A comparison of magnetic resonanceimaging and cardiac biopsy in the evaluationof heart iron overload in patients withb-thalassemia major

b-Thalassemia patients depend on adequate bloodtransfusions and continuous chelation to avoidtissue iron overload. Iron overload results fromboth excessive iron absorption and transfusionalsiderosis. Transfusional iron leads to iron depos-ition in the reticuloendothelial system of the spleen,

Mavrogeni SI, Markussis V, Kaklamanis L, Tsiapras D, ParaskevaidisI, Karavolias G, Karagiorga M, Douskou M, Cokkinos DV,Kremastinos DT. A comparison of magnetic resonance imaging andcardiac biopsy in the evaluation of heart iron overload in patients withb-thalassemia major.Eur J Haematol 2005: 75: 241–247. � Blackwell Munksgaard 2005.

Abstract: Objectives: To apply magnetic resonance imaging (MRI)for the assessment of myocardial iron deposition in patients withb-thalassemia and compare the results with cardiac biopsydata. Background: Myocardial iron accumulation is the main cause forcardiac complications in b-thalassemia. Methods: Twenty-five con-secutive thalassemic patients were studied using a 0.5-T (Tesla) system,ECG-gated, with echo time (TE) ¼ 17–68 ms. T2 relaxation time of theinterventricular septum was calculated assuming simple monoexponen-tial decay. A heart T2 relaxation time value of 32 ms was used for thediscrimination between high and low iron deposition. Heart biopsy wasperformed within a week after the MRI study. Patients with stainableiron in more than 50% of the myofibrils were graded as having severeiron deposition. A serum ferritin level below 2000 ng/mL was consideredas an indication of successful chelation. Results: Seven of the 25patients had heart biopsy indicative of low iron deposition (Group L)and the remaining 18 patients had heart biopsy indicative of high irondeposition (Group H). T2 relaxation time of the heart (T2H) was lowerin Group H compared to Group L (31.5 ± 3.9 (range: 28–40) ms vs.35.7 ± 3.7 (range: 29–40) ms, P ¼ 0.026). The T2H was in agreementwith heart biopsy in 86% of the patients in Group L and in 78% of thepatients in Group H (overall agreement 80%). Similarly, serum ferritinlevels were in agreement with heart biopsy in 28% and 88%, respectively(overall agreement 72%). In Group L, MRI was in better agreementwith biopsy compared to serum ferritin (86% vs. 28%, P < 0.05). Areceiver operating characteristic curve (ROC) analysis confirmed that aT2 relaxation time of 32 ms had the highest discriminating ability for thecorresponding biopsy outcome. Conclusions: Heart T2 relaxation timeappears in agreement with cardiac biopsy, both in high and lowiron deposition, and may become a useful non-invasive index inb-thalassemia.

Sophie I. Mavrogeni1, VyronMarkussis1, Loukas Kaklamanis1,Dimitrios Tsiapras1, IoannisParaskevaidis1, George Karavolias1,Markisia Karagiorga2, MarousoDouskou3, Dennis V. Cokkinos1,Dimitrios T. Kremastinos11Onassis Cardiac Surgery Center, Athens, Greece;2Aghia Sophia Children's Hospital, Athens, Greece;3Bioiatriki MRI Unit, Athens, Greece

Key words: heart biopsy; iron; magnetic resonanceimaging; thalassemia

Correspondence: Sophie I. Mavrogeni MD FESC,50 Esperou Street, 175-61 P. Faliro, Athens, GreeceTel/Fax: +30-210-9882797e-mail: soma@aias.gr

Accepted for publication 6 March 2005

Abbreviations: MRI, magnetic resonance imaging; T2H, T2

relaxation time of the heart; ROI, region of interest; SI, signal

intensity; TE, echo time; TR, repetition time.

Eur J Haematol 2005: 75: 241–247doi: 10.1111/j.0902-4441.2005.t01-1-EJH1876.xAll rights reserved

Copyright � Blackwell Munksgaard 2005

EUROPEANJOURNAL OF HAEMATOLOGY

241

liver and bone marrow. Despite chelation, irondeposition eventually occurs in all organs includingthe heart (1). Cardiac complications, such as heartfailure and arrhythmias are the major cause ofdeath in patients with thalassemia (2–4). Ironcardiomyopathy is reversible, if intensive chelationstarts in time (5), but diagnosis is often delayed bythe unpredictability of cardiac iron deposition andthe late development of symptoms and echocardi-ographic abnormalities (6). Once heart failuredevelops, the prognosis is usually poor with preci-pitous deterioration and death, despite intensivechelation. A direct measurement of myocardial ironwould allow early diagnosis and treatment and helpto reduce mortality.

