| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Dept of Medicine, Div of Digestive Disease and Nutrition (H.L.B., R.B.R., E.E.C.), Dept of Radiology (A.D., T.H.P.R., D.D.S.), and Center for Study of Disorders of Iron and Porphyrin Metabolism (H.L.B., R.B.R., E.E.C.), University of Massachusetts Medical School and University of Massachusetts Memorial Health Care, Rm S6-737, 55 Lake Ave North, Worcester, MA 01655. Received Feb 20, 1998; revision requested Apr 27; revision received Sep 8; accepted Dec 17. H.L.B. supported in part by the Magnetic Resonance Imaging Center of Central Massachusetts and by a grant from the National Institutes of Health (DK 38825). Address reprint requests to H.L.B. (e-mail: Herbert.Bonkovsky@ummhc.org).
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Thirty-eight patients underwent percutaneous needle biopsy of the liver with chemical measurement of the hepatic iron concentration and hepatic MR imaging with several spin-echo and gradient-recalled-echo (GRE) techniques. Correlations between MR imaging parameters and the hepatic iron concentration were determined.
RESULTS: Inverse curvilinear relationships were noted between several MR parameters and hepatic iron concentrations. GRE sequences with short repetition and echo times were more accurate and precise than spin-echo sequences for the estimation of hepatic iron concentration. A GRE sequence with a repetition time of 18 msec, an echo time of 5 msec, and a flip angle of 10° showed close correlation between the hepatic iron concentration and the natural logarithm of the ratio of the signal intensity of liver to the SD of background noise (r = -0.94) and low coefficient of variation (12%).
CONCLUSION: MR imaging with these parameters is a rapid, noninvasive, and accurate modality for estimation of hepatic iron concentration; it is sufficiently accurate and precise to obviate liver biopsy for the purpose of measuring hepatic iron concentration.
Index terms: Hemochromatosis, 761.659 • Iron • Liver, biopsy, 761.1261 • Liver, diseases, 761.659 • Liver, iron content, 761.659 • Magnetic resonance (MR), comparative studies, 761.121411, 761.121412
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
However, in the United States only about 80%–85% of patients with the hemochromatotic phenotype have these mutations in the HFE gene, and their prevalence is even lower in other geographic regions (4–7). In addition, an appreciable number of subjects who are homozygous abnormal, especially women, do not express the iron overload phenotype (8). Then too, there are clearly other types of hereditary hemochromatosis (eg, those occurring in black Africans or Melanesians), in which causative mutations occur in other still unidentified genes (9).
Because the liver is the principal organ responsible for the storage and detoxification of iron, and also the first and foremost organ damaged by heavy iron overload, the standard for the diagnosis of hemochromatosis is liver biopsy with determination of the hepatic iron index value (10). This index is defined as the hepatic iron concentration, expressed as micromoles of iron per gram of dry liver, divided by the age of the patient in years. Liver biopsy, however, is invasive and may be associated with complications. Furthermore, the variability in measured hepatic iron concentrations in small liver samples from patients with cirrhosis may be considerable (11,12).
To avoid liver biopsy, attempts have been made to devise noninvasive means of estimating the hepatic iron concentration. A dedicated device to measure magnetic susceptibilities, the superconducting quantum interference device, is highly sensitive for the detection of iron but is not widely available (13). Although computed tomography demonstrates an increase in the attenuation of the liver in hepatic iron overload, it is relatively insensitive to mild degrees of increased hepatic iron, especially if there is associated fatty change in the liver (14). Magnetic resonance (MR) imaging has been considered a promising noninvasive modality to estimate hepatic iron concentration (15–17). An increase in hepatic iron concentration decreases the T2 relaxation time, which decreases the liver's signal intensity and makes the liver appear dark on typical MR images.
