Portal de Periódicos da CAPES

Oral iron chelation therapy for thalassaemia: an uncertain scene

2000; Wiley; Volume: 111; Issue: 1 Linguagem: Inglês

10.1111/j.1365-2141.2000.02406.x

1365-2141

Martin J. Pippard, D. J. Weatherall,

Iron Metabolism and Disorders

There is now abundant evidence, reviewed by Olivieri & Brittenham (1997a) and Weatherall & Clegg (2000), that the otherwise fatal iron loading that results from long-term transfusion regimens in patients with β thalassaemia can be controlled by regular subcutaneous infusions of desferrioxamine. However, because of the expense and complexity of this form of treatment, it is not widely available in poorer countries and, even in those countries that can afford it, compliance remains an important problem. For this reason there has been an intensive search for an effective oral chelating agent. The only drug of this kind for which there is much clinical experience so far is deferiprone (1,2-dimethyl-3-hydroxypyridin-4-one). The development of deferiprone, including early animal and clinical studies, has been reviewed by Porter et al (1989). Preliminary studies in patients with thalassaemia major suggested that, at least as judged by its effect on serum ferritin levels, deferiprone might be capable of controlling iron loading (Kontoghiorghes et al, 1990; Olivieri et al, 1990; Tondury et al, 1990; Agarwal et al, 1992; Al-Refaie et al, 1992, 1995). This possibility was supported by observations that utilized hepatic biopsy to assess tissue-iron concentrations (Olivieri et al, 1992, 1995a). Starting in 1991, a preclinical evaluation of deferiprone was carried out by Ciba-Geigy (later Novartis). After a limited number of animal studies, the company concluded that, based on the information from these investigations and from the published accounts of toxicity in patients reported up to that time, further development of this compound was not justified (Berdoukas et al, 1993). However, this conclusion was strongly criticised by the International Study Group on Oral Iron Chelators, largely because the company's decision was based on toxicity studies in non-iron-loaded animals. Particularly in view of the lack of any other effective oral chelating agent on the horizon, the Group concluded that there was an urgent need for further well-controlled prospective clinical studies of deferiprone in sufficient numbers of patients to allow adequate judgement of its efficacy and safety for long-term clinical use (Hershko, 1993). Seven years later, the continuing lack of data from controlled trials of this kind is the main reason for the present uncertainty about the role of deferiprone in the treatment of β thalassaemia. Because of the particularly complex pathophysiology of β thalassaemia, it is vital that studies of new chelating agents are adequately controlled. This is particularly relevant to the question of potential tissue damage which, from the results of animal studies, might be a side-effect of deferiprone. For example, hepatic or cardiac damage occurring in a patient with β thalassaemia might reflect excessive tissue iron a side-effect of a drug (mediated by the drug itself, by its iron chelate, or through redistribution of iron between tissues), or, in the case of the liver, the results of viral hepatitis, which is particularly common in chronically transfused β thalassaemics. These considerations, together with the vagaries of the assessment of liver histology, require that the studies of the efficacy and toxicity of deferiprone are adequately controlled and that tissues are studied ‘blind’ by panels of pathologists with expertise in hepatic pathology. Furthermore, given the lack of consistency between serum ferritin levels and hepatic iron concentrations, adequate evaluation of any chelating agent requires serial hepatic iron measurements. Presumably based on these difficulties, the United States Food and Drug Administration (FDA), in a review of the drug carried out in 1993, stated that a prospective, randomized trial to compare therapy with deferiprone and desferrioxamine, and a second prospective study to estimate the incidence of serious adverse effects of deferiprone in an adequate sample of patients, would be required to assess the drug. Both these studies were initiated under partial support by the Canadian pharmaceutical company Apotex Pharmaceuticals. Unfortunately, the controlled comparison trial was terminated prematurely by the company, but the toxicity study was completed in 1996. It is from these limited data and a series of uncontrolled observational studies that we have to assess the current status of deferiprone for the prevention of iron loading in patients with β thalassaemia. The results of the randomized trial of deferiprone and desferrioxamine that was terminated prematurely have been reported in abstract form and were initially encouraging. In patients with continued iron loading from regular blood transfusions, hepatic iron concentrations (judged by liver biopsy or magnetic susceptometry) remained constant after a mean of 22 months of treatment with either subcutaneous desferrioxamine infusions (50 mg/kg/d) or oral deferiprone (75 mg/kg/d). In addition, there was also a suggestion from magnetic resonance imaging studies that the treatment with deferiprone might reduce cardiac iron (Olivieri et al, 1995b). However, further follow-up showed a mean increase in hepatic iron concentration of approximately 50% over baseline in the deferiprone-treated group, with no significant change in those treated with desferrioxamine (Olivieri, 1996; Olivieri & Brittenham, 1997b). These results mirror those obtained with deferiprone in an earlier cohort of patients (Olivieri et al, 1995a). Over a mean of 3·1 years of treatment, a substantial fall in liver iron concentration was seen in those patients who were initially most heavily iron-loaded, while in those who started with less hepatic iron, liver iron concentrations were maintained below 15 mg/g dry weight, (Brittenham et al, 1994; it is surprising that so few patients have hepatic