While it is known that myocardial iron accumu-lation is the main cause for cardiac complications(2, 3), little is known about the natural history ofiron deposition in the heart. Myocardial irondeposition is not homogeneous (7) and does notoccur until other organs such as spleen and liverbecome saturated (8). Autopsy studies havedemonstrated that the extent of body iron load,as evaluated by hepatic iron deposition, does notcorrelate with cardiac iron deposition (8). Althoughiron deposited in the organs has been previouslyreported to correlate with serum ferritin (9), morerecent studies do not agree with this observation(10). Moreover, serum ferritin levels may be affec-ted by other factors such as fever or inflammation(11, 12).

The reference method for evaluating the extent ofbody iron excess is the measurement of the hepaticstorage iron concentration (13). Excess storage ironis detectable in reticuloendothelial parenchymalsites. Although liver biopsy with chemical analysisof tissue iron content provides the most accuratequantitative direct measurement of iron status, thediscomfort and the risk of the procedure excludesits frequent use in serial patients evaluation.Moreover liver biopsy does not provide any infor-mation about heart iron deposition (8, 12). This isvery important, since a marked discordance be-tween liver and heart iron deposition has beenreported in recent studies using magnetic resonanceimaging (MRI) (10, 14–16).

The MRI is able to provide information aboutiron deposition in individual organs. Detection ofiron deposition by MRI is based upon the ability ofstored intracellular iron to enhance the magneticsusceptibility of the tissues (17). Recent studies inexperimental animals have shown that the T2relaxation time has a linear correlation with thetotal iron content for several individual organs,including the heart (18). Myocardial iron depos-ition has been also studied in b-thalassemic patientsusing the same technique and extensively iron-

overloaded thalassemic patients had a significantlylower T2 relaxation time of the heart (T2H),compared to those with a lower amount of ironoverload (14).

The aim of this study was to assess the amount ofiron deposited in the myocardium of thalassemicpatients non-invasively by MRI and to comparethese findings with myocardial biopsy data.

Patients and methods

Patient population

Twenty-five (15 males and 10 females) consecutivethalassemic patients aged 27 (19–40) yr werestudied. All of them presented clinical signs ofheart failure according to Framinham’s studycriteria and functional classes II and III accordingto NYHA criteria (19). Left ventricular ejectionfraction was measured by magnetic resonance.Twenty of the patients were splenectomised. Exclu-sion criteria were the co-existence of any othercardiopulmonary or systemic disease. All weretransfused regularly (every 2–3 wk) to maintainhemoglobin levels at 10–13 g/dL.

Iron chelation therapy with desferrioxamine wasstarted before the age of 7 yr, in order to keepserum ferritin at the lowest levels. In patients bornbefore 1973 iron chelation therapy was given byintramuscular or intravenous route. During the last20 yr all patients were on chelation treatment withsubcutaneous infusion of desferrioxamine in a dailydose 30–50 mg/kg, given 5–6 times per week.Chelation treatment was monitored by frequentestimation of ferritin. A serum ferritin level below2000 ng/mL was considered as an indication ofsuccessful treatment with desferrioxamine (20). Theaverage ferritin levels of the last 5 yr were used inthis study. There was no alteration in bloodtransfusions of the patients due to the presence ofheart failure. There was no intensification of ironchelation therapy since the patients were already onmaximal chelation therapy.

Clinical and MRI evaluation was performed 48 hafter the last transfusion in all patients. Aninformed consent was obtained from all subjectsand the study was approved by the Onassis CardiacSurgery Center ethics committee.

Imaging technique

All MR studies were performed using a 0.5 T(Tesla) superconducting imaging system (MR-Vectra, GE/CGR). A body quadrature coil wasused for both excitation and signal detection. Thesingle oblique orientation of the imaging slices(head-left to feet-right) was determined employing

Mavrogeni et al.

