Transfusional iron overload in children with sickle cell anemia on chronic transfusion therapy for secondary stroke prevention

Chronic transfusion reduces the risk of recurrent stroke in children with sickle cell anemia (SCA) but leads to iron loading. Management of transfusional iron overload in SCA has been reported as suboptimal [1], but studies characterizing monitoring and treatment practices for iron overload in children with SCA, particularly in recent years with the expansion of chelator options, are lacking. We investigated the degree of iron loading and treatment practices of 161 children with SCA receiving transfusions for a history of stroke who participated in the Stroke with Transfusions Changing to Hydroxyurea (SWiTCH) trial. Data obtained during screening, including past and entry liver iron concentration (LIC) measurements, ferritin values, and chelation were analyzed. The mean age at enrollment was 12.9 ± 4 years and the mean duration of transfusion was 7 ± 3.8 years. Baseline LIC (median 12.94 mg/g dw) and serum ferritin (median 3,164 ng/mL) were elevated. Chelation therapy was initiated after a mean of 2.6 years of transfusions. At study entry, 137 were receiving chelation, most of whom (90%) were receiving deferasirox. This study underscores the need for better monitoring of iron burden with timely treatment adjustments in chronically transfused children with SCA. Chronic transfusion therapy greatly reduces the risk of recurrent stroke in children with sickle cell anemia (SCA) [2]. A major concern with this therapy is the development of transfusional iron overload, which can cause endocrinopathies, liver inflammation and fibrosis, and cardiac toxicity [3-5], although some of these complications appear to occur less commonly in SCA compared with thalassemia patients [6, 7]. The use of modified transfusion programs that allow higher pretransfusion hemoglobin S levels [8] and automated exchange transfusions may limit iron loading in SCA [9, 10]. Chelation therapy also is an effective means of reducing iron burden, and the clinical availability of an oral chelator, deferasirox (DFX), since late 2005, may improve treatment adherence compared with the parenterally administered deferoxamine (DFO). Despite these effective means of controlling iron burden in children with SCA, many children receiving transfusions have elevated iron stores. The SWiTCH trial was a Phase 3 multicenter, randomized trial to compare transfusion therapy and iron chelation to hydroxyurea and phlebotomy for secondary stroke prevention and control of iron burden [11]. Children with SCA receiving red cell transfusions for a history of overt stroke with transfusional iron overload were screened to determine trial eligibility. In this report, we characterize the transfusional iron burden and treatment practices in this group of children. Patient characteristics of the 161 children screened for the study are shown in Table I. The mean serum ferritin level and LIC were elevated at study entry. Of the 151 children who underwent screening liver biopsy, 64 (42.4%) subjects had LIC ≥15 mg/g dw and 99 (65.6%) had serum ferritin levels ≥2,500 ng/mL, levels associated with a high risk of iron-related complications in other transfused populations [12-14]. Seventeen of 151 (11.3%) children who underwent screening liver biopsy had well controlled iron levels with LIC <5 mg/g dw so were ineligible for randomization. Their mean LIC was 2.85 mg/g dw (median, 2.781; range, 0.8–4.893) and mean ferritin was 1,102 ng/mL (median, 850; range, 219 – 3,315). No children had a history of iron-related heart disease but 15 of 79 (19.0%) with past liver biopsies had a history of hepatic fibrosis. Baseline ALT levels positively correlated with both ferritin (r = 0.27, P = 0.0007) and LIC (r = 0.26, P = 0.0012). The presence of endocrinopathies was not assessed in this study, but no children were receiving oral hypoglycemics, insulin, or thyroid hormone replacement at the time of study entry. Transfusion duration weakly correlated with both serum ferritin levels and LIC (Fig. 1); other factors such as method of transfusion and use of chelation affect this relationship. Serum ferritin and LIC also were positively correlated, although the correlation was not precise (Fig. 1), underscoring that other factors including inflammation affect serum ferritin levels. Prior reports have shown conflicting results regarding the correlation of ferritin and LIC in transfused SCA patients [15-17]. A recent report showed ferritin levels <1,500 ng/mL correlated with an LIC <10 mg/g dw, while serum ferritin levels above 3,000 ng/mL usually predicted LIC >10 mg/g dw [18]. This pattern generally held true in our patient population, although disparities occurred in a subset of children: 3 had ferritin <1,500 ng/mL with LIC >10 mg/g dw, while 14 had ferritin >3000 ng/mL with LIC <10 mg/g dw. Thus, serial assessment of the serum ferritin level is helpful and, together with monitoring of transfusional intake, may be