Description of the condition
Iron overload constitutes a major health problem for all people who require regular blood transfusions, such as people with thalassaemia. Effective iron removal, known as chelation therapy, is essential to the survival and quality of life of iron overloaded people. Before iron chelation therapy was available, the prognosis for transfusion-dependent individuals was poor and less than 5% of children survived beyond 15 years of age. Control of iron load using desferrioxamine (DFO) showed that if the children's total body iron was controlled adequately, survival was increased and the frequency of cardiac deaths fell (Brittenham 1994; Olivieri 1994). The clinical effectiveness of DFO was confirmed in long-term follow-up studies, where the cohort born after 1970 shows that 68% of the individuals are alive at the age of 35 years (Borgna-Pignatti 1998; Borgna-Pignatti 2004).
In people with transfusion-dependent thalassaemia major, failure to control the iron load has been demonstrated to be the main cause of death (Olivieri 1994). In these people, where serum ferritin was maintained at less than 2500 mg/L, there was a survival rate without cardiac disease of 91% at 15 years, whereas in those whose serum ferritin levels were greater than 2500 mg/L there was a survival rate without cardiac disease of only 20% over the same period (Olivieri 1994).
Description of the intervention
Currently, the iron chelator of first choice for clinical use is desferrioxamine (DFO). To be clinically effective, DFO must be administered as a subcutaneous infusion over 8 to 12 hours, 5 to 7 days a week. This regimen has been demonstrated to regulate iron balance, reduce the body iron load, prevent the onset of iron-induced complications, reverse some of the induced organ damage and increase survival as a result of iron excretion (Aldouri 1990; Borgna-Pignatti 1998; Brittenham 1994; Ehlers 1991; Gabutti 1996; Modell 2000; Olivieri 1999; Pippard 1978; Propper 1976; Propper 1977; Wolfe 1985).
Unfortunately, chronic therapy with DFO has had a number of problems, particularly the adherence to an arduous daily regimen of infusions (Olivieri 1997b; Weatherall 2002) and adverse events due to DFO may cause renal impairment (Bacon 1983; Richardson 1993), local skin reactions (Kushner 2001), growth retardation (Olivieri 1992; Piga 1988), increased susceptibility to Yersinia infections (Gallant 1986), high frequency sensorineural hearing loss (Kontzogolou 1996; Olivieri 1986; Porter 1989b) and retinal damage (Davies 1983; Olivieri 1986). However, reports of these adverse events, which appear frequently in the literature from the early years of DFO use, have declined in recent years as the relationship between toxicity and drug dose has been more clearly understood and managed (Porter 1989a; Porter 2002). The principle problem with desferrioxamine remains the low rate of compliance - only 70% in some series of participants (Olivieri 1995b). A Cochrane Review has examined the efficacy of different regimens of desferrioxamine in people with transfusion-dependent thalassaemia (Fisher 2013).
The introduction of a pharmacologically efficacious and well-tolerated oral iron chelator, which would negate some of the problems associated with DFO, has been widely seen as highly desirable. The drug 1,2-dimethyl-3-hydroxypyroid-4-one, or deferiprone (also known as L1, Kelfer or DMHP) has been the first oral iron chelator to be clinically evaluated. It was first synthesised in 1984 (Hider 1984) and was soon shown to be pharmacologically efficacious in achieving iron excretion (Agarwal 1992; Kontoghiorghes 1990; Tondury 1990). Since these early studies, a large number of centres have reported their experience with deferiprone (Cohen 2003; Del Vecchio 2000; Fischer 2003; Lucas 2002; Olivieri 1995a; Olivieri 1997a; Rombos 2000).
As with desferrioxamine, adverse events have been reported in people with thalassaemia taking deferiprone. These include gastrointestinal disturbances (Ceci 2002; Cohen 2003; Taher 2001), arthropathy (Cohen 2003; Hoffbrand 1998; Lucas 2002; Mazza 1998; Taher 2001), raised liver enzymes (Ceci 2002; Cohen 2003), neutropenia and agranulocytosis (Cohen 2003; Pati 1999). Progression of liver fibrosis during treatment with deferiprone has been cited by some studies (Berdoukas 2000; Olivieri 1998; Tondury 1998), but not others (Wanless 2002); and ensuing correspondence (Brittenham 2003). Variation in the length of treatment (Berdoukas 2000) and the failure to record baseline values of liver fibrosis (Hoffbrand 1998; Tondury 1998) has made precise evaluation of the progression and significance of liver fibrosis on treatment difficult.