fibrosis). However, in those patients with the lowest hepatic iron concentration, there was a tendency for this measure to increase during the period of study and, in further follow-ups of these deferiprone-treated patients, one-third were found to have hepatic iron concentrations above the toxicity threshold (Olivieri et al, 1998a). Two observational studies that included at least one hepatic iron measurement have confirmed these findings. In a long-term study of deferiprone in 51 transfusion-dependent β thalassaemics, reported by Hoffbrand et al (1998), 20 patients discontinued the drug and five patients died. Twenty-six patients continued deferiprone for a mean of 39·4 months; 50% of these individuals had hepatic iron concentrations at the end of the study of above the threshold of 15·0 mg/g dry weight. A similar limited efficacy, as judged by hepatic iron concentrations, was reported in a very small series of patients by Tondury et al (1998). However, the finding that deferiprone treatment could reduce liver iron to near normal in a patient with thalassaemia intermedia (Olivieri et al, 1992) indicates that it is the relationship between rate of iron loading (slower in thalassaemia intermedia than in transfusion-dependent thalassaemia major) and chelator efficiency that determines the change in liver iron concentration, and not a fundamental inability of deferiprone to chelate iron at low tissue iron concentrations. Although more data using sequential liver iron estimations are required, from these reports it is already clear that, in a significant proportion of chronically transfused patients with β thalassaemia, it is improbable that deferiprone will sustain a safe body iron load, at least at the dosage used in these studies. This conclusion was also reached by Hershko et al (1998) in their extensive review of these and related studies. The current position regarding the toxicity of deferiprone is less clear. Although there have been reports of deaths of patients who were receiving the drug, there have been no published cases that could be directly attributed to its action. For example, in the four deaths that occurred in patients who were receiving or had just received deferiprone described by Agarwal et al (1992), three cases were ascribed to infection and one case to iron overload. In the three who died of infection it was stated that there was no neutropenia, although the white cell counts at the time of death are not reported. In the 51 patients described by Hoffbrand et al (1998) there were five deaths, four of which were as a result of congestive heart failure associated with iron overload. The fifth patient died of adult respiratory distress syndrome. None of these patients were neutropenic. An early report of a systemic lupus erythematosus-like picture in an Indian patient who died while receiving deferiprone (Mehta et al, 1991) and a further paper from the same group suggesting that some of the deaths reported by Agarwal et al (1992) might have had a similar basis (Mehta et al, 1993) were the subject of a rebuttal by Agarwal et al (1993). The latter workers pointed out that no other cases resembling lupus erythematosus had been described up to 1993 and that their patient who died from a disseminated varicella infection had suffered from disseminated tuberculosis several years before receiving deferiprone, suggesting that this was indicative of ‘a certain amount of immune deficiency’. Interestingly, Berdoukas et al (1993) mentioned a patient of A. Cohen who had an overwhelming infection associated with thymic atrophy, but no further details were reported. Given the complex multiple pathology of iron-loaded β thalassaemics and their well-documented propensity to infection, particularly after splenectomy, and the lack of clinical data in some of these reports, it is impossible to determine whether deferiprone contributed to any of these deaths, particularly as none of the deaths occurred in controlled studies. The results of the prospective, observational study to estimate the incidence of adverse effects of the drug have been reported by Cohen et al (1998, 2000). Among 187 patients with thalassaemia major treated for 1 year, the incidence of agranulocytosis was 0·6 out of 100 patient years and that of milder forms of neutropenia was 5·4 out of 100 patient years. Apart from modest gastrointestinal symptoms and arthropathy in 20 of the patients, and some non-specific immunological abnormalities, the main observation of note was abnormal liver function, as reflected by alanine transaminase (ALT) levels. Although the summary of the second report (Cohen et al, 2000) stated that elevated ALT levels were generally transient and occurred more commonly in patients who showed serological evidence of hepatitis C infection, it appeared from the text that the mean levels of ALT were significantly higher than baseline at 3 months, 6 months and 9 months. Liver biopsies were not performed in this study. Curiously, the initial publication of this study (Cohen et al, 1998) describes a rise in ALT level to at least twice the baseline level in 48% of cases and this was reported not to be related to hepatitis C status. In another observational study, transient increases in serum ALT values to greater than twice the upper limit of normal were reported in 50 out of 84 patients, most of whom had negative hepatitis C serology, in the first 6 months of deferiprone therapy, but these apparently resolved without dose adjustment in all but one patient, who showed persistent liver dysfunction (Al Refai et al, 1995). In a controlled study designed to assess the possibility of hepatic damage by deferiprone, Olivieri et al (1998a) compared the hepatic histology from serial biopsies taken from a group of 14 patients who had been treated with deferiprone compared with that of 12 patients who had been receiving desferrioxamine, all over a period of up to or exceeding 4 years. The liver histology was assessed ‘blind’ by a panel of four hepatic pathologists. Five of the deferiprone-treated patients had progressive