242

a scout coronal T1 localizer image in order todepict axial slices of the heart. A mid-ventricularslice with optimal left ventricular delineation wasselected. The MR study was ECG-gated withrepetition time (TR): heart rate, slice thickness10 mm and echo time (TE) ¼ 17–68 ms in foursymmetrically repeatable echoes. These parameterswere chosen on the basis of phantom studies, wherea mean accuracy of 5.5% was demonstrated for T2values ranging from 10 to 80 ms. The field of viewwas 42 cm and the image reconstruction matrix was160 · 224 for both sequences. T2 relaxation timemeasurements of the left ventricle of the heart(T2H) were calculated using a single region ofinterest (ROI) that included the inter-ventricularseptum area. Heart biopsy specimens were alsotaken from the same area, under echocardiographicguidance. Assuming single exponential behavior,cardiac pixel signal intensity (SI) decays exponen-tially with TE in the base images of a multi-echosequence. The rate of exponential decay can thus becalculated by means of a mathematical fit on SI andTEs data values. In our lab, normal adult heart T2relaxation time values are 48.3 ± 5.5 (range: 37–59) ms (14). A heart T2 relaxation time value of32 ms was used for the discrimination between highand low iron deposition, based on a previous studyfrom our group (14). The observers were blinded tothe biopsy and ferritin data of the patients.

Biopsy

Endomyocardial biopsy was performed within 2–5 d after the MRI study under echocardiographicguidance in order to be as close as possible to theMR slices. Five to eight specimens were obtainedfrom the interventricular septum, using a percu-taneous right internal jugular vein approach.After positioning the end of the bioptome againstthe endocardium of the interventricular septum,the bioptome was opened, gently advancedagainst the endocardium, and then closed. Onwithdrawal of the bioptome, a small portion ofventricular myocardium with attached endo-cardium was obtained (21). The specimens werefixed in buffered formalin and embedded inparaffin. Tissue sections were examined andspecial stains were performed (Prussian blue) forthe detection of iron deposition. A great variationin the number of positive myocytes in differentareas of the same fragment was observed. Allbiopsy fragments of the same subject displayedthe same cell iron distribution and all thefragments showed stainable iron.Samples were analyzed semi-quantitatively

according to iron deposition. Myocardial irongrading was derived by averaging the percentage

of positive myofibrils in a section over the numberof biopsy fragments. Patients with stainable iron inless 50% (range: 1–49%) of the myofibrils weregraded as having mild and patients with more than50% (range: 50–99%) of the myofibrils were gradedas having severe iron deposition (22). Biopsymaterial with less than 200 myocytes present wasnot included in the analysis.

Statistical analysis

The data were expressed as mean ± SD. Meandifferences between groups of the study were testedusing the unpaired two-tailed Student’s t-test.Association between various continuous parame-ters was sought using the Pearson’s r-correlationcoefficient. Chi-square test was used to evaluaterelationships between categorical variables. Agree-ment between techniques was considered when theystratified a patient in the same group (high or lowiron overload).

Receiver operating characteristic curves (ROC)[sensitivity · (1 ) specificity)] were plotted and theareas under the curve (AUC) for each diagnosticmarker were calculated. Finally, cut-off pointanalysis was used in order to determine the optimalvalue of the investigated markers that differentiatehigh from low iron deposition. In particular, thecrucial point defined by the largest distance fromthe diagonal line of the curve (23).

All reported P-values are from two-sided tests.STATA release 8 (STATA Corp., College Station,TX, USA) was used for the statistical calculations.

Results

According to heart biopsy data, patients wereseparated in two groups. Group L (n ¼ 7) withlow iron deposition (<50% of myocytes affected)(range: 25–45%) and Group H (n ¼ 18) with highiron deposition (>50% of myocytes affected)(range: 55–85%) (Figs 1 and 2).

Patients in Group H were at similar age com-pared to those in Group L (29.5 ± 7.8 vs.26.4 ± 4.0 yr, P ¼ 0.197) and had similar serumferritin levels (2791 ± 489 vs. 2357 ± 690 ng/mL,P ¼ 0.090). Patients in both groups maintainedsimilar hemoglobin levels with transfusions(11.2 ± 0.8 vs. 11.5 ± 0.8 g/dL, P ¼ 0.409).