Standard MR imaging sequences, known as spin-echo sequences, have been shown to be sensitive, but not specific, for distinguishing mild from marked degrees of iron excess (18). Newer gradient-recalled-echo (GRE) sequences, which are more sensitive to the field inhomogeneities induced by paramagnetic substances such as iron, have shown more promise (15,19) but have not been widely used to study hemochromatosis. The present study was undertaken to develop a rapid, cost-effective, accurate test to estimate hepatic iron concentration without the need for performance of percutaneous needle biopsy of the liver.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
On the basis of a synthesis of the clinical history, hepatic iron index value, and serum study and hepatic histologic findings, final diagnoses were made. As shown in Table 1, 15 patients had hereditary hemochromatosis, 16 had chronic viral hepatitis, five had alcoholic liver disease, and two had steatohepatitis. Each patient in the study underwent both percutaneous needle biopsy of the liver with chemical measurement of the hepatic iron concentration and MR imaging. The two studies were carried out within 1 month of one another. Patients did not receive iron removal therapy during the interval between the two tests.
|
Liver Biopsy
Biopsy of the right lobe of the liver was performed with a 16-gauge Klatskin-modified Menghini needle (Becton-Dickinson, Rutherford, NJ) by using the intercostal, percutaneous approach. Biopsies were performed through the eighth right intercostal space at the midaxillary line. The majority of the specimen was placed in Carnoy solution or 10% buffered formaldehyde saline, processed routinely, embedded in paraffin, and sectioned. The embedded sections were stained with hematoxylin-eosin, Masson trichrome, and Perls stains. Another portion of the biopsy specimen was dried, weighed, ashed, and submitted to measurement of the hepatic iron concentration by a colorimetric procedure (20).
MR Imaging Procedures
Within 1 month of the time of liver biopsy, all patients underwent MR imaging with a unit (Signa; GE Medical Systems, Milwaukee, Wis) operating at a field strength of 1.5 T. In all but five patients, the MR study was performed prior to the biopsy. The following sequences were used for all patients (Table 2): Spin-echo sequence 600/17 (repetition time msec/echo time msec), 2,000/80, and 2,000/12; GRE sequence (18/5) with flip angles of 10° and 70°; GRE sequence (39/30) with flip angles of 10° and 70°; and GRE sequence (120/5) with flip angles of 10° and 70°. These sequences were selected by performing pilot experiments, which tested machine limits in various combinations of repetition time, echo time, and flip angle. Sequences outside these limits showed no improvement or substantial degradation of signal-to-noise ratios (SNRs) per unit of time. The GRE sequences (18/5) were performed with the patients holding their breath at the end of inspiration. A single 2-cm section was chosen that included both the liver and the spleen. The examinations were performed with two signals acquired, and the matrix was held constant at 256 x 128 pixels. One patient also was examined with a fast spin-echo (6,000/117) sequence (see Results).
|
In general, five regions of interest each were drawn on each liver and five on the air above the patient and other tissues. From these, mean values for signal intensities were obtained (Fig 1). The regions of interest included the region of the right lobe of the liver on which biopsy was performed. Ratios were calculated by dividing the mean signal intensity of the liver by that of the SD of the mean of the signal intensity of the overlying air, or noise, or the mean signal intensities of the other tissues (21). Initial plots of these ratios as functions of hepatic iron concentration revealed curvilinear relationships. Therefore, natural logarithms (ln) of the ratios were calculated. Plots of the ln of the ratios versus hepatic iron concentration were linear and inverse and were thus used in subsequent analysis. The coefficient of variation for the ln of the ratios was calculated as the SD of this parameter divided by its mean value multiplied by 100 to convert it to a percentage.
|
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Hepatic iron concentrations were highest in patients with untreated hereditary hemochromatosis (mean, 8,520 µg [153 µmol] of iron per gram of dry liver). The mean hepatic iron concentration was only mildly elevated in patients with chronic viral hepatitis or alcoholic liver disease (mean values, 2,000 µg [35.8 µmol]/g dry liver and 1,860 µg [33.3 µmol]/g dry liver, respectively). Hepatic iron concentration values were normal in the two patients with steatohepatitis.