Why it is important to do this review
Different iron chelators may have variable ability to chelate iron from specific tissues although the side effects of the drugs in combination may also be, in some way, additive. Therefore, the benefits and safety profiles of the combination of iron chelators compared with monotherapy cannot be predicted but can only be determined in high-quality, adequately-powered, long-term, randomised controlled trials. Indeed any new treatment must be evaluated against the best available treatment by randomised clinical trials.
Comparisons of iron chelation regimens face a number of methodological problems, including accurately assessing the clinical efficacy and effectiveness of therapy, determining the consequences of underlying disease, attributing side effects to therapy and monitoring participants for a sufficient period of time to reach clinically useful endpoints. It is therefore essential for trials to carefully report baseline values, choose appropriate outcome measures and consider observed changes in outcome measures in the context of underlying conditions. Moreover, the relatively short period of clinical use of deferiprone to date limits long-term comparison of not only safety but also effectiveness with desferrioxamine, which has been available for clinical use since the 1970s.
The need for long-term trials cannot be overemphasised. Iron chelators in thalassaemia are rather different to many drug trials simply because of the time-scale involved. Iron accumulates slowly and the onset of complications again is very variable in its timing. Thus the only reliable way to learn more about the relative value of the two iron-chelating drugs would be a long-term, i.e. more than a 5-year and preferably a 10-year prospective trial, starting in early life before there is really significant iron loading and carried through to the critical period of adolescence and beyond.
The importance of long-term considerations validate the need for a systematic review of the clinical efficacy and safety profile of the iron chelator deferiprone to determine and analyse the available evidence regarding such treatment.
One early systematic review examining the use of deferiprone in thalassaemia was published prior to the completion of 12 of the 17 relevant trials included in the current review and considered only one outcome, namely reduction of hepatic iron overload (Caro 2002). Two reviews have examined the limited number of studies looking at the psychosocial aspects of chelation therapy. Poor adherence to iron chelation therapy was documented to negatively impact survival (Abetz 2006), although older age was consistently associated with lower levels of chelation adherence (Evangeli 2010).
The role of iron chelation on cardiac function has been of general interest after reports that indirect measures of cardiac iron loading, the myocardial T2*, showed a statistically significant difference, favouring deferiprone over DFO in one trial (Pennell 2006) and favouring deferiprone combined with DFO over DFO alone in another trial (Tanner 2007). However, a meta-analysis of the influence of iron chelators on myocardial iron and cardiac function in thalassaemia reported that while DFO and deferiprone both reduced myocardial iron by similar amounts, there was no significant difference between the two chelators nor was there any improvement in the left ventricular ejection fraction (LVEF) as a measure of cardiac function (Mamtani 2008). They were able to analyse data on over 290 patients and also suggest there was a publication bias in smaller studies, favouring reporting of improvements in LVEF but not myocardial iron (Mamtani 2008).
The Italian Society of Haematology produced a systematic review of the available data surrounding the wider questions of iron chelation therapy in thalassaemia patients but did not undertake formal meta-analysis of the identified studies (Angelucci 2008). A later systematic review and meta-analysis of all iron chelation therapies in thalassaemia, focusing on the role of combined or sequential therapy versus monotherapy and examining 16 trials with 1500 patients, concluded that lower final liver iron concentrations were associated with combined chelator therapy compared with monotherapy. There were no significant differences in heart T2* signal during combined or sequential therapy versus monotherapy, although there was an improvement in the LVEF after combined or sequential therapy compared to monotherapy (P = 0.01 and P < 0.00001 respectively) (Maggio 2011).
These somewhat disparate findings in generally small series of patients highlight the need for continued systematic analysis of chelation therapy for thalassaemia.
The aims of this systematic review were to summarise data from trials on the efficacy and safety of deferiprone as an iron chelating agent in people with transfusion-dependent thalassaemia and to compare the efficacy and safety of deferiprone with desferrioxamine. Both aims were severely compromised by incompatible trial design or reporting of data, or both, and clinical diversity of participants between the trials.
