RESEARCH ARTICLE

A JH

Cardiac iron overload in sickle-cell disease Antonella Meloni,1,2 Mammen Puliyel,3 Alessia Pepe,1 Vasili Berdoukas,3 Thomas D. Coates,3 and John C. Wood2,4* Chronically transfused sickle cell disease (SCD) patients have lower risk of myocardial iron overload (MIO) than comparably transfused thalassemia major (TM) patients. However, cardioprotection is incomplete. We present the clinical characteristics of six patients who have prospectively developed MIO, to identify potential risk factors for cardiac iron accumulation. From 2002 to 2011, cardiac, hepatic, and pancreatic iron overload were assessed by R2 and R2* magnetic resonance imaging techniques in 201 chronic transfused SCD patients as part of their clinical care. At the time, they developed MIO, five of six patients had been on chronic transfusion for more than 11 years; only one was on exchange transfusion. The time to MIO was correlated with reticulocyte and hemoglobin S percentages. All patients had qualitatively poor chelation compliance (4600 ng/ml and liver iron concentration >22 mg/g. Pancreatic R2* was >100 Hz in every patient studied (5/6). Cardiac iron rose proportionally to pancreas R2*, with all patients having pancreas R2*>100 Hz when cardiac iron was present. MIO had a threshold relationship with liver iron that was higher than observed in TM patients. In conclusion, MIO occurs in a small percentage of chronically transfused SCD patients and is only associated with exceptionally poor control of total body iron stores. Duration of chronic transfusion is clearly important but other factors, such as levels of effective erythropoiesis, appear to contribute to cardiac risk. Pancreas R2* can serve as a valuable screening tool for cardiac iron in SCD patients. C 2014 Wiley Periodicals, Inc. Am. J. Hematol. 89:678–683, 2014. V

䊏 Introduction Regular red blood cell (RBC) transfusions have represented the cornerstone of treatment of patients with thalassemia major (TM) since the 1960s and much of our understanding of iron overload and its complications has come from studies in this population. In the last 15 years, however, transfusions have become an increasing integral part in the management of many sickle cell disease (SCD) patients to prevent neurovascular complications. In SCD, regular transfusions (every 3–4 weeks) improve oxygen carrying capacity and reduce the proportion of red cells capable of forming sickle hemoglobin (HbS) polymer [1]. Reducing post-transfusion HbS to below 30% is associated with a significantly decreased stroke risk in patients with abnormal cerebral blood flow velocities [2] or with prior strokes [3,4]. Moreover, transfusion is beneficial in other complications of SCD such as acute chest syndrome [5,6] and for the reduction of the frequency of pain episodes [7]. Although some SCD patients have increased iron excretion through renal elimination of decellularized hemoglobin [8], and others have lower iron intake because of the use of erythrocytaphoresis [9], long-term transfusion therapy nonetheless produces secondary state of iron overload in SCD patients. However, the onset and prevalence of iron-mediated complications appears to be delayed in SCD patients [10,11]. This could result from delayed iron uptake to sensitive organs or better buffering of organ iron stores. With increased availability of magnetic resonance imaging (MRI) based iron quantification methods, hepatic, cardiac, pancreatic and renal iron burden can be accurately and noninvasively assessed in SCD patients [12–17]. Using these techniques, our laboratory was the first to demonstrate that myocardial iron overload (MIO) was common among TM patients but absent in transfused SCD patients matched for age, sex and hepatic iron content [18]. The disparity in cardiac iron loading between the two patient populations was reconfirmed in many studies [19–21] and mirrors the observations of organ toxicity in these patients [11]. The common mechanism for this disparity is that transferrin saturation and circulating labile iron species are decreased in SCD patients [15,22,23]. Pathologic loading of the heart occurs primarily through circulating labile iron species [24]. There are at least three possible mechanisms for lower transferrin saturation in SCD. First, SCD patients are continually utilizing transferrin bound iron during red cell production, regenerating apotransferrin and lowering transferrin saturation. In fact, cardiac risk correlates inversely with endogenous marrow activity, with Blackfan-Diamond having the highest risk and SCD the lowest risk among chronically transfused anemias [23,25]. Second, SCD is a chronic