Preliminary MR Imaging Studies
In preliminary studies, regions of interest were drawn on the liver, as well as on the fat, spleen, paraspinous muscles, and air overlying the patient in the same transverse section (Fig 1). Regions of interest on the fat were difficult to draw in a standardized manner because of variations in body habitus of the patients. The coefficient of variation for the determination of the signal intensity of fat in an individual patient was usually greater than 30%. In contrast, the SD of pixel signal intensities of the overlying air, or noise, was easily measured in a standard manner in all patients and was not affected by body habitus. This SD was reproducible. Because variances were consistently lower when hepatic signal intensities were normalized to the overlying air, or noise, for routine use, the signal intensity ratio of the liver to that of this SD was used in all subsequent studies.
MR Imaging Characteristics of Patients in the Study
The GRE sequences, sensitive as they are to minute fluctuations in the local magnetic field, generally tended to perform better than the standard MR imaging sequences. An inverse linear correlation was noted with all MR imaging sequences between hepatic iron concentration and ln (SNR). The closest inverse linear correlation between hepatic iron concentration and ln (SNR) was with the GRE sequence (18/5) with a flip angle of 10° (r = -0.94, P < .001) (Table 2).
A scatterplot of hepatic iron concentration versus ln (SNR) in our optimal MR imaging sequence of GRE (18/5) with a flip angle of 10° is depicted in Figure 2. The straight line shows the least-squares best-fit of the data, and the curved lines indicate 95% confidence limits. No patient with ln (SNR) greater than 3.2 had hepatic iron concentration greater than three times the upper limit of normal (3,600 µg [64.5 µmol]/g dry liver). At levels of mild to moderate hepatic iron overload (2,500–7,500 µg [44.8–134 µmol]/g dry liver, n = 8), our optimal GRE sequence performed well in that a close inverse linear relationship between hepatic iron concentration and ln (SNR) was observed within this range. With this optimal GRE sequence (18/5, 10° flip angle), the determination of ln (SNR) was highly reproducible at hepatic iron concentration less than 10,000 µg (179 µmol)/g dry liver, with a coefficient of variation of only 8%–13%.
|
Figure 1 depicts an MR image of a man with HLA-linked hemochromatosis. A striking feature of this image is the markedly diminished signal intensity of the liver in comparison with surrounding structures such as the spleen and paraspinous muscles.
An example of the benefits of short GRE images with breath-hold techniques is shown in Figure 3. The conventional T2-weighted image (2,000/80) shows a dark liver. However, in this relatively long, non–breath-hold sequence, motion artifact from respiration contributes to appreciable background noise and degradation of contrast between liver and paraspinous muscle (Fig 3a). With more recently developed fast spin-echo T2-weighted images (6,000/117), data are rapidly acquired, and motion artifact is decreased. However, the difference between liver and muscle has been virtually eliminated because of the heavy T2 weighting (Fig 3b). In contrast, our optimal GRE sequence provides excellent contrast and limited motion artifact (Fig 3c).
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
We found that, with all of the MR imaging sequences studied, there was an inverse linear correlation between the hepatic iron concentration and the ln (SNR). Standard sequences performed less well than the newer GRE sequences (Table 2, Figs 3, 4), which is in agreement with other recent results (19). As recently noted by Stark (15), the inherent sensitivity of GRE sequences to paramagnetic substances is consistent with the superior performance of the GRE sequences in this study. Another major advantage of the GRE sequences is that, unlike the standard sequences that take 10 minutes to perform, they can be performed in a matter of seconds.
Of the various GRE sequences studied, the GRE sequence with the shortest repetition and echo times performed the best. This is probably because data acquisition was of sufficiently short duration that the patients were able to comfortably hold their breath during the sequence, thus minimizing motion artifact and other sources of noise. Noise has been noted previously to be a major problem in quantitative MR imaging (24).
From analysis of the scatterplot of hepatic iron concentration versus ln (SNR) with our optimal GRE sequence (Fig 2), we were able to establish a retrospective arbitrary threshold at ln (SNR) of 3.2; no patient with ln (SNR) greater than 3.2 had hepatic iron concentration greater than three times the upper limit of normal (3,600 µg [64.5 µmol]/g dry liver). The value of such a threshold for the clinician is that it allows one, by using only 19 seconds of imager time, to exclude clinically important hepatic iron overload in an individual patient.
