MRI evaluation of tissue iron burden in patients with β-thalassaemia major
- 1k Downloads
β-Thalassaemia major is a hereditary haemolytic anaemia that is treated with multiple blood transfusions. A major complication of this treatment is iron overload, which leads to cell death and organ dysfunction. Chelation therapy, used for iron elimination, requires effective monitoring of the body burden of iron, for which serum ferritin levels and liver iron content measured in liver biopsies are used as markers, but are not reliable. MRI based on iron-induced T2 relaxation enhancement can be used for the evaluation of tissue siderosis. Various MR protocols using signal intensity ratio and mainstream relaxometry methods have been used, sometimes with discrepant results. Relaxometry methods using multiple echoes achieve better sampling of the time domain in which relaxation mechanisms take place and lead to more precise results. In several studies the MRI parameters of liver siderosis have failed to correlate with those of other affected organs, underlining the necessity for MRI iron evaluation in individual organs. Most studies have included children in the evaluated population, but MRI data on very young children are lacking. Wider application of relaxometry methods is indicated, with the establishment of universally accepted MRI protocols, and further studies, including young children, are needed.
Keywordsβ-Thalassaemia MRI Iron Relaxometry
β-Thalassaemia major is a hereditary anaemia characterized by ineffective erythropoiesis and haemolysis . The underlying mechanism is defective production of haemoglobin β-chains, resulting in excess of α-chains, which are unstable and precipitate to form intracellular inclusion bodies [2, 3]. This excessive intracellular deposition of α-chain material is responsible for accelerated apoptosis of the erythroid precursors and for peripheral haemolysis of the erythrocytes . By the age of 3 months, severe anaemia develops leading to increased intestinal iron absorption. To maintain haemoglobin at a level of 10–12 g/dl, patients suffering from β-thalassaemia major need to be given repeated blood transfusions . A major drawback of this treatment is transfusion siderosis, which, in association with the increased intestinal iron absorption, apoptosis of the erythroid precursors and peripheral haemolysis, leads to iron overload .
Iron is ubiquitous in eukaryotic organisms. It is essential for cellular survival and proliferation and for haemoglobin synthesis [4, 5]. Human demands for iron are covered partly by intestinal absorption, but mainly from the recycling of iron from old or abnormal erythrocytes phagocytized by macrophages . Iron released from macrophages binds mainly to transferrin, but also to citrate and albumin to form non-transferrin-bound iron (NTBI) which is a toxic form [7, 8, 9]. Extracellular iron bound to transferrin enters the cell via transferrin receptors by a process of endocytosis. Once iron is released into the cytoplasm, it enters a poorly defined cellular compartment termed the labile iron pool (LIP). Iron in this compartment is loosely bound and therefore is highly toxic . Intracellular iron that is not needed for immediate use is stored in the form of ferritin which consists of an apoprotein storing up to 4,500 atoms of iron (loading factor 4,500) [5, 11], and is not cytotoxic. When ferritin storage capacity is exceeded the LIP increases and haemosiderin is generated from ferritin denaturation . Iron in the form of haemosiderin is thought to be more cytotoxic . The iron in the LIP is in both the ferrous (Fe2+) and ferric (Fe3+) forms . Fe2+ reacts with hydrogen and lipid peroxides and generates highly toxic hydroxyl and lipid radicals (Fenton reaction) that damage cellular membranes, proteins and nucleic acids . Iron overload, just as lack of cellular iron, may lead to cell death and organ dysfunction.
Iron overload is a major cause of morbidity and mortality in β-thalassaemia major . Iron accumulates initially in the reticuloendothelial system (bone marrow, spleen, liver) and then in the hepatocytes, the heart (myocytes) and the endocrine glands [1, 9, 13]. In contrast to the reticuloendothelial cells, the turnover of iron in the hepatocytes, myocytes and endocrine glands is very low . Chelation therapy has been used to eliminate excess iron [1, 7]. Chelatable iron is derived from the catabolism of haemoglobin in macrophages . Chelators remove NTBI from the plasma, but they do not interact with ferritin and haemosiderin at clinically relevant rates [10, 15]. Desferrioxamine (DFO), which is the chelating agent most widely used over the last 30 years, requires parenteral administration. It removes mainly extracellular iron and only a fraction of the intracellular LIP iron [14, 15]. Deferiprone is an oral iron chelator that penetrates the cellular membrane and chelates intracellular species of toxic iron . Cellular uptake of chelators takes place at different rates in different cells . DFO is of low toxicity, mainly affecting the optic, auditory and skeletal systems . Skeletal changes develop mainly due to the toxic effects of DFO on the growth cartilage and are manifested as disproportionate truncal shortening with loss of seated height . Adverse effects of deferiprone are arthralgia, gastrointestinal symptoms, elevated liver enzymes and rarely agranulocytosis [7, 16].
