59Fe-DF), the biliary excretion of 59Fe-DF is clear evidence of intrahepatic chelation of biliary radioiron. In contrast, the interpretation of urinary radioiron excretion is more complicated. Such excretion is derived entirely from circulating ferrioxamine which in turn may originate from the chelation of iron in RE stores, or iron-in-transit to or from circulating transfen'in (Pippard et al., 1982a), which may be contributed by either RE or parenchymal cells. Considering the enormous volume of iron recycled by RE cells through haemolysis and ineffective erythropoiesis in thalassaemia and other iron loading anaemias, the bulk of iron-intransit for urinary iron excretion in such patients is most probably derived from RE cells. A number of experimental and clinical observations support the assumption that the urinary excretion of chelated iron is derived mainly from RE cells. Studies in hypertransfused rats using continuous DF infusion to capture all chelatable iron have shown that in contrast to hepatocellular
Iron Chelators in Medicine
119
radioiron excretion which is confined entirely to the bile, most of the radioiron excretion derived from the RE label is recovered in the urine (Hershko, 1975, 1978; Hershko et al., 1978). Moreover, when DTPA, a water-soluble synthetic chelator which does not enter cells, is employed in the same experimental model, there is no enhancement at all of hepatocellular iron excretion, but the enhancement of urinary RE radioiron excretion is very similar to that observed previously with DF. According to the "alternative pathways" hypothesis derived from these observations (Hershko and Weatherall, 1988) DF obtains iron for chelation by one of two alternative mechanisms; (a) in situ interaction with hepatocellular iron and subsequent biliary excretion, and; (b) chelation of iron derived from RBC catabolism in the RE system with subsequent urinary excretion. The observations described do not permit a firm conclusion as to whether RE-derived iron is chelated by DF within the R.E. cell or following its release into the plasma or upon the hepatocyte surface (Pippard et al., 1982a). Clinical observations in patients with primary and secondary haemochromatosis lend strong support to the "alternative pathways" hypothesis. In thalassaemic patients in whom the specific activity of storage iron in the spleen, representing RE stores, and the liver has been studied during splenectomy 10 to 12 days after ferrokinetic studies, the specific activity of splenic non-heme iron was identical with that of chelated urinary iron and both were 15 times higher than the specific activity of hepatic iron. These measurements indicate that chelated urinary iron is most probably derived from splenic and not from hepatic iron stores (Hershko and Rachmilewitz, 1979). In another study of thalassaemic patients (Pippard et al., 1982b), a reciprocal relation between urinary and faecal DF-induced iron excretion has been demonstrated. Hypertransfusion resulted in the suppression of plasma iron turnover, indicating reduced ineffective erythropoiesis and reduced RE breakdown of nonviable erythrocytes. This was associated with reduced urinary, and increased faecal excretion of chelated iron. The same study has also shown that increasing doses of DF administered to thalassaemic patients in whom the greatly increased plasma iron turnover is attributed to increased haemoglobin iron catabolism in RE cells, result in a predominant urinary iron excretion. In contrast, in hereditary haemochromatosis where plasma iron turnover is normal but hepatocellular iron stores are increased, the same treatment results in a predominant increase in faecal iron excretion. All of these studies indicate that urinary iron excretion is closely related to RE iron metabolism and is enhanced by increased RE haemoglobin catabolism. In contrast, faecal iron excretion depends mainly on hepatocellular iron concentrations and is actually increased by reduced plasma iron turnover. The chemical nature of the chelatable iron pool has been the subject of a number of studies. In cultured hepatocytes, radioiron supplied as ferric citrate is rapidly incorporated into cytosol ferritin. Such radioiron is readily available for chelation by DF (Octave et al., 1983). In subsequent studies employing tritiated DF-analogues, these chelators were shown to accumulate within plasma-membrane related structures and in lysosomes (Laub et al., 1985). The authors proposed that autophagy of cytosolic ferritin may greatly facilitate the chelation of ferritin iron by DF due to the acidic pH and hydrolytic enzymes in lysosomes, and that this may represent the chelatable intracellular iron pool. This assumption is supported by the observation that the size of haemosiderin iron cores within lysosomes is smaller than that of ferritin (Weir et al., 1984) which, in addition to the loss of their protective apoferritin coating makes lysosomal haemosiderin iron even more accessible for chelation. Further support for the identification of lysosomal ferritin as the target of DF in hepatocytes is the observation that the phase of hepatocellular lysosomal
120
C. Hershko