For completion of data acquisition with this optimal sequence, the total facility time, including turnover time between patients, was approximately 15 minutes. At the current rate of charge for imager time in the United States of $800 per hour, the charge for this test might be only about $200, which is far less than the charges for liver biopsy and measurement of hepatic iron concentration at our and most other centers. At our center (University of Massachusetts Memorial Health Care), for example, typical charges for liver biopsy with quantitative hepatic iron concentration measurement are as follows: The short-stay facilities charge is $170, the hepatologist's fee for performing biopsy is $330, the fee for processing the specimen is $237, the pathologist's fee for interpretation of biopsy results is $220, the fee for hepatic iron concentration measurement is $146, and the total is $1,103. The average actual reimbursement to our clinical system and group practice is considerably lower because of discounts negotiated with third-party insurers. In addition, because the MR imaging procedure is noninvasive, it is more comfortable than liver biopsy for many patients, although some who are claustrophobic are unable to lie in the MR imaging machine, even for 15 minutes.
Chemical measurement of hepatic iron concentration is accurate and precise when adequate samples are assayed under ideal conditions (25). However, errors occur in standard practice, probably owing to inadequate specimen sizes, contamination, or laboratory error. Then, too, the variability of hepatic iron concentration estimation in cirrhotic livers is substantial, with coefficients of variation as high as 30%–40% (11,12). With respect to the concern of possible sampling error, MR imaging has an advantage in that the entire liver is visualized, and several truly representative regions of interest can be selected and results averaged among them (Fig 1). For example, in our study the mean coefficient of variation for the determination of ln (SNR) was only 12%, and the coefficients of variation for patients with cirrhosis in our study were not different from those without cirrhosis.
Another potential advantage of MR imaging over liver biopsy is that, due to the excellent visualization of the liver, important additional information may be obtained. In this study, for example, an operable hepatocellular carcinoma was found in one patient, and unanticipated hepatic lesions were discovered in another.
Because of the inability of MR imaging to show hepatic histopathology, liver biopsy retains its importance in the assessment of hepatic fibrosis and inflammation. As a rule, we continue to recommend liver biopsy in all patients with palpable livers, increased serum aminotransferase, or serum ferritin levels greater than 1,000 ng/mL (1,000 µg/L). Biopsy results that show fibrosis or cirrhosis provide important prognostic information and suggest the need to evaluate the severity of portal hypertension and the need for close follow-up for development of hepatocellular carcinoma. However, our results establish that MR imaging is useful to exclude iron overload in patients in whom there is only a low to moderate suspicion of the disease (eg, in a sibling of a person known to have hemochromatosis with normal or only mildly elevated iron study results).
MR imaging can also be useful in patients who adamantly refuse liver biopsy or those with decompensated liver disease and severe coagulopathy, in whom needle biopsies pose increased risks (eg, heavy drinkers with elevated liver chemistries, prothrombin times, and serum ferritin levels or transferrin saturations). Another potential role for quantitative MR imaging is to follow patients noninvasively through the course of phlebotomy or chelation therapy for primary or secondary hemochromatosis. Such uses for quantitative hepatic MR imaging have already been described (26,27). Figure 5 shows the usefulness of MR imaging in following up a patient during the course of phlebotomy therapy.
|
|
Bonkovsky et al (18) used a heavily T2-weighted spin-echo MR imaging sequence to examine 48 patients with abnormal iron parameters and found a correlation coefficient of -0.8 between the hepatic iron concentration and the signal intensity ratio of liver to paraspinous muscle. However, the 95% CIs still were thought to be too wide for this MR imaging method to be practical clinically. In addition, the spin-echo MR imaging sequences showed a lack of specificity. Figure 3 provides an example of the limitations of standard T2-weighted spin-echo or fast spin-echo sequences compared with rapid GRE sequences.
