European Heart Journal (20010) 22, 2971-2179
`doi:i0. 1053/cuhj.2001.2822, available online at http:/fwww.idealibrary.com on IDE mL
`
`Cardiovascular T2-star (T2*) magnetic resonance for
`the early diagnosis of myocardial iron overload
`
`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? and
`D. J. Pennell’
`
`' Cardiovascular MR Unit, Royal Brompton Hospital, London; * University College Hospital, London:
`‘Whittington Hospital, London, U.K.
`
`Aims To develop and validate a non-invasive method for
`measuring myocardial iron in order to allow diagnosis and
`treatment before avert cardiomyopathy and failure develops.
`
`Methods and Results We have developed a new magnetic
`fesonance T2-star (T2*) technique for the measurement of
`lissuc iron, with validation to chemical estimation ofiron in
`patients undergoingliver biopsy. To assess the clinical value
`of this technique, we subsequently correlated myocardial
`iron measured by this T2* technique with ventricular
`function in 106 patients with thalassaemia major. There
`wis a significant, curvilinear, inverse correlation between
`iron concentration by biopsy and liver T2* (r=0-93,
`P<0-0001). Inter-study cardiac reproducibility was 5-0%.
`As myocardial
`iron increased,
`there was a progressive
`decline in cjection fraction (r=0-61, P<0-001). All patients
`with ventricular dysfunction had a myocardial T2* of
`<20ms. There was no significant correlation between
`myocardial T2* and the conventional parameters of iron
`status, serum ferritin and liver iron. Multivariate analysis of
`clinical parameters to predict the requirement for cardiac
`
`medication identified myocardial 12* as the most signifi.
`cant variuble (odds ratio 0-79, P<0-002),
`
`iron deposition can be repro-
`Conclusions Myocardial
`ducibly quantified using myocardial T2" and this is the
`most significant variable for predicting the need for ven-
`tricular dysfunction treatment. Myocardial
`iron content
`cannot be predicted from serum fesritin or liver iron, and
`canventional assessments of cardiac function can only
`detect those with advanced disease. Eurly intensification of
`iron chelation therapy, guided by this technique, should
`reduce mortality from this reversible cardiomyopathy.
`(Eur Heart J 2000; 22: 2171-2179, doi:10.1053fewhj.2001,
`2822)
`(©) 2001 The European Society of Cardiclogy
`
`Key Words: Cardiomyopathy, magnetic
`imaging, heart failure and thalassaemia.
`
`resonance
`
`See page 2140, doi:10.1053/euhj.2000.2951 for the Editorial
`comment on this article
`
`
`
`Introduction
`
`Cardiac failure secondary to transfusional iron overload
`remains the commonest cause of death in patients with
`thalassaemia major!'*!_ In the United Kingdom, approxi-
`mately 50%of patients die before reaching the age of
`3581, The cardiomyopathyis reversible if intensive iron
`chelation treatmentis instituted in time! “, but diagnosis
`is often delayed by the unpredictability of cardiac iron
`deposition and the late development of symptoms, and
`echocardiographic abnormalities!”"!, Once heart failure
`
`Revision submitted 11 June 2001, accepted 13 June 2001, and
`published 19 October 2001.
`Correspondence: Professor D. J. Pennell, Cardiovascular MR Unit,
`Royal Brompton Elospital, Sydney Street, London SW3 6NP,
`U.K.
`
`0195-668X/01/222171 +09 $35.00/0
`
`develops, the outlook is usually poor”! with precipitous
`deterioration and death, despite intensive chelation.
`Direct measurement of myocardial
`iron would allow
`earlier diagnosis und treatment and help to reduce
`mortality from ihis reversible cardiomyopathy.
`The aim of this study was to develop a reproducible
`magnetic resonance (MR) method for quantifying
`myocardial
`iron concentration, For this purpose, we
`investigated myocardial T2-star (T2*), a relaxation par-
`ameter arising principally from local magnetic field
`inhomogeneities that are increased with iron deposition.
`Although a rare disease in the U.K.,
`thalassaemia
`the commonest genetic disorder worldwide, with
`is
`approximately 94 million heterozygotes for beta thalas-
`suemia and 60.000 homozygotes born each year, Iron
`overload cardiomyopathy is also a complication of
`hereditary haemochromatosis, which predominantly
`
`¢) 2061 The European Society of Cardiology
`
`Apotex Tech.
`Ex. 2015
`
`