hepatic fibrosis as compared with none of the patients who had been treated with desferrioxamine. Importantly, the mean hepatic iron concentration of those patients who developed fibrosis was only half that of the patients who showed no progression to fibrosis. While this study was criticized because of the small numbers involved and the possibility that hepatitis C might have contributed to the fibrosis, the worrying observation that the liver iron concentrations were significantly lower in the deferiprone-treated group who had developed fibrosis, and that fibrosis had occurred in the absence of positive serology for hepatitis C in one case, seemed to justify the authors conclusions (and those of an accompanying editorial; Kowdley & Kaplan, 1998) that further controlled studies along these lines are required. The only other studies to address the problem of hepatic fibrosis were based on a single biopsy taken at various times after exposure to deferiprone. In a study by Tondury et al (1998), the liver pathology was assessed in 11 patients who had received deferiprone from 54 to 60 months, six of whom were hepatitis C-positive and five who were negative. Four of the patients who were positive showed chronic active hepatitis and, as expected, there was a significantly higher fibrosis score in this group than in the hepatitis C-negative patients. However, of the latter, four out of five patients showed perisinusoidal fibrosis, severe in one case. Only one of this latter group had serum iron values in excess of 15 mg/g dry weight. Hoffbrand et al (1998) reported the liver histology in 17 patients receiving deferiprone for a mean period of 39·4 months, five of whom were hepatitis C-RNA positive, while 12 patients were negative. Of the former group, all showed cirrhosis, while among the latter group, none had cirrhosis and showed either no fibrosis or, in one case, mild fibrosis. As the range of hepatic iron in the 17 patients ranged from 5·9 to 41·2 mg/g dry weight and 50% had levels above 15 mg/g dry weight, it is surprising that so few patients showed no hepatic fibrosis. Several other small studies have described hepatic histology in patients who received deferiprone, but the follow-up period was too short to draw any conclusions (Stella et al, 1998; Mazza et al, 1998; Aydinok et al, 1999). Another major hiatus in our knowledge about the properties of deferiprone is its ability to control cardiac iron loading and whether it has any potential toxic effects on the myocardium (Nathan, 1995). The only well-documented patient to have been reported to date who had undergone cardiac biopsy during deferiprone therapy had developed acute congestive cardiac failure after the drug had been administered for 3–5 years; cardiac biopsy showed extensive fibrosis. These findings are of concern, particularly as the hepatic iron level in this patient had been maintained within concentrations associated with a reduced risk for cardiac disease (Olivieri et al, 1998b). Clearly, the role of deferiprone in the prevention of iron loading in patients with β thalassaemia maintained on blood transfusion is still uncertain. From the few adequate efficacy studies that have been carried out, it seems improbable that deferiprone will maintain hepatic iron concentrations at the levels achieved by desferrioxamine and many transfusion-dependent patients will be left with levels above the ′toxic threshold', at least with the dose regimens that have been used so far. The belief that any iron chelation treatment must be better than none (Tricta & Spino, 1998) is undermined by the serious doubts that remain about deferiprone's potential for long-term toxicity. In view of the possibility that it might cause tissue fibrosis even at low levels of hepatic iron, a concern that is underlined by the ability of related agents to produce this effect in experimental animals (reviewed by Olivieri & Brittenham, 1997a), it seems premature to encourage the widespread use of deferiprone as a second-line agent for the control of iron loading without a further and adequately controlled clinical trial comparing its effect on liver histology and cardiac function with that of desferrioxamine over 4–5 years. While the acquisition of information of this type is important for the richer countries, it is even more vital for those of the developing world. The temptation to prescribe deferiprone because of its lower cost and ease of administration is particularly strong in countries in which the prohibitive cost of desferrioxamine and the pumps required to administer it make it very difficult to manage thalassaemic patients adequately. Furthermore, the existence of an agent of this kind reduces pressures on the governments of these countries and international health agencies to provide the funding for desferrioxamine and, hence, deprives patients of a form of treatment which is of proven value. While deferiprone is an alternative option for those who cannot tolerate desferrioxamine, genuine allergy to the latter is very rare (Weatherall & Clegg 2000); it would be a particularly worrying trend if those looking after patients who find it difficult to comply with the current regimens for administering desferrioxamine were to be tempted to change their patients to a less effective agent rather than continue to persuade them to pursue a less pleasant, but more effective regimen. This concern extends to the proposed use of combined therapy with desferrioxamine and deferiprone (Wonke et al, 1998) because, if the drug has an independent fibrogenic effect, deferiprone might worsen tissue damage despite improved iron balance. As evidenced by the problems that have been faced in interpreting the pathological basis of the complex published interactions of iron loading, infection and deferiprone in the generation of liver fibrosis in patients with thalassaemia, establishment by the thalassaemia community of a further controlled trial seems essential if the true role of deferiprone is to be established; further observational studies will not answer this question.