Additional Supporting Information may be found in the online version of this article. 1

CMR Unit, Fondazione G. Monasterio CNR-Regione Toscana, Pisa, Italy; 2Division of Cardiology, Children’s Hospital Los Angeles, Los Angeles, California; 3Division of Hematology-Oncology, Children’s Hospital Los Angeles, Los Angeles, California; 4Department of Radiology, Children’s Hospital Los Angeles, Los Angeles, California

Conflict of interest: AM: none; MP: none; AP: speaker honoraria from Chiesi Farmaceutici, ApoPharma, Inc. and Novartis; VB: consultancy and Advisory Board for ApoPharma, Inc., confidentiality agreement with Novartis for the development of ICL 670; TDC: research funding and consulting with Shire, Advisory Board with Shire and ApoPharma, Inc., speaker honoraria from Novartis.; JCW: research for Shire, Advisory Board for ApoPharma, Inc. and Shire, speaker honoraria from Novartis, ApoPharma, Inc. and Shire. *Correspondence to: John C. Wood, Division of Cardiology, MS#34, Children’s Hospital of Los Angeles, 4650 Sunset Blvd, Los Angeles, CA 90027-0034. E-mail: [email protected] Received for publication: 27 January 2014; Revised: 19 March 2014; Accepted: 21 March 2014 Am. J. Hematol. 89:678–683, 2014. Published online: 25 March 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/ajh.23721

C 2014 Wiley Periodicals, Inc. V

678

American Journal of Hematology, Vol. 89, No. 7, July 2014

doi:10.1002/ajh.23721

RESEARCH ARTICLE inflammatory disease [22,26]. Inflammation and cytokines (IL-10) stimulate the uptake and retention of iron into monocytes and reticuloendothelial cells, isolating it from the blood [27]. In addition, IL6 stimulates hepcidin release from the liver, independent of erythropoetic and iron-sensing regulation, further impairing iron release from the gut and reticuloendothelial [28]. Lastly, SCD patients have lower transfusional exposure because they start transfusions later in life, they have greater utilization of erythrocytaphoresis and they often have less intensive simple transfusion regimes [29]. However, cardioprotection in SCD is incomplete. Of the 201 pediatric and young adult chronically transfused patients followed by MRI at CHLA, we have identified six patients (3%) who have prospectively developed cardiac iron. In the present study, we present the clinical characteristics of these patients to identify potential risk factors for cardiac iron accumulation.