This updated review complements a concurrent systematic review of iron chelation entitled 'Desferrioxamine mesylate for managing transfusional iron overload in people with transfusion-dependent thalassaemia', which we previously undertook and which we have now also updated (Fisher 2013). All but one of the trials presented in this current review are common to both papers (Choudhry 2004).
Seventeen RCTs were eligible for analysis in this review. Within these trials there were five different intervention comparisons: five trials compared deferiprone to DFO; two trials compared deferiprone alone to deferiprone combined with DFO; six trials compared deferiprone combined with DFO to DFO alone; three trials included a three-way comparison of deferiprone, DFO and deferiprone with DFO; and one trial compared different doses of deferiprone.
There was little opportunity for meta-analysis for the majority of outcomes; analysis of data was principally a commentary on the findings from each included trial. Few trials measured long-term outcomes, mortality was only measured in four trials and evidence of reduced end-organ damage was measured in six trials. Although measures of iron overload were reported in all but one trial, meta-analysis was not possible for most outcomes due to different methods of assessment and measurement of outcomes used, variation in time points and a lack of sufficient data reporting. Trials aimed at evaluating iron chelators may benefit from the development and application of agreed standardised sets of outcomes to be collected and reported, ideally through the Core Outcome Measures in Effectiveness Trials (COMET) initiative (www.comet-initiative.org), which will better enable results of trials to be compared, contrasted and combined as appropriate (see also 'Implications for research' below).
Mortality was reported by four trials; each trial reported the death of one individual receiving deferiprone (with or without DFO). Most of these deaths were due to cardiac complications and could not be attributed to therapy. However, one patient receiving deferiprone presented with leucopenia with neutropenia, developed a severe respiratory tract infection after 11 months and died as a result of this infection (Choudhry 2004). One trial reported five further deaths in patients who withdrew from randomised treatment (deferiprone with or without DFO) and switched to DFO alone (Maggio 2009).
Reporting of long-term outcomes was limited and inconclusive; few trials reported evidence of reduced end organ damage as an outcome. The effect of deferiprone and DFO has attracted some interest since reports that cardiac iron load may be reduced by deferiprone (Pennell 2006). The earlier trials measuring cardiac iron load indirectly by measurement of the MRI T2* signal had suggested deferiprone may reduce cardiac iron more quickly than DFO. In a direct comparison of patients with low T2* (high cardiac iron) receiving deferiprone or DFO, T2* values increased in both treatment arms; the increase was two-fold higher at 12 months in patients who received deferiprone (26.9% increase from baseline) compared with those who received DFO (12.8%) (Pennell 2006). In a trial of DFO compared with DFO and deferiprone, increases in myocardial T2* (representing a reduction in cardiac iron) was significantly in favour of the combined treatment group, with an estimated 10% (95% CI 2% to 19%) increase in the combined group compared with the DFO group (Tanner 2007).
In the comparison of deferiprone versus DFO monotherapy, meta-analysis of three trials which reported LVEF as a measure of cardiac function showed no significant differences between treatment arms (Maggio 2002; Olivieri 1997; Pennell 2006). However, in a meta-analysis of two trials which compared LVEF between patients who received combined deferiprone with DFO and those who received DFO alone, LVEF was shown to be significantly reduced in patients who received DFO alone compared with combination therapy (Abdelrazik 2007; Tanner 2007).
There have been reports of deaths due to heart disease in trial participants. A recent trial did report a sudden death, probably due to a cardiac arrhythmia, although the death was not attributed to deferiprone (Ha (ii) 2006). Two further cardiac-related deaths were reported in patients receiving deferiprone (with or without DFO) (Aydinok 2007; Maggio 2009). There are many reports of the incidence of cardiac disease in observational studies, but it is impossible to compare the rates of cardiac dysfunction or disease in such studies.
The significance of differences in improvements in LVEF and MRI T2* and how they translate into the ability to prevent cardiac disease is unclear in the absence of more data and longer follow up of randomised participants (Neufeld 2006). It is clear from retrospective and prospective clinical studies that intensified DFO treatment, by either subcutaneous or intravenous route, can reverse cardiac dysfunction because of iron overload and increase survival in thalassaemia-major patients with early or overt cardiomyopathy (Maggio 2007). Any benefit of deferiprone to prevent heart disease needs to be evaluated in appropriately-powered trials with robust measures of cardiac iron load and function.