The effective management of patients, and especially of children, with thalassaemia requires optimal monitoring of the toxic effects of both iron overload and excessive chelation therapy. Serum ferritin has been widely used as a surrogate marker and a target ferritin level of 1,000 μg/l is generally recommended [7, 17]. However, serum ferritin represents only 1% of the total iron pool, and as an acute-phase protein it is not specific because the levels can be raised in inflammation (e.g. hepatitis) and liver damage [7, 18, 19]. Chronic liver inflammation is not rare in patients with thalassaemia, since over 40% of them have positive anti-hepatitis C virus (HCV) antibodies and more than 50% have chronic (persistent or active) hepatitis [20, 21]. Liver iron concentration (LIC) measured on needle biopsy is currently considered the gold standard for the evaluation of siderosis [7, 18, 19]. However, needle biopsy is an invasive technique, it is not easily repeatable and the accuracy of the resulting LIC measurement is greatly affected by hepatic inflammation-fibrosis and uneven iron distribution . Furthermore, it appears that cardiac iron overload, which is the leading cause of death in thalassaemia, cannot be predicted from the degree of liver siderosis, probably because of differences in iron kinetics between liver and heart cells . The use of non-invasive techniques for monitoring iron overload in each of the affected organs would be preferable, and to this end MRI has been used increasingly over the last two decades [19, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44]. Other non-invasive methodologies are too complex, too costly and not readily available (i.e. magnetic susceptometry), or lack imaging capability (i.e. magnetic resonance spectroscopy) [45, 46].
MRI assessment of tissue iron
MRI evaluation of tissue iron overload is based on T2 relaxation enhancement induced from the interactions between high-molecular-weight iron complexes such as ferritin, haemosiderin and ferrioxamine with water molecules [5, 47, 48]. Ferrioxamine is produced after cellular uptake of DFO, and in extrahepatic tissues, where there is no active excretion and it may remain for days . Ferrioxamine-based agents have been used in the past as contrast media because of their T2* effect . “Inner sphere” and “outer sphere” relaxation theories have been proposed to explain the T2 relaxation enhancement induced in siderotic tissues [50, 51, 52, 53, 54, 55]. According to the inner sphere theory, iron electron spins enhance the relaxation of protons of water molecules buried in iron-containing proteins. Enhanced relaxation of the bound water protons is then transferred to the free water protons by water or proton exchange [48, 56, 57]. According to the outer sphere theory, magnetic field gradients induced at the periphery of iron-containing proteins create loss of phase and relaxation enhancement of free water protons diffusing in the neighbourhood [53, 54].
MR methods for assessing tissue iron can be separated into two groups: signal intensity ratio (SIR) methods and relaxometry methods. Various techniques have been described, including: (a) methods measuring SIR based on T2-weighted (spin-echo) or T2*-weighted (gradient-echo) sequences [19, 23, 24, 25, 26, 30, 58, 59, 60, 61, 62], (b) relaxometry methods measuring absolute T2, (c) relaxometry methods measuring absolute T2*, and (d) hybrid relaxometry methods [30, 31, 32, 34, 35, 36, 37, 39, 40, 41, 42, 43, 58, 63].
These have been used for the study not only of the liver but also of other organs such as the spleen, pancreas, pituitary gland, bone marrow and abdominal lymph nodes [19, 23, 25, 27, 30, 58, 59, 60, 61, 62]. For SIR assessment the signal intensity of the target organ is divided by the signal intensity of a reference tissue (e.g. fat, muscle) or noise. Signal intensity measurements are performed in the same slice by using the region-of-interest (ROI) method. For large organs such the liver, spleen and pancreas more than one ROI is used, positioned in areas lacking vascular structures and movement artefacts [19, 23, 25, 27, 30, 58, 61, 62]. The mean signal intensity from the different ROIs is then divided by the signal intensity of the reference tissue. One ROI is used for signal intensity measurement of the pituitary gland, bone marrow, abdominal lymph nodes and the reference tissue [58, 59, 60]. For SIR evaluation of the liver various reference tissues have been used, of which paraspinal muscle appears to be the best choice because it combines good sensitivity with minimum intersite variability [19, 64]. A disadvantage of the SIR methods is that in most cases they use only one echo time (TE) and thus lose their detection sensitivity in tissues with heavy siderosis, where transverse relaxation is much faster than the TE. This occurs particularly in the liver at the upper range of LIC values where signal intensities are widely dispersed . Gandon et al. , by using an algorithm that combined signal intensity ratios from multiple sequences with different TEs, achieved extension of the detection range up to about 21 mg Fe per gram dry liver tissue, with sensitivity and specificity similar to those of biochemical analysis.