Left ventricular ejection fraction (%) was notdifferent between the high and low iron group(42.9 ± 2.4 vs. 44.1 ± 2.8, P ¼ 0.286). Heart T2Hin Group H was lower compared to Group L(31.5 ± 3.9 vs. 35.7 ± 3.7 ms, P ¼ 0.026)(Table 1). The intra-observer and inter-observercoefficient of variation for T2 measurements was7% and 13%, respectively.

Heart MRI vs. biopsy in b-thalassemia

243

In Group L biopsy was in agreement with T2relaxation time in six out of seven patients (86%),and in agreement with serum ferritin only in twoout of seven patients (28%) (P < 0.05). In GroupH, biopsy was in agreement with T2 relaxation timein 14 out of 18 patients (78%) and with serumferritin in 16 out 18 patients (88%) (Tables 2 and3). Heart T2 relaxation time had 80% and serumferritin levels 72% overall agreement with biopsy.

Serum ferritin levels correlated negatively withthe T2H (r ¼ )0.51, P < 0.01) in the studypopulation. The ROC for serum ferritin levelsand T2H are presented in Fig. 3. The AUC ofserum ferritin levels was by 10% lower compared tothe AUC of T2 relaxation time (66%, 95% confid-ence interval 41–91% vs. 76%, 95% confidenceinterval 54–97%). Cut-off point analysis revealedthat T2 relaxation time levels <32 ms had thehighest discriminating ability for the correspondingbiopsy outcome (sensitivity ¼ 78% andspecificity ¼ 86%), while serum ferritin con-centrations >2650 ng/mL were the optimal

discriminating level (sensitivity ¼ 72% andspecificity ¼ 57%).

Discussion

There are limited studies at present comparingbiopsy and T2 relaxation time for iron depositionestimation. Human data are available only for liver(16, 24–27) where an excellent correlation was

Fig. 1. Cardiac biopsy (top) (·200, Prussian blue) and MRIof the heart (bottom) (T2H ¼ 40 ms) from a patient withmild iron deposition.

Table 1. Clinical and MRI parameters of thalassemia major patients with low(Group L) and high (Group H) iron deposition according to heart biopsy

Group L Group H P value

n 7 18Age (yr) 26.4 € 4.0 29.5 € 7.8 0.197Hb (g/dL) 11.5 € 0.8 11.2 € 0.8 0.409Serum ferritin (ng/mL) 2357 € 690 2791 € 489 0.090LV EF (%) 44.1 € 2.8 42.9 € 2.4 0.286T2H (ms) 35.7 € 3.7 31.5 € 3.9 0.026

Hb, hemoglobin; T2H, T2 relaxation time of the heart; LV EF, ejection fraction of leftventricle measured by magnetic resonance.

Fig. 2. Cardiac biopsy (top) (·200, Prussian blue) and MRIof the heart (bottom) (T2H ¼ 29 ms) from a patient withsevere iron deposition.

Mavrogeni et al.

244

found between T2 relaxation time of the liver andliver biopsy results. Concerning the heart, there isonly one study in experimental animals (18) wherea linear correlation was documented between T2Hand iron in heart samples. To our knowledge, this isthe first study comparing T2H and cardiac biopsydata in a cohort of thalassemic patients.Endomyocardial biopsy is a helpful means of

assessing structural changes in a variety of cardiacdiseases. Postmortem studies have clearly demon-strated that iron is distributed in the heart in aninhomogeneous way (28). It is known that whensome iron was visible, then all the biopsy fragmentsshowed stainable iron. Thus the local iron inhomo-geneity is relative and may be overcome by aver-

aging iron grading over a number of biopsysamples (29).

In this study we stratified our thalassemic popu-lation in two groups, based on biopsy results. Theevaluation of iron accumulation was performed on5–8 endomyocardial fragments. In relation tovariation between different myocardial regions, weperformed the biopsy under echo guidance in orderto be sure that all fragments came from theinterventricular septum area. Specimens with>50% and <50% iron loaded myocytes wereconsidered indicative of high and low iron over-load, respectively (22). Thereafter, we comparedthese findings with the MRI and serum ferritindata. In the MRI study a heart T2 relaxation timeof 32 ms was used for discrimination between lowand high iron deposition, based on a previous studyfrom our group (14). Similarly, serum ferritin levelsbelow 2000 ng/mL were considered as an index ofsuccessful chelation and low iron overload (20).