`

`OL. J. Anderson et al.
`2172
`
`
`uffects these of northern European ancestry, where
`homozygous mutations of the HFE gene approximate
`0-5%!"" Patients receiving multiple transfusions during
`chemotherapy and bone marrow transplant, or for other
`indications such as sickle cell anaemia, may also benefit
`from assessment with this technique.
`
`Methods
`
`Study populations
`
`Liver biopsy patients
`We prospectively studied 30 beta-thalassaemic patients
`(12 females and 18 males, aged 18-38, mean 27-1 + 6-7
`years) undergoing liver biopsy for routine clinical man-
`agement. The biopsy iron concentrations were compared
`with the liver T2* measurements derived by MR. All
`scans were performed within 21 days of the liver biopsy
`(mean 10+ 7-0 days} and no adjustments to chelation
`trealment were made between investigations. In 27 cases,
`a section of the biopsy specimen was also examined
`histologically (cirrhosis 3 patients, periportal fibrosis
`10 patients).
`
`Thalassaemia major cohort
`A total of
`[09 regularly transfused patients with
`thalassaemia major were scanned. Three patients were
`excluded from comparison analysis of ventricular func-
`tion due to cardiac anomalies (1 corrected tetralogy of
`Fallot,
`1 subaortic shelf and | peripheral pulmonary
`urtery stenosis), The residual cohort of 106 paticnts
`included 50 males and 56 females, aged 13-41, mean
`27 +7 years. All patients had received iron chelation
`therapy since the mid-to-late 1970s, or from early child-
`hood in patients born after this time, with a broad
`range of compliance to treatment (serum ferritin 262
`7624 pg .1>', mean 20954 1559ue.17'), Seventeen
`patients required medication for ventricular dysfunc-
`tion (antiarrhythmics or angiotensin-converting-enzyme
`inhibitors).
`
`Normal subjects
`Normal ranges for T2* values in the liver, heart, spleen
`and skeletal muscle were established in
`15 healthy
`volunteers (9 males, 6 females, aged 26-39, mean
`3E+3-7 years),
`
`Magnetic resonance
`
`Patients were scanned with a Picker 1-ST Edge Scanner
`(Marconi Medical Systems, Ohio, U.S.A.). Each scan
`lasted approximately 45 min and included the measure-
`tment of liver and heart T2*, and left and right ventricu-
`lar
`function, volumes
`and mass
`using standard
`techniques!"*],
`The liver T2* was determined as follows: a single
`10mm slice through the centre of the liver was scanned
`at eight different echo times (TE 2-2-20-1 ms). Each
`image was acquired during a 10-13 breath-hold using
`a gradient-echo sequence (repetition time 200 ms, flip
`angle 20°, matrix 96 x 128 pixels, field of view 35 cm,
`sampling bandwidth of 125 kHz), The signa! intensity of
`the liver parenchyma and the background noise were
`measured in each of the cight images using in-house
`software (CMRtools, «:) Imperial College), Background
`noise was subtracted from theliver signal intensity, and
`the net value was plotted against the echo time for each
`image. A trendline wasfitted to the resulting exponential
`decay curve, with an equation of the form y=Ke~ TT"
`where K represents a constant, TE represents the echo
`time and y represents the image signal intensity.
`For the measurement of myocardial T2*, a single
`short axis mid-ventricular slice was acquired at nine
`separate echo times (TE 5-6-17-6 ms). The repetition
`time between radiofrequency pulses was between [1:8
`23-8 ms, depending on the echo time used. A gradient
`echo sequence wasused {flip angle 35°, matrix 128 x 256
`pixels, phase encode group 8,
`field of view 35cm,
`sampling bandwidth of 250 kHz). The repetition time
`was adjusted to the patient’s heart rate. Each image was
`acquired during an 8-135 breath-hold. A gating delay
`time of O ms after the R-wave was chosen in order to
`obtain myocardial images in a consistent position in the
`cardiac cycle irrespective of the heart
`rate. A full-
`thickness region of intercst was measured in the left
`ventricular myocardium, encompassing both epicardial
`and endocardial regions. This was located in the septum,
`distant from the cardiac veins, which can cause suscep-
`tibility artefacts!""!, The myocardial T2* was calculated
`using the same method as that in the liver.
`
`Statistical analysis
`
`Liver biopsyiron assays
`
`All biopsy specimens were analysed at the Royal Frec
`Hospital, London!'*!, The dry weights ofall specimensin
`this study exceeded 0-5 mg (mean 1-33 4 0-59 mp).
`
`Serum ferritin measurements
`
`Measurements of serum ferritin were carried out by
`enzyme immunoassay (WHO Ferritin 80/602 First
`International Standard, normal range 15-300 pg . 17").
`
`Summary data are presented as mean+ 1 standard
`deviation. Pearson’s and Spearman's tests were used to
`assess the correlation between liver iron and liver T2*.
`For reproducibility data, the coefficient of variation was
`defined as the standard deviation of the differences
`between the two separate measurements, divided by
`their mean and expressed as a percentage. T2* values
`measured in healthy volunteers showed a normal distri-
`bution and are expressed with 95%reference ranges.
`Pearson’s cocfiicient of correlation was used to assess the
`degree of association between myocardial T2* and liver
`
`