T2 relaxometry methods
These assess T2 relaxation time or R2 (1/T2) by using the Carr-Purcell-Meiboom-Gill (CPMG) spin-echo sequence, which employs multiple (2–32) equidistant refocusing 180° pulses, each followed by an echo [65, 66]. Most scanners, by using a pixel-by-pixel, log-linear fitting model, automatically derive the corresponding T2 maps. Signal intensity measurements in the T2 maps correspond to the mean T2 relaxation time of the included voxels . To avoid motion artefacts from respiration, studies of abdominal organs should be performed with respiratory triggering. To this end a pressure-sensitive pad is attached to the abdomen and depending on the patient’s respiratory cycle a delay time is chosen. Data acquisition takes place at the end-expiration phase [39, 40]. During cardiac MRI, in addition to respiratory triggering, cardiac triggering is necessary. To obtain a minimum repetition time (TR) of 2,000 ms two to three R–R intervals should be allowed between successive excitations [39, 40]. The mid-systole phase and a cardiac trigger delay of around 250–300 ms are usually chosen to image the myocardium with sufficient thickness for the best signal intensity measurements [39, 40]. A pseudo four-chamber or short-axis view of the heart is generally used. A T2 relaxometry method that received FDA approval for clinical liver iron estimation has recently been developed by St. Pierre et al. . This method uses multiple T2-weighted single spin-echo sequences with different TEs acquired in half-Fourier mode to reduce acquisition time. The calculated mean R2 values combined with LIC values, obtained from liver biopsies, are used to create calibration curves.
T2* relaxometry methods
These evaluate T2* or R2* (1/T2*) by using multiple gradient-echo sequences with different TEs. These methods have been developed to further accelerate acquisition, in order to increase sensitivity and eliminate artefacts related to respiration or cardiac motion. To obtain R2* (1/T2*) values, the signal decay curve is usually fitted with an exponential model: S=S0e−TE/T2*, where S is the net image signal intensity, TE is the echo time and S0 is a constant . T2* relaxometry methods have been used mainly for myocardial iron assessment and cardiac gating is always applied [35, 42, 68, 69, 70, 71]. Breath-hold sequences have greatly eliminated motion artefacts [35, 42, 68].
Hybrid relaxometry methods
Hybrid approaches have been applied in high fields and measure both R2 and R2* to calculate the inhomogeneity factor R2′=R2*−R2 [34, 72, 73]. These approaches assume that R2′ is more specific to mechanisms of relaxation related to iron than R2 [34, 73].
Comparison of the MRI methodologies
SIR versus relaxometry
T2, T2* and hybrid methods
There is no general consensus on which relaxometry method (spin-echo or gradient-echo) or index (R2, R2*, R2′) is best for tissue iron quantification. Theoretically, R2* and R2′ are more sensitive than R2 to iron-induced field inhomogeneities. Gradient-echo sequences evaluating R2* are preferred for cardiac relaxometry, because of their short acquisition time, but gradient-echo sequences are more prone to artefacts and R2* is more dependent on factors unrelated to iron, such as susceptibility artefacts from the lungs and the blood oxygenation level-dependent (BOLD) effect [35, 74]. According to the BOLD effect, the T2* decreases as the concentration of intracapillary deoxyhaemoglobin increases . A pronounced BOLD effect may be observed in early diastole and results in signal loss [75, 76]. Data collection during mid-systole helps to overcome BOLD effect interference with accurate T2* measurements of the myocardium. In liver and pituitary studies both T2 and T2* have been assessed. T2* is more sensitive to low iron content, but studies in the liver show that these methods suffer from inaccuracies at high iron concentrations . Furthermore, in high fields susceptibility artefacts from the sphenoid sinus are more pronounced with gradient-echo sequences and may lead to inaccurate measurements of pituitary siderosis .