Table 2. The T2H and serum ferritin levels in thalassemia major patients with low(Group L) and high (Group H) iron overload according to heart biopsy

T2H (ms) Serum ferritin (ng/mL)

Group L29 300040 120036 210034 250040 300036 180035 2900Group H28 250029 280028 320031 310030 295028 310029 220029 280031 400030 300030 280030 300031 280030 290037 180039 200038 250040 2800

T2H, T2 relaxation time of the heart.

Table 3. Agreement between MRI, serum ferritin levels and biopsy data

Heart biopsyLow iron(Group L)

High iron(Group H) All patients

(A) MRI dataLow iron by MRI 6 4High iron by MRI 1 14Agreement with biopsy (%) 86* 78 80

(B) Serum ferritin dataLow iron by serum ferritin 2 2High iron by serum ferritin 5 16Agreement with biopsy (%) 28* 88 72

*P < 0.05, chi-square test (MRI vs. serum ferritin: agreement with biopsy in GroupL).

0.0 0.2 0.4 0.6 0.8 1.0

1–specificity

1–specificity

0.0

0.2

0.4

0.6

0.8

1.0

Sen

siti

vity

Sen

siti

vity

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

1.0

Fig. 3. The ROC for T2 relaxation time (top) and serumferritin (bottom). T2 relaxation time at a value of 32 ms hadthe highest discriminating ability for the correspondingbiopsy outcome.

Heart MRI vs. biopsy in b-thalassemia

245

In our patient population, T2H was in agreement(concordance between techniques) with heart biopsydata in the majority (80%) of the patients. Theagreement between biopsy and MRI data was 86%in the low-iron group (according to biopsy) and78% in the high-iron group. Heart iron deposition ispatchy and may lead to sampling error (7). How-ever, heart biopsy tissue was sampled from theinterventricular septum where iron is depositedmore heavily that in the rest of myocardium (30)andMRImeasurements were performed at the samearea, in order to be closer to biopsy.

In this population the T2H was lower in the highiron overload group, compared with the low ironoverload group (P ¼ 0.026). A ROC analysisbetween biopsy and T2 relaxation time confirmedthat the T2 relaxation time, at a value of 32 ms, hadthe highest discriminating ability for the corres-ponding biopsy outcome in our sample. Theejection fraction of the left ventricle did not differbetween the two groups. These findings arecomparable with other studies (10), where MRIparameters are an early indicator of cardiacinvolvement in iron overload, before any changein ejection fraction becomes evident.

When affected tissues are exposed to a magneticfield, the presence of iron overload causes localirregularities that lead to signal loss and this effectis concentration dependent (31, 32). In this way thepresence of iron affects tissue T1 and T2 relaxationtime. Decrease in T2 relaxation time is caused bydephasing of water protons as they diffuse throughfield inhomogeneities created by magnetic bodies inthis case ferritin molecules (33). The measurementof T2 relaxation time in thalassemic patients hassome peculiarities. Although generally T2 relaxa-tion time is independent of field strength, there is anexception. When the alteration of tissue relaxationtime is due to iron deposition, there is a lineardependence of 1/T2 on field strength (33). In thepast, T2 relaxation time in most published studieshas been measured using 0.5-T systems, where theeffect of field is lower and the measurement of thisparameter is feasible using the available TEs. Inanother study from our group using a 1.5-T system,heart and liver T2 relaxation time was not meas-urable in severe iron overloaded patients because SIwas equal to background noise, applying thecommercially available TEs (34).

Recently a newmagnetic resonance T2* techniquehas been used for the measurement of tissue iron,with validation to chemical estimation of iron inpatients undergoing liver biopsy. A significant cur-vilinear, inverse correlation between iron concen-tration by biopsy and liver T2* was found (10). Atthe moment results comparing T2* and myocardialbiopsy are missing, but it seems that T2* is a new

promising technique, with good reproducibility, ableto overcome the field strength problem. However,the sequence applied in T2* measurement is notcommercially available at present.