`

`CMR for carlydiagnosis of cardiac iron overload 2173
`
`
`
`15
`Liver T2* (ms)
`
`(b)
`
`°
`
`-_battsLdjeooooroonocou
`
`a
`
`0
`
`a
`
`eSGe
`
`te
`
`cda
`
`0
`
`-1
`
`0-5
`
`0
`
`Od
`
`1
`
`15
`
`2
`
`2-5
`
`3
`
`a5
`
`Log,liver T2*
`
`(a) Regression curve for the relationship between liver T2* and fiver
`Figure |
`biopsy iron concentration. Black circles depict fibrotic biopsies, and squares depict
`non-fibrotic biopsies. The fibrotic samples show increased variability, compatible
`with previous reports, (b) There was a close linear relation between T2* and liver
`iron concentration in the non-fibrotic samples following lep,
`transformation
`(r=0-93, P<()-0001), sce text for details.
`
`T2* and myocardial T2* and serum ferritin. Stata
`statistical software was used for computations (Stata
`Corporation, Texas, U.S.A.).
`
`Results
`
`Validation of T2* values as a measurement
`oftissue fron concentration
`
`There was a significand, curvilinear, inverse correlation
`between liver T2* and the liver iron content for all
`samples (r=081, Fig. I(a)). There was a better corre-
`
`the fibrotic samples (r=0-68), as would be predicted
`from the known variability of iron measurements
`from fibrotic biopsies!"*""l, Therefore we subsequently
`employed non-fibrotic samples to generate predictions
`ofliver iron content from the measured T2* values. As
`liver iron concentration and liver T2* measurements
`were positively skewed, the values were log, transformed
`in order to analyse the correlation (Fig. I(b)). For the
`non-fibrotic samples, both Pearson's and Spearman's
`tests gave a correlation coeflicient of 0-93 which is highly
`significant (P<0-0001), Regression analysis shows that a
`one unit increase in log, T2* is associated with a §-07
`unit increase in log, iron concentration (95%confidence
`
`