In agreement with the findings of an earlier pilot study (31), the results of this study showed that the ratio of signal intensities of liver to overlying air, or noise, had a closer relationship to the hepatic iron concentration values than the ratio of signal intensities of liver to that of fat, muscle, or spleen (Table 2). Our results are also in accord with those of Gandon et al (19) who also found the GRE sequences more sensitive than the standard sequences. Our results differ, however, in that Gandon et al (19) found that the liver-to-fat signal intensity ratios generally correlated more closely with hepatic iron concentration values than did the liver-to-air ratios. We speculate that the reason we found the liver-to-air ratios more precise is because of the artifactual variability of subcutaneous fat on MR images, which is related to variations in body habitus, radio frequency, and the positioning of the magnet.
Because higher magnetic field strengths provide larger magnetization gradients, higher signal contrasts can be achieved with 1.5-T than with 0.5-T magnets (24). Therefore, our 1.5-T unit, theoretically, is better suited for quantitation of hepatic iron concentration than the 0.5-T unit used by Gandon et al (19). The magnetization of ferritin itself is proportional to the strength of the ambient magnetic field (32), and previous studies (33) have suggested that higher magnetic field strength increases transverse relaxation. Thus, higher field strength imaging systems are more sensitive and accurate for the estimation of hepatic iron concentration in the range of clinical interest.
With regard to future directions and developments, the most challenging problem is finding a way to standardize noninvasive estimation of hepatic iron concentrations from one MR imaging facility to another. Measurements of signal intensity may vary from one machine to another, even if the machines are of the same type. Therefore, other investigators will need to calibrate their MR imaging machines against chemical measurements, as was done in this study. Ideally, phantoms that could be used to establish standard curves for iron measurements by means of MR imaging instruments would be developed, thus obviating the use of human subjects for machine calibration. Such phantoms would have to be stable, durable, and available at a reasonable cost and in sufficient quantity such that they would be generally available.
In summary, this study demonstrates that rapid GRE MR imaging at high magnetic field strength is a sensitive, accurate, and noninvasive modality for the estimation of hepatic iron concentration and for ruling out clinically important hepatic iron overload (ie, greater than three times the upper limit of normal). Because of the greater sensitivity and reduced random variability of results obtained with GRE sequences, such methods are preferred. Indeed, the variability of results with such noninvasive methods applied to cirrhotic livers appears to be less than that of traditional direct chemical measurements performed on small portions of biopsy specimens. Furthermore, they should be less costly than liver biopsy with direct measurement of hepatic iron concentration, although liver biopsy retains an important place in evaluation and management because of the other information it can uniquely provide.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
3 Current address: Dept of Microbiology and Immunology, Penn State College of Medicine, Hershey, Pa.
4 Current address: Dept of Radiology, Onze Lieve Vrouwe Gasthuis, Amsterdam, the Netherlands.
5 Current address: Dept of Radiology, University of Nebraska Medical Center, Omaha, Neb.
The opinions expressed in this paper are those of the authors; they do not necessarily reflect the official views of the National Institutes of Health.