In the absence of long-term follow up in all but one trial (Maggio 2009) and with limited data on mortality or end organ damage, data on the effects of chelation therapy rely on markers of iron overload or measurements of iron loading or excretion.
In the direct comparison of deferiprone and DFO, after six months, a statistically significant difference in mean change in serum ferritin concentration in favour of DFO was observed in two trials (Gomber 2004; Pennell 2006); a third trial found no significant difference between treatment arms (Ha (ii) 2006). However, none of the trials which reported longer-term follow up showed a significant difference in mean change in serum ferritin concentration between treatment arms, at 12 months (Maggio 2002; Pennell 2006) or 24 months (Olivieri 1997).
In a single large trial with four years of follow up, there was a significant difference in mean change of serum ferritin concentration in patients receiving deferiprone in combination with DFO compared to deferiprone alone (Maggio 2009). Furthermore, in three trials comparing DFO to DFO and deferiprone, meta-analysis showed a significant difference between treatment arms, favouring the combined therapy (Abdelrazik 2007; Mourad 2003; Tamaddoni 2010).
These results suggest an advantage of combined therapy with DFO and deferiprone over monotherapy to reduce iron stores as measured by serum ferritin. Can such advantages for combination therapy be supported by other measurements of iron fluxes or stores, or both? Measures of urinary iron excretion are difficult to interpret as comparative measures of efficacy since these estimates do not include biliary iron excretion, and so would underestimate the total iron excretion for DFO. Hence no real conclusion can be made examining measurements of urinary iron excretion in five trials comparing deferiprone with DFO (Aydinok 2005; El-Beshlawy 2008; Gomber 2004; Maggio 2002; Olivieri 1990).
In three trials which compared deferiprone alone with combination therapy, no significant differences were observed in serum ferritin concentration between treatment groups either at the end of the trial or as mean change from baseline, with the exception of one trial with a planned follow up of five years (Maggio 2009) In this trial, a significant difference in mean change from baseline was observed in favour of patients receiving deferiprone combined with DFO over deferiprone monotherapy; this significant difference was maintained over four years of study. Of note, this large trial was terminated early due to the beneficial effects of combined treatment in terms of serum ferritin reduction compared with deferiprone alone.
In the comparison of combination therapy with DFO alone, all three trials which reported serum ferritin concentration after six months reported significantly lower serum ferritin values in patients who received DFO combined with deferiprone (Gomber 2004; Mourad 2003; Tamaddoni 2010). Furthermore, after 12 months, meta-analysis of three trials showed a significant difference between treatment arms in favour of deferiprone combined with DFO (Abdelrazik 2007; Mourad 2003; Tamaddoni 2010). Two trials which reported a significant difference between treatment arms in mean change in serum ferritin concentration from baseline favoured different treatment regimens.
Urinary iron excretion was measured in six trials comparing deferiprone with DFO (Aydinok 2005; El-Beshlawy 2008; Gomber 2004; Maggio 2002; Olivieri 1990; Olivieri 1997). However, in all but one of these trials, no baseline data were presented for this outcome and therefore mean change from baseline to end of trial could not be calculated (Maggio 2002). As already stated, these estimates of urinary iron excretion do not include biliary iron excretion, and so would underestimate the total iron excretion for DFO.
Of three trials which compared deferiprone with combination therapy, only one showed a significant difference in urinary iron concentration at the end of the trial in favour of deferiprone (Aydinok 2005). However, in the comparison of combined therapy with DFO alone, two trials showed a significant difference in urinary iron concentration at the end of the trial in favour of deferiprone combined with DFO (Abdelrazik 2007; El-Beshlawy 2008). In addition, one trial which reported mean percentage urinary iron excretion over the trial period showed a significantly greater level of urinary iron excretion in patients who received combined therapy compared with DFO alone (Aydinok 2005). Only one other trial reported urinary iron excretion for this comparison and this trial found no significant difference between treatment arms (Gomber 2004). Data from the majority of trials did not allow analysis of mean change from baseline between treatment arms. Nevertheless, the data suggest addition of deferiprone to DFO increases urinary iron excretion over DFO alone.