The measurement of serum ferritin provides anindirect estimate of body iron stores, but theusefulness of this measurement is limited by manyclinical conditions such as inflammation, infection,liver disease, in which serum ferritin is not a reliableindicator of body iron (12). Serum ferritin levelswere in agreement with heart biopsy data in 72% ofthe patients. The agreement between biopsy andserum ferritin levels was 88% in the high-irongroup and 28% in the low-iron group (significantlylower agreement, compared to heart T2 relaxationtime, in the low-iron group). This may be due to thefact that variations in serum ferritin mainly corres-pond to changes in reticuloendothelial systemstorage and not to changes in parenchymal ironcontent (25), while heart T2 relaxation time repre-sents a myocardial index of iron deposition. In ourpatient population, serum ferritin levels correlatednegatively with heart T2 relaxation time as previ-ously published (14).

There are some limitations concerning this study.The heart biopsy analysis was semi-quantitative,like all published human cardiac biopsy studies iniron overload. The magnetic field used was 0.5 T,whereas the majority of currently available systemsare 1.5 T. Therefore, our findings should be alsoevaluated in a 1.5-T system, using very short TEs.A single mid-ventricular slice used in the MRIstudy does not allow assumptions about irondeposition in the rest of the cardiac muscle.However, the T2* measurement study (10) has thesame limitation.

In conclusion, this is a first study in humanscomparing heart biopsy data with MRI for heartiron deposition. Heart T2 relaxation time, as anindex of iron deposition, measured by MRI appearsin good agreement with cardiac biopsy. Moreover,it is closer to biopsy compared to serum ferritin inpatients with lower iron deposition. MRI, being anoninvasive index, may prove useful in b-thalasse-mia evaluation.

References

1. Modell B, Khan M, Darlison M. Survival in beta-tha-lassaemia major in the UK: data from the UK Thalas-saemia Register. Lancet 2000;355:2051–2052.

2. Sankul D, Thakerngpal K, Parcharee P. Cardiacpathology in 76 thalassemic patients. Birth Defects1988;23:177–181.

3. Ohene-Frempong K, Schwartz E. Clinical features ofthalassemia. Pediatr Clin North Am 1980;27:403–411.

4. Chrysohoou C, Panagiotakos D, Barbetseas Y, BrilliS, Lambrou S, Karagiorga M, Ladis VA, Pitsavos C,Stefanadis C. Echocardiographic and electrocardio-

Mavrogeni et al.

246

graphic prognostic factors of heart failure in young patientswith b-thalassemia major: a 10-year (1995–2004) follow-up.Int J Hematol 2004;80:336–340.

5. Wacker P,Halperin DS, Balmer-Ruedin D, Oberhansli

I, Wyss M. Regression of cardiac insufficiency after ambu-latory intravenous deferoxamine in thalassemia major.Chest 1993;103:1276–1278.

6. Henry WL, Nienhuis AW, Wiener M, Miller DR,Canale VC, Piomelli S. Echocardiographic abnormalitiesin patients with transfusion-dependent anemia and secon-dary myocardial iron deposition. Am J Med 1978;64:547–555.

7. Olson LJ, Edwards WD, McCall JT, Ilstrup DM,Gersh BJ. Cardiac iron deposition in idiopathic hemo-chromatosis: histologic and analytic assessment of 14 heartsfrom autopsy. J Am Coll Cardiol 1987;10:1239–1243.

8. Buja LM, Roberts WC. Iron in the heart: etiology andclinical significance. Am J Med 1971;51:209–221.

9. Jacobs A. The pathology of iron overload. In: Jacobs A,Worwood M, eds. Iron in Biochemistry and Medicine, II.New York: Academic Press, 1980:439–452.

10. Anderson LJ, Holden S, Davis B et al. CardiovascularT2-star (T2*) magnetic resonance for the early diagnosis ofmyocardial iron deposition. Eur Heart J 2001;22:2171–2179.

11. JacobsA,MillerF,WorwoodM,BeamishRW,Wardrop

CA. Ferritin in the serum of normal subjects with iron defi-ciency and iron overload. Br Med J 1972;4:206–208.

12. Crosby WH. Serum ferritin fails to indicate hemochroma-tosis: nothing gold can stay.NEngl JMed1976;294:333–334.

13. Barry M, Sherlock S. Measurement of liver iron con-centration in liver biopsy specimens. Lancet 1971;1:100–103.

14. Mavrogeni S, Maris T, Gouliamos T, Vlahos L,Kremastinos DT. Myocardial Iron deposition inb-thalassemia studied by magnetic resonance imaging. Int JCardiac Imaging 1998;14:117–122.

