Multicenter Validation of Spin-Density Projection-Assisted R2-MRI for the Non-Invasive Measurement of Liver Iron Concentration

Iron overload is a consequence of several clinical disorders, such as thalassemia, sickle cell disease, hereditary hemochromatosis (HH), aplastic anemia, and myelodysplasia. Some of the major determinants of clinical outcome in iron-overloaded patients include the total amounts and concentration of iron in different tissues, irrespective of whether the iron overload results from transfusion therapy, increased dietary iron absorption, or both. Chelation therapy or phlebotomy is necessary in order to avoid the serious clinical consequences of uncontrolled iron overload, which include liver cirrhosis and cardiac and endocrine dysfunction (1). Accurate assessment of body iron burden is therefore an important element of management for patients at risk of iron overload. For patients receiving iron chelation therapy, regular monitoring of body iron burden is essential to prevent iron toxicity, while avoiding potential adverse effects of excess chelator administration. Accurate determination of body iron stores can also be used in patients with HH to identify patients at risk of iron-induced organ damage who would benefit from phlebotomy therapy (2). The reference standard for evaluating the magnitude of body iron overload in systemic iron overload is measurement of the liver iron concentration (LIC) (3,4). Until recently, the conventional clinical method for measuring LIC was through chemical analysis of biopsy specimens and this has been considered the “gold standard” of LIC measurement (5). However, the invasive nature and risks associated with multiple liver biopsies preclude serial observations. Measurement of LIC and detection of fibrosis or cirrhosis in biopsy specimens are also subject to sampling variability, primarily as a result of the natural spatial variation of LIC and fibrosis throughout the liver relative to the small size of the biopsy (6–8). Furthermore, the variation in LIC throughout the liver increases as iron loading increases and with the development of cirrhosis (6). To counter the shortcomings of liver biopsy, several noninvasive techniques have been applied to estimate LIC, including biomagnetic susceptometry based on superconducting quantum interference device (SQUID) technology and magnetic resonance imaging (MRI) techniques. However, biomagnetic susceptometry is currently only accessible at a limited number of centers (2,9–12). Since the 1990s, several MRI-based techniques for assessing LIC have been proposed (13). These techniques are based either on the measurement of the ratios of signal intensities from the liver and from a non-iron-loaded reference tissue (14) or the measurement and imaging of proton transverse relaxation rates (R2 or R2*) within the liver (15–17). Radiofrequency pulse recalled echo imaging produces images for calculating T2 and R2 MRI parameters, and gradient echo imaging produces images for calculating T2* and R2* MRI parameters (18). However, one of the major limitations of these techniques is that they have typically been calibrated against LIC measurement by biopsy or other reference standard using a single scanner only (14,17,19). Furthermore, methods of image data acquisition and analysis are often not standardized. With regard to liver T2* measurement, evidence indicates that calibrations are not transferable from one method to another. For example, a liver T2* of 2.5 ms measured on a 1.5-T scanner gives vastly different LIC results depending on the calibration used, as follows: 5.4 mg Fe/g dry weight (dw) (20), 10.4 mg Fe/g dw (17), 10.7 mg Fe/g dw (21), 12.7 mg Fe/g dw (22), and 26.4 mg Fe/g dw (23). These wide-ranging LIC values for the same value of liver T2* cover the range from low-risk ( 15 mg Fe/g dw) categories (24), making interpretation of the absolute value of the T2* measurement problematic unless the specific method and scanner have been calibrated. Although monitoring the relative values of uncalibrated liver, T2* in serial measurements may be useful in clinical practice to facilitate dose titrations of iron chelators in response to ongoing iron burden, the high cost of using MRI scanner time for such qualitative monitoring needs to be assessed in relation to the benefits gained above and beyond those obtained from serial serum ferritin measurements. Other factors that can potentially confound calibrations of MRI methods of liver iron measurement are the presence of liver fibrosis and inflammation because of the associated change in water diffusion within the tissue (13). We previously reported on calibration of the spin-density projection-assisted (SDPA) R2-MRI (FerriScan®) technique, using data from five different scanners on 105 patients with a range of iron-loading disorders (16). None of the subjects contributing to the calibration curve had been chelated with deferasirox (EXJADE®; Novartis Pharma AG, Basel, Switzerland), an oral iron chelator that has been approved for clinical use by the Food and Drug Administration since the original calibration study. We now report on the validity of the SDPA R2-MRI calibration using five more different scanners by comparing biopsy iron assays from a larger group of regularly transfused, heavily iron-overloaded patients with β-thalassemia treated with the iron chelator, deferasirox with the SDPA R2-MRI LIC measurements, allowing any impact of this chelation therapy, scanner type, age of patient, stage of liver fibrosis, and grade of necroinflammation on the calibration to be determined. Specifically, the study is aimed at determining (i) whether there is any significant bias between the SDPA R2-MRI and biopsy methods of measuring LIC and (ii) what are the limits of agreement between the two methods in order to determine whether SDPA R2-MRI can be used in place of biopsy for the purpose of measuring LIC in patients with thalassemia being treated with deferasirox.