One trial used three other measures of iron burden and chelation, namely total iron excretion, chelation efficiency and plasma NTBI (Aydinok 2005). There was a small statistically significant difference in total iron excretion in favour of deferiprone compared with deferiprone and DFO. The results of other comparisons were not significant.
Liver iron concentration is a valuable measure of stored iron. Indeed, direct measurement of liver iron is the gold standard by which other assays are validated. Five trials comparing deferiprone with DFO reported liver iron concentration (El-Beshlawy 2008; Ha (ii) 2006; Maggio 2002; Olivieri 1997; Pennell 2006). However, conflicting results were observed between trials. Liver iron concentration at the end of the trial was significantly lower in the deferiprone treatment group in one trial (El-Beshlawy 2008), significantly lower in the DFO group in a second trial (Olivieri 1997), and showed a non-significant difference in favour of DFO in a third trial (Maggio 2002). None of three trials reporting mean change from baseline found significant differences between treatment groups (Ha (ii) 2006; Olivieri 1997; Pennell 2006).
Neither trial which reported liver iron concentration for the comparison of deferiprone alone versus combination therapy found significant differences in liver iron concentration at the end of the trial (Aydinok 2007; El-Beshlawy 2008). Similarly, in the comparison of deferiprone with DFO versus DFO alone, neither trial reporting this outcome found a significant difference in mean change from baseline between treatment arms (Galanello 2006; Ha (i) 2006).
Pooling of liver iron concentration was generally was prevented by differences in techniques used for measurement of liver iron content between trials (SQUID, atomic spectrophotometry, liver T2*) as well as the variable presence of hepatitis C and the wide variation in treatment duration between and possibly within trials. In addition, the degree of reporting of results (displayed graphically or a lack of SDs or CIs) in a number of trials was inadequate to formally assess differences between treatment groups.
Only one trial compared different doses of deferiprone (Choudhry 2004). This trial only reported serum ferritin concentration as a measure of iron overload. No difference in serum ferritin concentration at the end of the trial was observed between patients who received 50 mg/kg/day and those who received 75 mg/kg/day. Evidence of reduced end organ damage was not reported as an outcome in this trial.
There is therefore no conclusive or consistent evidence for the improved efficacy of combined deferiprone and DFO therapy against monotherapy from direct or indirect measures of liver iron.
Given the apparently similar efficacy of deferiprone and DFO, the safety, compliance and cost of these drugs are of special interest. Indeed the safety of chelation therapy has been somewhat controversial. Treatment with DFO has been in clinical usage for some 20 years longer than deferiprone and adverse events as a result of DFO therapy have been observed, with epidemiological data suggesting that adverse events may be dose-related (Porter 1989a; Porter 2002). The longer-term safety profile is much less certain for deferiprone than for DFO, although in the late 1990s liver fibrosis appeared to be associated with deferiprone during a clinical trial (Olivieri 1998).
Adverse event data were reported in 14 trials, but differences in how adverse events were measured (number of events or events per person), the time points for measurement and the lack of standard reporting of the severity of reactions precluded pooling of data across the trials in many instances.
In a direct comparison of DFO against deferiprone, meta-analysis showed an 11.7-fold increased risk of gastrointestinal side effects associated with deferiprone (El-Beshlawy 2008; Maggio 2002), an 8.9-fold increased risk of raised liver transaminase levels (El-Beshlawy 2008; Maggio 2002) and a 2.6-fold increased risk of joint pain or arthralgia (El-Beshlawy 2008; Maggio 2002; Pennell 2006). Furthermore, the FDA review of the primary data from the trial conducted by Pennell concluded that "Regarding safety, adverse events related to elevation of serum alanine aminotransferase levels were reported in 38% of the deferiprone group and in 18% of the deferoxamine group. In the context of additional concerns, this observation signals the potential for deferiprone induced liver toxicity." (Pennell 2006).