`

`OL. #. Anderson etat.
`2174
`
`
`Reproducibility
`
`Ten patients were scanned on two occusions to assess the
`inter-study reproducibility of the T2* technique (interval
`1-2t days, mean 7-1 days). The coellicient of variation
`was 3:3%for the liver and 50%for the heart. This
`compared favourably with coefficients of variation for
`signal
`intensity ratio measurements from these same
`images (liver-to-muscie 7-9%, liver-to-noise 8:8", heart-
`to-muscle 12-6%, and heart-to-noise 14-1"), techniques
`that have previously been used.
`The images from 10 patients were studied indepen-
`dently by two observers to assess inter-observer varia-
`bility. The coefficient of variation was 4-5%for the liver
`and 6-4% for the heart. This compared favourably with
`signal
`intensity ratio measurements (liver-to-muscle
`54%,
`liver-to-noise 6°1%, heart-to-musele (0-8%, and
`heart-10-noise 7-5%).
`
`Normal T2* values
`
`The normal values for T2* using the technique described
`above were: Heart 52 + 16 ms, liver 33+ 7 ms, skeletal
`muscle 30+ 5 ms, spleen 56 + 22 ms.
`
`Heart iron, liver iron, serumferritin and the
`relationships between these variables
`
`In many patients we found a marked discordance
`between liver and heart iron concentration (Fig, 2) and
`no significant correlation could be found between liver
`and heart T2* in this large cohort (r=0-15, P=0-11).
`Similarly, no significant correlation was found between
`heart T2* and serum ferritin level at the lime of the scan
`(r=0-10, P=0-32). To confirm that this finding was net
`due to spurious individual ferritin readings,
`the mean
`ferritin for 12 months prior to the scan was also com-
`pared to heart T2*, and once aguin there was no
`significant correlation (r=0-09, P=0-35).
`
`Myocardial iron and parameters of
`ventricular function
`
`In the normal range of myocardial T2* (lower 95%
`confidence interval 20 ms}, parameters of ventricular
`function (ejection fraction, volume and mass)fell within
`the normal range!) (Fig. 3). Below a myocardial T2* of
`20 ms, there was a progressive and significant decline in
`left ventricular ejection fraction (r=0-61, P<0-0001) and
`an increase in the left ventricular end-systolic volume
`index {r=0-50, P<0-0001), and left ventricular mass
`index (r=0-40, P<0-001).
`
`Myocardial T2* andclinical outcome
`
`Logistic regression was performed to relate the require-
`
`
`
`
`Figure 2. Discardance of liver and heart iron deposition.
`Short axis plane, including the adjacent liver (TE 5-6 ms).
`The top panct shows a patient with severe cardiac iron
`deposition but minimal liver iron deposition (beart darker
`than liver). The lower panel shows a patient with normal
`myocardial iron but severe liver iron overload (liver darker
`than beart}.
`
`Of 106 patients, 17 patients required medication for
`ventricular dysfunction, and univariate analysis identi-
`fied myocardial T2*,
`Ieft ventricular ejection fraction
`and ieft ventricular end systolic volume as significant
`variables (Table 1). Using multivariate backward step-
`wise regression analysis, with a cut-off of P=0-1 for
`removing variables and P=0-05 for including variables,
`only myocardial T2* (odds ratio 0-79, 95% confidence
`interval 0-67-0-92, P=0-002) and serum ferritin (odds
`ratio 095, 95%confidence interval 0-91-1-00, P=0-05)
`were significant. Depite the lack of correlation between
`
`

`

`CMRfor carly diagnosis of cardiac iron overload 2175
`
`
`1
`*
`s a
`ComoneataiwalesGBaaseenCaeOEE
`ton * ¢ *e,
`*
`s
`=
`$y
`
`0
`
`10
`
`20
`
`30
`
`60
`50
`40
`Heart T2* (ms)
`
`70
`
`80
`
`90
`
`100
`
` 0
`
`10
`
`20
`
`30
`
`60
`60
`40
`Heart T2* (ms)
`Figure 3 Relationships between myocardial T2* values and parameters of ventricular
`function: (a) left ventricular ejection fraction, (b} left ventricular mass index, {c)
`left
`ventricular end-systolic volume index. The broken lines represent
`the normal reference
`ranges for myocardial T2* and parameters of cardiac function.
`
`100
`
`70
`
`80
`
`90
`
`#3 aaeeeeeeeeLe
`atteeeeee
`
`160
`
`140
`
`120
`
`100
`
`80
`
`60
`
`40
`
`20
`
`0
`
`10
`
`20
`
`30
`
`~2)
`
`end-systolicvolume
`
`index(ml.m
`
`60
`50
`40
`Heart T2* (ms)
`
`70
`
`80
`
`90
`
`100
`
`

`

`2176
`
`L. J. Anderson ct al.
`
`Figure 4 MR gradient echo images of differential tissuc iron clearance before and during
`intravenous chelation therapy: before treatment {top row), after 3 months (middle row) and
`after 6 months of treatment (bottom row). The left column shows the liver (transaxial view,
`TE 4°5 ms), and the right column shows the heart (horizontal Jong axis, TE 14 ms), which is
`dilated (RY — right ventricle 182 mt end-diastolic volume, LV — left ventricle 183 ml end
`diastolic volume). Thereis severe iron loading (dark tissue signal) in the liver (L), pancreas
`(P) and beart prior to treatment(liver T2*=1-2 ms, myocardial T2*=10 ms). By 3 months,
`the liver iron is noticeably improved (liver ‘T2*=5-1 ms), but cardiac iron deposition has
`changed little (T2*=10-1 ms). Myocardial iron deposition only shows improvement at 6
`months (T2*=12-1 ms) and even ut
`this time the heart remains dilated (end-diastolic
`
`