MRI studies of individual iron overloaded organs
The degree of siderosis, the crystalline structure of ferritin, the rate of iron elimination under chelation therapy and the degree of ferrioxamine formation are all organ-specific [7, 15, 18, 78]. All these parameters may be responsible for differences in the T2 relaxation enhancement induced in the various organs. Individual organs should be considered separately, and the effect of age on iron overload should be taken into account. Higher survival probabilities have been reported in patients with thalassaemia born in the last 30 years . Patient compliance with treatment regimens and effective chelation therapy are thought to be the main factors associated with improved survival . The combination of DFO, deferiprone and the new oral chelators is considered very promising, but will require effective monitoring by non-invasive methods . An increasing number of studies have evaluated iron in the various affected organs by MRI [19, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44]. Many of these studies have included paediatric patients, but very few have investigated children younger than 10 years and none has included children younger than 5 years [17, 18, 28, 39, 43, 58, 60, 61, 62, 77, 81, 82, 83, 84]. In addition, very few studies have evaluated the effect of age on individual organ siderosis and only one has evaluated and compared MRI data from younger and older patients with thalassaemia [28, 81, 82, 84]. Studies have tended to focus on specific organs.
In iron overload states, over 70% of body iron is found in the liver and LIC has been considered to be the best marker of total body iron burden [7, 18, 19]. Based on the good correlation between hepatic R2 or SIR and LIC a number of recent studies have tested the relationship between siderosis of the liver and other organs [28, 39, 40, 58, 59, 60, 61, 87]. No correlation has been found between liver and pituitary siderosis [28, 58]. With regard to the heart, a correlation with liver siderosis has been found only in cases of heavy myocardial iron deposition [39, 40]. This lack of correlation can probably be explained by differences in transferrin receptor concentration, iron kinetics, the crystalline structure of ferritin and the degree of organ inflammation or fibrosis [7, 78, 88, 89, 90, 91, 92]. Furthermore, under chelation therapy with DFO intracellular paramagnetic ferrioxamine is formed, which exits slowly from cells unless there is an active excretion pathway as is present in the hepatocytes .
Young patients with thalassaemia studied longitudinally have shown absence of substantial improvement in the MR parameters of liver siderosis under different chelation therapy regimens . This may be explained by the fact that liver siderosis progresses very fast in thalassaemia patients, and iron overload develops after only 2 years of transfusion therapy . Therapy with the most widely used chelating agent is started at the age of about 3 years, and until growth is completed DFO should not exceed a dose of 40 mg/kg per day . An early start to monitoring the progress of tissue iron deposition with MRI might be useful in deciding whether to begin chelation therapy at a younger age and when to introduce new chelating agents.
Another important issue is whether myocardial iron evaluated by MRI could predict which patient will develop cardiac failure and arrhythmia. Thalassaemia patients with significantly reduced left ventricular ejection fraction (LVEF) have myocardial T2* values about 20 ms [22, 37, 38, 68], although most patients with T2* values less than 20 ms have normal LVEF . Furthermore, the relationship between myocardial R2 and LVEF is either non-existent or weak [22, 38, 39, 40]. Differences in cardiac functional indices among patients with similar MRI measurement of iron might also be explained by genetic factors that predispose to the more rapid development of cardiac failure when a critical level of myocardial siderosis is reached [99, 100].
Heterogeneous iron distribution in the myocardium has been demonstrated in histological studies, and this would suggest the necessity for global MRI estimation of myocardial iron [42, 101]. Pepe et al. , using T2* multiecho cardiac MRI, demonstrated that global T2* shows a close correlation with midseptal T2*, suggesting that the latter may be used for a quick assessment of myocardial siderosis.
MR studies have shown that cardiac siderosis increases with age, but that siderosis progresses more slowly in the heart than in the liver [28, 82]. Systematic MRI studies in children are necessary in order to evaluate the age at the start of cardiac iron overload.
Pituitary gland and brain
The single study assessing iron in the brain of thalassaemia patients by MRI  showed significantly higher R2 values in the cortex, the putamen and the caudate nucleus in patients than in controls. A lack of correlation between R2 and serum ferritin was found, possibly because serum ferritin, which does not cross the blood–brain barrier, is not a good marker of brain siderosis .