15. Johnston DL, Rice L, Vick GW, Hedrick TD, Rokey R.Assessment of tissue iron overload by nuclear magneticresonance imaging. Am J Med 1989;87:40–47.

16. Jensen PD, Jensen FT, Christensen T, Eiskjaer H,Baandrup U, Nielsen JL. Evaluation of myocardial ironby magnetic resonance imaging during iron chelationtherapy with deferrioxamine: indication of close relationbetween myocardial iron content and chelatable iron pool.Blood 2003;101:4632–4639.

17. Brittenham GM, Badman DG. Noninvasive measurementof iron: report of an NIDDK workshop. Blood2003;101:15–19.

18. Liu P, Henkelman M, Joshi J et al. Quantification ofcardiac and tissue iron by nuclear magnetic resonancerelaxometry in a novel murine thalassemia-cardiac ironoverload model. Can J Cardiol 1996;12:155–164.

19. McKee PA, Castelli WP, McNamara PM, Kannel WB.The natural history of congestive heart failure, theFramingham Study. N Engl J Med 1971;285:1441–1446.

20. Olivieri NF, Nathan DG, MacMillan JH, Wayne AS,Liu PP, McGee A, Martin M, Koren G, Cohen AR.

Survival in medically treated patients with homozygousb-thalassemia. N Engl J Med 1994;331:574–578.

21. Caves P, Coltart J, Billingham MB, Rider A, StinsonE. Transvenous endomyocardial biopsy-application of amethod for diagnosing heart disease. Postgrad Med J1975;51:286–290.

22. Barosi G, Arbustini E, Gavazzi A, Grasso M, Pucci A.Myocardial iron grading by endomyoacardial biopsy. Aclinico-pathologic study on iron overloaded patients. Eur JHaematol 1989;42:382–388.

23. Guangqin MA, Hall WJ. Confidence bands for receiveroperating characteristic curve. Med Decis Making1993;13:191–197.

24. Stark DD, Moseley ME, Bacon BR, Moss AA, Gold-

berg HI, Bass NM, James TL. Magnetic resonance ima-ging and spectroscopy of hepatic iron overload. Radiology1985;154:137–142.

25. Gomori JM, Horev G, Tamary H et al. Hepatic ironoverload: quantitative MR imaging. Radiology1991;179:367–369.

26. Papakonstantinou O, Maris T, Kostaridou V et al.Assessment of liver iron overload by T2-quantitativemagnetic resonance imaging. Correlation of T2-QMRImeasurements with serum ferritin concentration andhistologic grading of siderosis. Magn Res Imaging1995;13:967–977.

27. Gandon Y, Olivie D, Guyader D, Aube C, Oberti F,Sebille V, Deugnier Y. Non-invasive assessment of hep-atic iron stores by MRI. Lancet 2004;363:357–362.

28. Cutler GJ, Isner JM, Bracey AW, Hufnagel CA,Conrad PW, Roberts WC, Kerwin DM, Weintraub

AM. Hemochromatosis heart disease: an unemphasizedcause of potentially reversible restrictive cardiomyopathy.Am J Med 1980;69:923–928.

29. Arnett EN, Nienhuis AW, Henry WL, Ferrans VJ,Redwood DR, Roberts WC, Roberts WC. Massive myo-chondrial hemosiderosis: a structure function conference atthe National Heart and Lung Institute. Am Heart J1975;90:777–787.

30. Schellhammer PF, Engle MA, Hagstrom JWC. Histo-chemical studies of the myocardium and conduction systemin acquired iron-storage disease. Circulation 1967;35:631–39.

31. Stark DD. Hepatic iron overload: paramagnetic patho-logy. Radiology 1991;179:333–305.

32. Hardy PA, Henkelman RM. Transverse relaxation rateenhancement caused by magnetic particulates. Magn ResonImaging 1989;7:265–275.

33. Vymazal J, Brooks RA, Zak O, McRill C, Shen C, Di

Chiro G. T1 and T2 of ferritin at different field strengths:Effect on MRI. Magn Res Med 1992;27:368–374.

34. Mavrogeni S, Gotsis E, Markussis V, Tsekos N, PolitisC, Vretou E, Kermastinos D. T2 relaxation time study ofiron overload in b-thalassemia. MAGMA 1998;6:7–12.

Heart MRI vs. biopsy in b-thalassemia

247