`

`CMR for early diagnosis af cardiac iron overlaad 2177
`rrrra
`
`for cardiac medication. This may be related to increased
`clinical vigilance in the treatment of patients with high
`serum ferritins, As cardiac medication had been initiated
`in all patients prior to CMR scanning, physicians were
`blinded to myocardial T2* measurement. Myocardial
`T2* values were between 49 ms and 13 ms in this
`group.
`
`Discussion
`
`Iron overload pathophysiology
`
`Iron overload occurs either duc to excess gastrointestinal
`absorption or secondary to repeated blood transfusions.
`The human body has no mechanism forexcreting excess
`iron, which is stored as crystalline iron oxide within
`ferritin and hacmosiderin in the body. The actiology of
`the iron overload effects the tissue distribution of iron.
`In hereditary haemochromatosis, iron is carried from
`the intestine to the liver via the portal vein (as transfer-
`tin} and deposited in the periportal hepatocytes,
`In
`severe disease,
`iron is also deposited in the pancreas.
`heart and endocrine organs. In thalassaemia, iron over-
`load results from both excessive iron absorption and
`transfusional siderosis. Transfusional iron leads to iron
`deposition in the reticulo-endothelial system of the
`spleen, liver and bone marrow. In advanced cases iron
`also accumulates in parenchymalcells ofthe liver, heart,
`pancreas and endocrine organs, which are sensitive to
`the toxic effects of iron, When the iron-binding capacity
`of transferrin is cxhausted, free iron appears as non-
`transferrin bound iron (NTBI). The toxicity of NTBI is
`much higher than bound iron, and promotes hydroxyl
`tedical formation resulting in peroxidative damage to
`membranelipids and proteins, In the heart this results in
`impaired function of the mitochondrial
`respiratory
`chain and is manifested clinically as heart failure!!*),
`The presence of two types of iron explains the nature
`of the relationship between myocardial function and
`iron concentration (as shown in Fig, 3). As iron accu-
`mulates in the normal storage form in the heart, the T2*
`falls, but there is little effect on cardiac function until a
`threshold is reached where the iron storage capacity is
`exhausted. At this point NTBI starts to appear, which
`profoundly affects cardiac function. Thus the relation-
`ship between the measured T2* and cardiac function is
`shallow until a critical level is reached, after which rapid
`deterioration occurs. This explains why identification of
`abnormalsystolic function is a late sign of iron toxicity.
`Iron clears more slowly from the heart than the liver
`(Fig. 4), which may contribute to the high mortality of
`patients with established cardiomyopathy despite inten-
`sive chelation, Using this T2* technique,it is possible to
`identify much cartier those patients who require inten-
`sive chelation prior to the onset of systolic dysfunction
`and this should avoid the mortality associated with overt
`heart failure.
`
`Table | Univariate analysis of clinical variables to test
`the strength of their relationship to the need for cardiac
`medication
`
`
`
` Variable Odds ratio (95%CI) P value
`
`
`
`0-003
`0-81 (0-71, 0-93)
`Myocardiul T2* (ms)
`<0-001
`0-88 (0-82, 0-94)
`LVEF (%)
`0-001
`1-05 (1-02, 1-08}
`LVESY¥ (ml}
`O17
`0-97 (6-93, 1-01)
`Scrum ferritin (ug .1~*)
`ORS
`1-01 (0-91, 1:12)
`Liver T2* (inl)
`0-39
`1-58 (8-56, 4-51)
`Diabetes mellitus
`
`
`1-01 (0-94, 1408)Age 0-85
`
`ClSconfidence interval; LVEF =Ieft ventricular ejection fraction:
`LVESV=left ventricular end systolic volume.
`
`MR T2* technique to measure myocardial
`iron
`
`We chose a gradient-echo T2* sequence rather than a
`spin-echo T2 sequence becauseof the greatersensitivity
`to iron deposition. T2* is relaicd to T2 by summation of
`lissuc relaxation (T2), and magnetic inhomogeneity,
`known as T2 prime (T2"), in the form:
`1/72" = 1/12 + 1/T2’
`
`loss in affected tissues
`Tron overload causes signal
`because iron deposits become magnetized in the scanner,
`inducing focal irregularities in the magnetic field, which
`cause water protons around these deposits to lose phase
`coherencel’!, This effect is concentration dependent".
`An additional benefit of the shorter acquisition times
`of gradient-echo images is minimization of motion
`artefacts from myocardial contraction and respiratory
`movement, which greatly affect
`the accuracy and
`reproducibility of T2 images,
`Previous work using spin-echo techniqueswith signal-
`intensity-ratios or T2 measurements, have shown an
`inverse relationship to liver iron concentration!?!-™i.,
`However, in practice, the limited sensitivity of spin-echo
`techniques, motion artefacts and poor signal to noise
`at longer echo times?!775] have made quantification
`of myocardial
`iron unsatisfactory, Because of
`these problems, gradient-echo techniques using signal-
`intensity-ratios have recently been used to quantify liver
`iron?" but no studies in the heart have been
`reported. We have used multiple echoes to generate T2*
`instead of relying on signal
`ratios between tissues,
`and for the first time have apptied the technique to the
`heart. This range of echo times improves quantification
`of severe iron overload, provides high sensitivity at
`low and normal
`tissue iron levels, and gives greater
`teproducibility than signal intensity ratios.
`The normal value of myocardial T2* in this study was
`52+ f6ms,. There is limited literature with which to
`compare these results. Li ef af. studied 13 normals and
`reported 4 T2* of 33 + 6-5 ms, but only two echo times
`were used", Wacker et ai. reported the normal myo-
`cardial value in six patients with coronary disease
`
`