Abbreviations: GRE = gradient-recalled echo ln = natural logarithm SNR = signal-to-noise ratio
Author contributions: Guarantor of integrity of entire study, H.L.B.; study concepts, H.L.B., A.D., D.D.S.; study design, H.L.B., E.E.C., R.B.R.; definition of intellectual content, all authors; literature research, T.H.P.R., H.L.B., R.B.R., E.E.C.; clinical studies, T.H.P.R., H.L.B., R.B.R., A.D.; experimental studies, T.H.P.R., H.L.B., D.D.S.; data acquisition, T.H.P.R., H.L.B., R.B.R., A.D., D.D.S.; data analysis, T.H.P.R., H.L.B., R.B.R., E.E.C.; statistical analysis, H.L.B., R.B.R., E.E.C.; manuscript preparation, H.L.B., R.B.R., E.E.C., A.D., D.D.S.; manuscript editing, H.L.B., A.D., D.D.S.; manuscript review, H.L.B., R.B.R., A.D., D.D.S.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. P. O'Regan, M. F. Callaghan, M. Wylezinska-Arridge, J. Fitzpatrick, R. P. Naoumova, J. V. Hajnal, and S. A. Schmitz Liver Fat Content and T2*: Simultaneous Measurement by Using Breath-hold Multiecho MR Imaging at 3.0 T--Feasibility Radiology, May 1, 2008; 247(2): 550 - 557. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. L. Lauzon, H. Mahallati, and R. Frayne Time-efficient breath-hold abdominal MRI at 3.0 T. Am. J. Roentgenol., September 1, 2006; 187(3): 649 - 657. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. W. Swinkels, M. C.H. Janssen, J. Bergmans, and J. J.M. Marx Hereditary Hemochromatosis: Genetic Complexity and New Diagnostic Approaches Clin. Chem., June 1, 2006; 52(6): 950 - 968. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. G. St. Pierre, P. R. Clark, W. Chua-anusorn, A. J. Fleming, G. P. Jeffrey, J. K. Olynyk, P. Pootrakul, E. Robins, and R. Lindeman Noninvasive measurement and imaging of liver iron concentrations using proton magnetic resonance Blood, January 15, 2005; 105(2): 855 - 861. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. B. Salem, J.-P. Cercueil, F. Ricolfi, and D. Krause Case 75: Erythropoietic Hemochromatosis Radiology, October 1, 2004; 233(1): 116 - 119. [Full Text] [PDF]
|
|
|
|
|
|
J. M. Alustiza, J. Artetxe, A. Castiella, C. Agirre, J. I. Emparanza, P. Otazua, M. Garcia-Bengoechea, J. Barrio, F. Mujica, and J. A. Recondo MR Quantification of Hepatic Iron Concentration Radiology, February 1, 2004; 230(2): 479 - 484. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. D. Jensen, F. T. Jensen, T. Christensen, H. Eiskjaer, U. Baandrup, and J. L. Nielsen Evaluation of myocardial iron by magnetic resonance imaging during iron chelation therapy with deferrioxamine: indication of close relation between myocardial iron content and chelatable iron pool Blood, June 1, 2003; 101(11): 4632 - 4639. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. M. Brittenham and D. G. Badman Noninvasive measurement of iron: report of an NIDDK workshop Blood, January 1, 2003; 101(1): 15 - 19. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L.J. Anderson, S. Holden, B. Davis, E. Prescott, C.C. Charrier, N.H. Bunce, D.N. Firmin, B. Wonke, J. Porter, J.M. Walker, et al. Cardiovascular T2-star (T2*) magnetic resonance for the early diagnosis of myocardial iron overload Eur. Heart J., December 1, 2001; 22(23): 2171 - 2179. [Abstract] [PDF]
|
|
|
|
|
|
L. Pieroni, L. Khalil, F. Charlotte, T. Poynard, A. Piton, B. Hainque, and F. Imbert-Bismut Comparison of Bathophenanthroline Sulfonate and Ferene as Chromogens in Colorimetric Measurement of Low Hepatic Iron Concentration Clin. Chem., November 1, 2001; 47(11): 2059 - 2061. [Full Text] [PDF]
|
|
|
|
 |
|
 J. P. Kushner, J. P. Porter, and N. F. Olivieri Secondary Iron Overload Hematology, January 1, 2001; 2001(1): 47 - 61. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. L. Bonkovsky, A. Davidoff, D. D. Stark, M. Kami, T. Hamaki, Y. Kishi, E. Angelucci, G. M. Brittenham, and G. Lucarelli Hepatic Iron Concentration and Total Body Iron Stores in Thalassemia Major N. Engl. J. Med., November 30, 2000; 343(22): 1656 - 1657. [Full Text]
|
|
|
|
|
|
E. Angelucci, G. M. Brittenham, C. E. McLaren, M. Ripalti, D. Baronciani, C. Giardini, M. Galimberti, P. Polchi, and G. Lucarelli Hepatic Iron Concentration and Total Body Iron Stores in Thalassemia Major N. Engl. J. Med., August 3, 2000; 343(5): 327 - 331. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