䊏 Methods Patient population. From July 2002 to June 2013, 201 regularly transfused SCD patients had undergone multiple MRIs for iron overload assessment at the Children’s Hospital of Los Angeles, providing a total of 606 available exams. We retrospectively reviewed all these exams and identified the patients who developed MIO during the monitoring period. This was defined as initially having a cardiac R2* < 50 Hz (T2* > 20 ms), followed by at least one exam with a R2* > 50 Hz. The protocol for this analysis was approved by the Committee for the Protection of Human Subjects. The medical records of the selected patients were reviewed for demographic data, for transfusion and chelation history and for hematologic and biochemical parameters, determined by routine laboratory methods. Hemoglobin S percentage and lactate dehydrogenase (LDH) were considered valid only if measured within 3 months from the MRI. Serum ferritin and reticulocyte count were calculated from values collected during transfusion visits over 1, 3, or 5 years at the time of the MRI in which cardiac iron was detected. Sickle cell patients on chronic transfusion therapy at CHLA are transfused 10– 15 ml/kg every 3 or 4 weeks, typically delivering between 150 and 235 cc/kg/year of packed red blood cells. Transfusions volume and intervals are adjusted to maintain a hemoglobin S percentage below 30%. If target hemoglobin S levels could not be achieved, manual or automated exchange was initiated. Given the retrospective nature of this study, the long observation interval, and the incomplete documentation of transfusional volumes prior to initiation of the electronic record, it was not possible to accurately estimate transfusional burdens in these six patients at the time they developed cardiac iron. To place our practices in clinical context, a recently published study from our chronically transfused patient cohort indicated a median transfusion burden of 169 ml/kg/year for SCD (N 5 35) and 199 ml/kg/ year for TM (N 5 35) [29]. MRI. MRI exams were performed on two different 1.5 Tesla scanners: a Philips Achieva (Philips Medical Systems, Best, The Netherlands) running system 2.5.1 and a GE Signa CVi (GE Healthcare, Waukesha, WI) running system 9.1. Similar phased array torso coils were used on both scanners. Multiecho gradient-echo sequences were used to collect the images for R2* analysis. Myocardial R2* was assessed in a single midpapillary short axis slice; black blood images were acquired on the Philips scanner (16 echo times, TEs, starting from 1.28 ms with an echo spacing of 1.09 ms) while bright blood images were acquired on the GE scanner (8 echo times starting from 1.40 ms with an echo spacing of 2.38 ms). A mid hepatic slice was acquired at 15 echo times logarithmically spaced between 0.95 and 16 ms in the GE scanner and at 16 echoes evenly spaced from 1.0 to 13.4 ms on the Philips scanner. Liver R2 was also collected using single spin echo sequences having a TR of 300 ms, slice thickness 10 mm, and TEs of 3, 3.5, 5, 8, 12, 18, and 30 ms (GE) or 3.9, 4.1, 4.5, 5, 8, 12, 18, and 30 ms (Philips). For pancreatic R2* assessment multiple slices with zero gap were acquired from the diaphragmatic angle to the renal-collecting systems (8 echoes equally spaced between 1.1 and 15.1 ms on the GE scanner and 16 echoes equally spaced between 1.1 and 17.9 ms on the Philips scanner). Kidney R2* was measured in five coronal slices (8 echoes equally spaced from 1.2 to 13.9 ms on the GE scanner and 16 echoes equally spaced from 1.0 to 16 ms on the Philips scanner). All R2* images were processed using custom MATLAB routines (The Mathworks, Natick, MA). Region of interest (ROI) in the heart was restricted to the interventricular septum [30] while in the other organs it encompassed the entire visible tissue [14–16]. The signal in each pixel within a ROI was fit to an exponential plus a constant model and the median of the distribution of R2* values was calculated. A cardiac R2* measurement  50 Hz was considered indicative of iron overload [12]. As recommended [31], liver R2 values [32] and R2* values [16] were converted to predicted liver iron concentration (LIC) using appropriate calibration curve. A

doi:10.1002/ajh.23721

Cardiac iron in SCD LIC  3 mg/g dry weight was considered indicative of a significant load [33]. For the pancreas R2* the cutoff of normality, derived from normal controls as 95% confidence interval, was 28.1 Hz and in TM patients a pancreatic R2* > 100 Hz was a powerful negative predictor of cardiac iron [14]. A renal R2* value < 34 Hz was considered normal [15]. For the evaluation of left ventricular (LV) ejection fraction (EF), 15 serial shortaxis steady-state free precession cine images were acquired during 8-sec breath holds with slice thickness adjusted to span the heart. Twenty to 30 cardiac phases were acquired per heart-beat. Images analysis was performed in a standard way, as previously described [34].

䊏 Results Five patients (2.5%) developed MIO (cardiac R2*>550 Hz). We included in our study group also a sixth patient because he showed a R2* of 49 Hz that was increasing rapidly on serial examinations. Table I describes the clinical characteristics of the six patients (four females and two males) at the time they developed MIO. One patient was 17 years old but all the others were older than 18 years of age. In four patients, the spleen was not detectable on the MRI images. Five of the six patients were managed on simple transfusions. Five patients had been on chronic transfusion for more than 11 years; this corresponds to the “threshold” duration observed in TM patients during previous studies [18,35]. The patient who had received transfusions for only 4 years had received many intermittent transfusions for recurrent acute chest syndrome and had a LIC of 15 mg/g dry weight prior to initiating chronic transfusion therapy. In five of six patients, the hemoglobin S was maintained at