`

`L. J. Andersonet al.
`2478)
`rrPre
`
`(remote from ischaemia) as 48 +9 ms using a JO echo
`time technique!*"!, Reeder reports normal T2* values of
`38 + 6 ms in the mid septum in five normal volunteers,
`and showed reduced values adjacent to the cardiac veins
`due to their local susceptibility!The variation in these
`values may result from residual T! effects associated
`with the short repetition times imposed in breath-hold
`wcquisitions, and may lead to over-estimation of T2*.
`However,
`this effect
`is much less significant
`in the
`presence of short myocardial T2* values
`in
`iron
`overload.
`
`Validation of T2* measurements, and
`variability between tissues
`
`We have shown a significant curvilinear correlation
`between liver T2* and biopsy iron concentration
`(r=0-93, P<0-000i, for non-fibrotic livers), Using the
`T2* technique in a large cohort of patients, we also
`found that
`there is no reliable relationship between
`myocardial T2* and serum ferritin or liver T2*. This
`indicates that cardiological management based on these
`established parameters of iron status is unreliable.
`
`addition, relaxation parameters such as T2* may be
`machine and sequence dependent, and further vatidation
`work is required before widespread use.
`
`Conclusions
`
`Gradient-echo T2* MR provides a rapid, non-invasive,
`reproducible means
`for assessing myocardial
`iron.
`Myocardial
`iron content cannot be predicted from
`scrum ferritin or liver iron, and conventional assess-
`ments of cardiac function can only detect
`those with
`advanced discase. Early diagnosis and treatment of
`myocardial
`iron overload is
`likely to prevent
`the
`mortality seen in patients with established ventricular
`dysfunction.
`
`Melissa Wright of the Hammersmith Hospital, London gave
`valuable statistical advice. This work was supported by a British
`Heart Foundation Junior Fellowship Grant
`(FS/98064),
`the
`Wellcome Trust, and CORDA,the heart charity.
`
`References
`
`Heart failure and causality
`
`The poorpredictive value of serum ferritin and liver iron
`measurements have made heart disease difficult to detect
`in thalassaemia, raising questions over the causal re-
`lationship between cardiac iron overload and cardiac
`failure??"4), Recently, myocarditis has been implicated
`in the developmentofheart failure in thalassaemial®?-""1,
`This
`study demonstrates
`the
`relationship between
`deterioration in ventricular function and myocardial
`iron loading and illustrates clear evidence for the causal-
`ity of iron overload and heart failure in thalassacmia
`patients. Whilst our study supports the aetiological role
`of iron in thalussaemic cardiomyopathy, other factors
`such as antioxidant statc!*?! may also be important.
`
`Study limitations
`
`{l] Zurlo MG,De Stefano P, Borgna-Pignatti C ef af. Survival
`and causcs of death in thalassacmia major. Lancct 1989; 2:
`27-30.
`{2) Olivieri NF, Nathan DG, MacMillan JH et af. Survival
`in medically
`treated paticnts with homozygous
`beta-
`thalassacmia. N Engl J Med 1994; 331; 574-8,