Abnormalities in adrenal function have been reported in patients with thalassaemia . There is only one study evaluating adrenals for iron overload with MRI (Fig. 5), which showed a significant correlation between adrenal and liver siderosis .
Spleen, lymph nodes and bone marrow
In spite of the fact that the spleen, lymph nodes and bone marrow, which all contain reticuloendothelial cells, are among the first organs to be affected by iron overload [1, 13], there have been very few studies evaluating their iron overload in thalassaemia by MRI, and these used mainly SIR techniques [59, 60, 62, 84]. SIR of the spleen shows a significant correlation with serum ferritin but not with SIR of the liver (Fig. 5) . The absence of correlation between liver and spleen siderosis could be explained by differences in iron kinetics, by differences in the cluster size of iron proteins, by haemochromatosis gene mutations in β-thalassaemia major carriers, and by the presence of extramedullary haemopoietic tissue in the spleen [78, 110, 111]. Intraabdominal lymph nodes in β-thalassaemia have been related to chronic hepatitis C . Lymph node siderosis correlates with liver, but not with spleen siderosis .
In the few studies that have been reported, SIR and relaxometry methods have shown discordant results for MR parameters of bone marrow siderosis and serum ferritin [59, 84]. Normal bone marrow signal associated with liver siderosis has been reported in a few patients with thalassaemia and this may be due to differences in genotype or differences in chelation therapy regimens .
It is evident that MR relaxometry has the potential to become the method of choice for non-invasive, safe and accurate assessment of iron load. Further theoretical research, along with studies monitoring wider age groups of patients are needed before a generally accepted protocol can be established. Until then, extreme caution is needed in its clinical application and interpretation. Experimental artefacts, non-optimized protocols, poor data analysis and unawareness of the inherent limitations of current methodologies in assessing a heavy body iron burden can result in misleading diagnosis and inappropriate management of thalassaemia patients with iron overload. Future experimental developments in relaxometry based on a better theoretical understanding of the contribution of the iron-containing proteins to the MR signal are expected to further strengthen its clinical role in the monitoring of patients with β-thalassaemia major.
- 6.Brittenham GM (2005) Iron balance in the red blood cell donor. Dev Biol (Basel) 120:77–82Google Scholar
- 47.Vymazal J, Urgosik D, Bulte JW (2000) Differentiation between hemosiderin- and ferritin-bound brain iron using nuclear magnetic resonance and magnetic resonance imaging. Cell Mol Biol (Noisy-le-grand) 46:835–842Google Scholar
- 48.Halle B (1999) Water in biological systems: the NMR picture. In: Bellisent-Funel M-C (ed) Hydration processes in biology. IOS Press, Amsterdam, pp 221–232Google Scholar
- 50.Bloembergen N (1957) Proton relaxation time in paramagnetic solutions. J Chem Phys 25:572–573Google Scholar
- 51.Gueron M (1975) Nuclear relaxation in macromolecules by paramagnetic ions: a novel mechanism. J Magn Reson 19:58–66Google Scholar
- 56.Denisov V, Venu K, Peters J et al (1997) Orientational disorder and entropy of water in protein cavities. J Phys Chem B 101:9380–9389Google Scholar
- 62.Papakonstantinou O, Drakonaki EE, Maris T et al (2006) MR imaging of spleen in beta-thalassemia major. Abdom Imaging. DOI 10.1007/s00261-006-9138-4
- 65.Carr HY, Purcell EM (1954) Effects of diffusion on free precession in nuclear magnetic resonance experiments. Phys Rev 94:630–638Google Scholar
- 66.Meiboom S, Gill D (1958) Modified spin-echo method for measuring nuclear relaxation times. Rev Sci Instrum 29:688–691Google Scholar
- 84.Argyropoulou MI, Kiortsis DN, Astrakas L et al (2007) Liver, bone marrow, pancreas and pituitary gland iron overload in young and adult thalassemic patients: a T2 relaxometry study. Eur Radiol. DOI 10.1007/s00330-007-0683-1
- 88.Kushner JP, Porter JP, Olivieri NF (2001) Secondary iron overload. Hematology Am Soc Hematol Educ Program 47–61Google Scholar
- 90.Atkin SL, Burnett HE, Green VL et al (1996) Expression of the transferrin receptor in human anterior pituitary adenomas is confined to gonadotrophinomas. Clin Endocrinol (Oxf) 44:467–471Google Scholar