degradation of injected 59Fe-labelled exogenous ferritin coincides with the timing of maximal availability of 59Fe for in vivo chelation by DF (Pippard et al., 1982a; Unger and Hershko, 1974). In contrast to hepatocytes, there is no available information on the chemical nature of chelatable RE iron although the ability of DF to mobilize iron directly from peritoneal macrophages has been clearly demonstrated in vitro (Esparza et al., 1981). It is possible that, similar to parenchymal cells, DF may enter secondary lysosomes in RE cells to interact directly with iron derived from haeme catabolism. However, it is also possible that RE iron may be chelated following its release from these cells. The DTPA studies mentioned earlier (Hershko, 1975) have shown that RE iron may be chelated by a drug which is unable to enter these cells with an efficiency which is equal to that of DF both in experimental animals and in man (Hershko etal., 1978b; Pippard etal., 1986). The existance of a chelatable, low molecular weight plasma iron fraction (Hershko and Peto, 1987) has been documented in patients with severe iron overload by a number of investigators (Hershko et al., 1978a; Batey et al., 1978; Anuwatanakulchal et al., 1984; Wang et al., 1986; Wagstaff et al., 1985; Gutteridge et al., 1985; Singh et al., 1990). Such non-transferrin plasma iron (NTPI) is only found after complete saturation of circulating transferrin. NTPI was shown to promote the formation of free hydroxylradicals and to accelerate the peroxidation of membrane lipids in vitro (Gutteridge etal., 1985). The rate of low molecular weight iron uptake by cultured rat heart cells is over 300-times greater than that of transferrin iron (Link et al., 1985). Such uptake was shown to result in increased myocardial lipid peroxidation and abnormal contractility, and these effects were reversed by in vitro treatment with DF. Recognition of NTPI as a potentially toxic component of plasma iron in haemoch romatosis may be useful in designing better strategies for the effective administration of DF and other iron chelating drugs.
Chapter2
Effects of Long-term Chelating Therapy
There are no prospective randomized studies available to indicate the ability of iron chelating therapy to prevent or reverse the complications of chronic iron overload. In the absence of better information, one is limited to the comparison of morbidity and mortality figures in DF-treated patients with similar data in non-compliant patients or in historical controls. Such comparison, however, is inaccurate as historical controls are different from patients receiving DF treatment by present-day technology in many respects, the most important of which are the intensity and effectiveness of transfusion programs. It is also difficult to compare the results of DF therapy reported from various centers as the manner of DF administration, dosage, use of supplementary vitamin C and tocopherol, age, duration of treatment and methods of documenting response to therapy vary from group to group, and even from report to report within the same group. For the sake of comparison, it may be useful first to review information on a large group of thalassaemic children of the pre-DF era reported by Engle eta/. in 1964. The most impressive data are the very low proportion of patients surviving beyond 20 y of age, the onset of heart failure between 10 and 15 y, the fact that only 1 patient survived long enough to receive more than 100 g iron by cumulative blood transfusions, and the low levels of haemoglobin maintained by infrequent transfusions. Chronic anaemia is an important factor contributing to the early onset and severity of congestive heart failure, making comparison with present-day non-chelated thalassaemics difficult. Survival in thalassaemic patients in recent years is definitely better than in those reported by Engle et al. 27 y ago. ModeU et al. have found in 1982 that over 50% of patients lived longer than 20 y (1982). Similarly, Giardina et al. have shown recently that in thalassemic patients treated at the Comell Medical Center receiving a hypertransfusion program combined with regular s.c. DF chelation the median survival has increased to 28 years compared to 18 years in patients treated formerly by a low-transfusion regimen and no chelation (1990). However the most extensive and detailed information on survival in thalassemia comes from Italy, where over 5000 patients with thalassaemia major are treated and monitored by a remarkably well coordinated national study of the National Association for Pediatric Ontology and Hematology (AIEOP). In Italy continuous subcutaneous DF treatment has been introduced and used systematically since 1978 (Gabutti 1990). In a report by Borgna-Pignatti et al. (1989) on 942 thalassaemic patients from Italy, a steady improvement in survival has been shown for cohorts born in 1960-1964, 1965-1969 and 1970-1974 (Table 1). The total mortality was 15.7% and heart disease was the cause of death in 2/3 of cases. Overall survival from birth for patients 121
122
C. Hershko
born in 1970-1974 and therefore subjected to chelation therapy from an early age, was 97% at 10 years and 94% at 15 years (Zurlo et al., 1989). In another recent report on 211 thalassaemic patients from Australia and England receiving regular iron chelating therapy since 1977, the predicted survival at age 36 was 85% (Hoffbrand and Wonke, 1989). These impressive figures illustrate the significant improvement that has taken place in recent years in the survival of patients with thalassaemia major receiving adequate therapy. However, it is difficult to determine how much of this improvement in life expectancy may be attributed to iron chelating therapy, hyportransfusion programs avoiding the additional strain of chronic anaemia and ineffective erythropoiesis, or better management of infection. Table 1 Years of birth
Years after 10th birthday
5
10
15
78.3
55.4
40.8
(69.4-87.2)
(44.7-66.1)
29.8-51.8
1965-69
83.6 (78.5-88.6)
69.7 (62.9-76.5)
n.e.
1970-74
96.6 (94.4-98.8) (p