utilized to monitor iron overload, but periodic LIC measurements are advisable, particularly if MRI is available. The correlation between the duration of transfusion and (A) baseline study serum ferritin level and (B) baseline liver iron concentration and the correlation between liver iron concentration obtained by liver biopsy and serum ferritin (C) are shown. r = Pearson's correlation coefficient. Please see the accompanying Supporting Information for a list of participating centers. Only 79 of 161 (49.1%) had undergone liver biopsy in the past. Transfusion duration was longer in those who underwent liver biopsy (8.2 vs. 5.7 years, P < 0.0001). This liver biopsy rate is strikingly similar to a rate of 46.5% in transfused pediatric patients with SCA in a prior report, which was significantly lower than the rate of 69.6% in transfused children with thalassemia [1]. The mean LIC at the time of past biopsy was 6.8 mg/g dw (n = 73, range, 1.15–44.5) and had risen to 15.4 mg/g dw at study entry (mean, 2.6 years). The mean study entry LIC of children without prior liver biopsy was similar at 15.5 mg/g dw. Magnetic resonance imaging (MRI) techniques for estimation of liver iron loading were not available at most sites at the time of SWiTCH study enrollment. The increasing availability of R2 and other MRI techniques to measure LIC should allow more timely and frequent assessment of LIC in this patient population. One hundred forty-one (87.6%) children had received chelation therapy in the past. The mean transfusion duration prior to chelation initiation was 2.6 years (median, 2.0; range, 0–12.6), which is substantially lower than the duration of 4.1 years for sickle cell patients and 5.5 years for thalassemia patients in a prior report [1]. Treatment guidelines generally recommend initiating chelation after 10–20 transfusions have been administered [19], or after one to two years of transfusions, similar to the time of initiation of chelation in our overall population. Twenty of 161 (12.4%) children had never received chelation therapy before study entry; these children were younger (9.6 ± 2.8 vs. 13.4 ± 3.9 years, P < 0.001), had a shorter transfusion duration (3.2 ± 1.9 vs. 7.5 ± 3.7 years, P < 0.001) and had a trend towards both lower serum ferritin (2,622 vs. 3,525 ng/mL, P = 0.084) and LIC (11.6 vs. 15.95 mg/g dw, P = 0.096) than children who received chelation. Most (105) subjects had received DFO in the past, but by study entry, only 14 of the 105 (13.3%) continued to take DFO (mean dose 33.9 mg/kg/day), including two children who were taking both DFO and DFX. Ninety-nine (94.3%) children switched from DFO to DFX monotherapy (mean dose 25 mg/kg/day) before study entry after a mean of 4.7 years (median, 4.25; range, < 1 month to 12.33 years) of DFO therapy. The predominant use of DFX is not unexpected given that it is an oral agent and easier to administer than parenteral DFO. There was no significant difference in either ferritin levels (P = 0.6913) or LIC (P = 0.7811) between the DFO and DFX groups at study entry. However, an inverse relationship between duration of DFX, but not DFO, treatment and both serum ferritin level (r = −0.21, P = 0.0018) and LIC (r = −0.21, P =0.0137) was seen. This beneficial effect on reducing iron load was evident despite a short overall duration of DFX use, and could be related to superior adherence to treatment with this drug (data not shown). Further follow-up is required to see if this effect will persist, or perhaps be augmented over time. The type of transfusion also had an effect on LIC: median LIC was 7.87 mg/g dw (range, 3.17–31.94, N = 18), 11.71 mg/g dw (0.8–61.3, N = 96), and 18.94 mg/g dw (range, 1.89–42.39, N = 37) among children receiving erythrocytapheresis, simple, and partial exchange transfusions, respectively, P = 0.0417; duration of transfusion therapy was similar between these groups (7.4, 6.6, and 7.7 years, respectively, P = 0.24). The finding of the lowest iron burden in children receiving erythrocytapheresis is not surprising, given that this method is associated with a reduction in net red cell loading [9, 10]. Overall treatment costs also are lowered if chelation therapy can be avoided [20], further supporting the early use of erythrocytapheresis for long-term transfusion therapy in SCA when feasible. The finding of the highest iron loading in the subjects receiving partial exchange transfusion is unexpected. It is possible that these subjects were switched to partial exchange in response to high iron burden, and their iron burden will improve with continued treatment, which could not be assessed in this study. In a prior report, with long-term manual partial exchange to maintain hemoglobin S levels below 35–50%, six of seven subjects maintained ferritin levels below 1,000 ng/mL and did not require chelation; costs also were lowered compared with automated erythrocytapheresis [21]. Thus, manual partial exchange may be beneficial, particularly where chelation is not readily available or the