Local adverse events specifically related to the administration of DFO are well described. Local reactions at infusion sites occurred in 17.8% (n = 45) (El-Beshlawy 2008), 27.5% (n = 40) (Tamaddoni 2010), 85.7% (n = 14) (Mourad 2003) and 38.7% (n = 31) (Pennell 2006) of patients; local abscesses at the site of infusion occurred in 3.3% (n = 30) (Galanello 2006) and 2.5% (n = 40) (Tamaddoni 2010) of patients; and systemic allergy occurred in 4.3% (n = 23) of patients (El-Beshlawy 2008).
Three trials reported data that enabled a comparison of the risk of experiencing any adverse event. In a comparison of DFO versus deferiprone, there was a statistically significant 2.2-fold increased risk of experiencing an adverse event in participants receiving deferiprone compared with those receiving DFO (Maggio 2002). Furthermore, two trials observed a greater proportion of adverse events in patients receiving deferiprone with DFO than those receiving DFO alone; meta-analysis showed a statistically significant three-fold increased risk of experiencing an adverse event in participants receiving deferiprone with DFO compared with those receiving DFO alone (Abdelrazik 2007; Galanello 2006). The danger of raised liver enzymes or agranulocytosis with deferiprone means that close monitoring of full blood counts and liver function is required and precludes its use where close monitoring is not available.
However, these conclusions must be viewed with some caution, given that RCTs are not designed to measure the adverse effects of an intervention and thus data from RCTs does not represent a formal comparison of adverse events caused by iron chelation therapy. A complete review of adverse events should be the subject of a separate, formal analysis incorporating data from non-RCTs and observational studies. Nevertheless, reporting of serious adverse events from post-marketing surveillance, has shown that use of deferiprone may be associated with fatal agranulocytosis and when used at more than 100 mg/kg/day a neurological syndrome of cerebellar and psychomotor retardation that has progressively regressed after deferiprone has been discontinued (Henter 2007; Swedish Orphan Drug 2007). It is now recommended that weekly full blood counts are monitored in people receiving deferiprone (Swedish Orphan Drug 2007).
A further major concern of DFO is compliance. It has been suggested that participants given an oral iron chelator may be more compliant with treatment than those given a continuous subcutaneous infusion of DFO. Eleven trials reported compliance, with this suggestion being supported in two trials (Maggio 2009; Olivieri 1997) although reporting of compliance in other trials was often purely descriptive. The observed differences may be medically important, especially as the difference in compliance associated with these two iron chelators may become greater outside the rigours and discipline of a formal RCT setting. Thus, objective assessment of safety and compliance with these chelating agents will require studies with long-term follow up in a variety of settings.
Several factors have contributed to the disappointing absence of meta-analyses of trial results. Trials included participants with substantial clinical and demographic diversity which was not always fully defined. In particular, there were differences in the measurement of outcomes across trials, notably the different techniques used to assess liver iron concentration between trials, the presence of hepatitis C in participants in one trial (Maggio 2002), the skewed nature of the data in two trials (Maggio 2002; Mourad 2003) and the undermining of the randomisation process (Gomber 2004). Some trials failed to report baseline data, making valid comparisons impossible between trials. Moreover, where reported, baseline values for serum ferritin and liver iron concentration differed substantially between trials, and in two trials between treatment arms (Gomber 2004; Pennell 2006). The impact of these differences on any overall results is difficult to evaluate due to the limited number of trials and the small sample sizes of the included trials. The skewed nature of the data in two trials gives rise to concerns of the likely skewing of the data in the remaining trials. Without obtaining the IPD from all trials, this issue could not be confirmed.
Variation in the time points used for outcome measurement was also a factor that precluded meta-analysis. While very short-term outcome measurement was the appropriate objective of the two iron balance trials and measures of iron overload can be assessed over a six or twelve month period, much longer follow up is needed to examine the effect of iron chelation on mortality and reduced end-organ damage or other toxicity.
The validity of the data is a third issue to note. With the exception of two trials (Maggio 2002; Maggio 2009), the sample sizes of the included trials were small with 40 or fewer per treatment arm. Only six trials presented information on sample sizes required to power the trial around a main outcome, whilst several trials failed to clearly state the primary endpoint for analysis. Data were presented in abstract form for one trial, presumably with a lack of peer review and limiting the amount of data that can be reported.
The risk of bias of the included trials was difficult to assess given the general absence of information on randomisation and blinding to treatment allocation. As such, the influence of the quality of the included trials was not explored with respect to the robustness of any results. With the interventions in these trials, blinding to treatment allocation would be impossible for clinician and participant in all but one trial in which different doses of deferiprone were compared. However, blinding of outcome assessors would be desirable as a means of minimising bias, particularly where subjective measurements such as functional status and histological results are concerned.
Although no quantitative assessment of publication bias was undertaken, attempts were made to minimise the likelihood of publication bias by the use of a comprehensive search strategy, the handsearching of relevant conference abstract books and contacting the manufacturers of deferiprone and other iron chelators.
This review has highlighted the difficulty of undertaking an analysis in such intervention comparisons, not least the limited ability to pool and meta-analyse crucial aspects of measures of iron stores and chelation across the trials. These difficulties seem to be inherent in the population being studied. Children with thalassaemia may enter trials at different points in their clinical course, after variable amounts of blood transfusion, with variable endogenous iron loading, with different prior treatment and with differing biochemical levels as a result of past treatment and their thalassaemia. It seems unlikely that any one particular methodology could overcome these problems in future trials. It would therefore be important that each RCT comparing iron chelators or different schedules, doses or methods of administration of iron chelators should provide extensive details of baseline measures, collect follow-up data at times appropriate to the outcomes of interest and be sufficiently powered to provide a significant comparison within the trial.
These difficulties highlight the importance not only in designing trials to optimise subsequent meta-analysis, but of ensuring IPD are made available to the scientific community. Perhaps deposition of the IPD for an RCT will ultimately become a requirement for publication in the same way as molecular data for experimental data is deposited in public databases now. Moreover, consensus on the appropriate methods and sampling to measure iron overload would greatly help data from future trials to be combined. Finally, accurate recording of compliance and adverse events for different schedules of iron chelation should be an essential feature of future trials.
There is clearly an absence of adequate RCTs for understanding the relative benefit of DFO and deferiprone in long-term studies. Nevertheless, there has been some considerable use of combined DFO and deferiprone therapy. There are now many reports of a decline in the incidence of cardiac disease and mortality in observational studies since the implementation of regular measurement of myocardial T2* to detect those patients at risk of developing cardiac disease and subsequent intensification of chelation therapy in those with MRI evidence of myocardial iron loading. It is unclear whether the observed declines in mortality have followed improved detection of those at risk of cardiac disease or specific chelation regimens.
The absence of high quality data from RCTs to support specific recommendations for the use of deferiprone is not just the conclusion of this review but also the considered view of the FDA in the United States of America. The FDA reviewed an application to license deferiprone in 2011, having previously rejected an application in 2009. At that time Apotex, the manufacturer of deferiprone, was advised that the FDA would require at least one new RCT as well as a full audit of the original trial led by Dr Olivieri in Toronto. However, neither further RCTs nor other requested audits were submitted by Apotex, and Apotex abandoned its application for full approval despite previous and continued claims of efficacy and safety. Downgraded approval of deferiprone under the lower 'accelerated' standards, as "last resort treatment of iron overload in thalassemia, myelodysplasia and sickle cell disease" was later provided. This decision arose from the lack of new RCT evidence and the failure to provide answers to the FDA's questions on efficacy and safety.
In summary, there were insufficient data available to fulfil the aim of this review; namely to determine the effectiveness of the iron chelating agent deferiprone in people with transfusion-dependent thalassaemia and to compare the efficacy and safety of deferiprone with other iron chelating agents. Deferiprone is indicated for the treatment of iron overload in patients with thalassaemia major when DFO is contraindicated or inadequate. Intensified DFO treatment by either subcutaneous or intravenous route and/or use of other oral iron chelators remains the established treatment to reverse cardiac dysfunction because of iron overload. Indeed, the FDA have approved deferiprone only as "last resort treatment of iron overload in thalassemia" and combination therapy with DFO and deferiprone has been used in spite of the lack of RCT data in specific clinical situations, particularly where there is evidence of myocardial iron deposition or where compliance with continuous DFO is a severe problem, or both. Factors are suggested that should be considered in further trials to improve the efficacy and applicability of deferiprone in people with iron overload.