treatment costs are prohibitive. A number of limitations of this study exist. First, study inclusion criteria required evidence of iron loading and thus excluded patients with well-controlled iron burden. Nonetheless, we found that iron burden was much higher than required for study entry. Second, baseline data collection was retrospective and some subjects may have had prior monitoring that was not reported in the study. Finally, the mean duration of DFX use was short (1.58 years), limiting the ability to assess the impact of this chelator on iron burden. In summary, the majority of the children in this multicenter cohort had suboptimal control of iron burden at study entry. Although iron overload was required for study inclusion, the LIC criterion for entry was only 5 mg per g dw while most of the participants had LIC well above this value. Most children initiated chelation in a timely fashion; thus, the high iron burden might be explained by other factors such as poor adherence to therapy or inadequate chelator dosing, which deserves further study. Health care providers should be aware of the importance of monitoring iron burden with timely initiation of treatment and ongoing adjustments to transfusion methods and/or chelation regimens in response to these measurements. The primary aim of SWiTCH, a multicenter intervention trial, was to compare alternative therapy (hydroxyurea and phlebotomy) with standard therapy (transfusions and chelation) for the prevention of secondary stroke and reduction of transfusional iron overload in pediatric subjects with SCA and previous stroke. The SWiTCH study design is described in detail elsewhere [11]. Briefly, children ages 5 to < 19 years old, were eligible to be screened for participation in the randomized portion of the SWiTCH study if they had SCA (type SS, S-β0-thalassemia, or S-OArab), a history of overt stroke, had received chronic transfusion therapy (simple, partial exchange, or erythrocytapheresis) for at least 18 months with an average pretransfusion hemoglobin (Hb) S level of ≤45% during the prior six months, and had evidence of iron overload defined as a LIC of at least 5 mg/g dry weight (dw) liver or serum ferritin of at least 500 ng/mL. At study entry, medical history was taken including data on stroke and other medical complications such as heart and liver disease. Data on transfusion duration and method were collected. Transfusion methods were characterized as simple, partial exchange (manual removal of a determined amount of blood followed by packed red cell transfusion), and erythrocytapheresis (automated red cell exchange). Additional data collection included chelation history (chelator type, dosage and duration of use), and current and prior medications used. Baseline laboratory studies included complete blood count with white blood cell differential and reticulocyte count obtained locally and hemoglobin electrophoresis with quantification of Hb A, S and F levels, chemistry panel, and serum ferritin measured at a central laboratory (Georgia Health Sciences University, Augusta, GA). Liver biopsy was performed and analyzed for LIC by inductively coupled plasma-mass spectrometry at Mayo Medical Laboratories, Rochester, MN. Descriptive statistics (mean, standard deviation, median, range, and frequency) were performed for demographic variables, transfusion history, chelation history, and baseline laboratory values. LIC values were log-transformed due to a skewed distribution of values. Pearson product-moment correlations were implemented for continuous variable relationships and for significance testing. Independent sample t-tests were computed for comparisons of categorical predictors (chelation type, splenomegaly, transfusion type) with continuous outcomes (ferritin, LIC, and hemoglobin S levels). T-tests were also used to examine the relationship of baseline chelator type (DFO versus DFX) and both LIC and serum ferritin levels. All P-values should be considered descriptive. Additional Supporting Information may be found in the online version of this article. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article. Janet L. Kwiatkowski MD, MSCE kwiatkowski@email.chop.edu*, Alan R. Cohen*, Julian Garro , Ofelia Alvarez , Ramamorrthy Nagasubramanian§, Sharada Sarnaik¶, Alexis Thompson**, Gerald M. Woods , William Schultz , Nicole Mortier , Peter Lane§§, Brigitta Mueller , Nancy Yovetich , Russell E. Ware MD, PhD*, * Children's Hospital of Philadelphia and University of Pennsylvania School of Medicine, Philadelphia PA, Rho, Chapel Hill, North Carolina, University of Miami Miller School of Medicine, Miami, FL, § Nemours Children's Clinic, Orlando, FL, ¶ Childern's Hospital of Michigan and Wayne State University School of Medicine, Detroit, MI, ** Children's Memorial Hospital, Chicago, IL, Children's Mercy Hospital and University of Missouri-Kansas City School of Medicine, Kansas City, MO, Baylor College of Medicine, Houston, TX, §§ Emory University/Children's Healthcare of Atlanta, Atlanta, GA.