`(3) Modcil B, Khan M, Darlison M. Survival in beta thalassacmia
`major in the UK: Data from the UK Thalassaemia Register,
`Lancet 2000; 355 (9220): 2051-2,
`(4) Aldouri MAWB, Hoffbrand AV, Flynn DM, Ward SE,
`Agnew JE, Hilson AJ. High incidence of cardiomyopathy
`in beta-thalassaemia patients receiving regular transfusion
`and iron chelation:
`reversal by intensificd chelation, Acta
`Haematol 1990; 84: 113-7.
`I, Wyss M.
`(5) Wacker PHD, Balmer-Rucdin D, Oberhansli
`Regression of cardiac insufliciency after ambulatery intra-
`venous deferoxeminc in thalassacmia major. Chest 1993; 103:
`1276-8.
`[6] Davis BA, Porter JB. Long-term outcome of continuous
`34-hour deferozumine infusion via indwelling intravenous
`catheters in high-nsk beta-thalussacmia. Blood 2000; 95:
`1229-36.
`[7] Henry WLNA, Wiener M, Miller DR, Canale VC, Piomelti §
`Echocardiographic abnormatlitics
`in patients with trans-
`fusion-dependent anaemia and secondary myocardial
`iron
`deposition. Am J Med 1978; 64: 547-55,
`(8) Nienhuis AW, Griffith P, Strawezynski Her af. Evaluation
`of cardiac function in patients with thalassacmia major. Ann
`NY Acad Sci 1980; 344: 384.96.
`It is not possible definitively to predict myocardial iron
`[9] Felker GM, Thompson RE, Hare JM et af. Underlying causes
`concentration from the myocardial T2* value, because
`and long-term survival in patients with initially unexplained
`no validation has been performed with cardiac tissue.
`cardiomyopathy, N Engl J Med 2000; 342: 1077-84,
`This requires myocardial biopsies and will be difficult
`(10] Weatherall DJ. Anaemia us a World Health Problem. Oxford
`because of inhomogencus myocardial deposition?*!
`Textbook of Medicine. Oxford University Press 1996; 22:
`3463-82.
`and small samples. Nonetheless, the data presented in
`[il] Olynyk JK, Cullen DJ, Aquilia S, Rossi E, Summerville L,
`this study, showing the strong relationship between
`Powell LW. A population-based study of the clinical cxpres-
`declining myocardial T2* and impaired ventricular func-
`sion of the hemochromatosis genc. N Engl J Med 1999, 341:
`THR-24
`tion, clearly indicates the empirical value of myocardial
`T2*. and the validation data from the liver biopsies
`
`
`
`[12] Barry M, of—liver-ironSherlock S$. Measurement
`concentration in needle-biopsy specimens. Lancet 1971;
`1:
`supports the relationship between tissue iron and T2*.
`[60-3,
`The F2* of iron-loaded tissue decreases with increas-
`[13] Pattynama PM, De Roos A, Van der Wall EE, Van
`ing field
`strength and therefore the
`threshold of
`Voorthuisen AE. Evaluation of cardiac
`function with
`
`

`

`CMR for early diagnosis of cardiac iron overload 2179
`rrhr
`
`(28] Emst OSG, Bonvarlcet P, Canva-Delcambre V, Paris JC,
`L’Hermine C, Hepatic iron overloud: diagnosis and quantili-
`cation with MR imaging. AJR Am J Roentgenol 1997; 168:
`1205-8.
`f29] Bonkovsky HLRR, Cuble EE, Davidoff A, Rijcken TH, Stark
`DD. Hepatic iron concentration: noninvasive estimation by
`means of MR imaging techniques. Radiology 1999; 232:
`227-34.
`
`In
`(I4) Reeder SB, Faranesh AZ, Boxerman JL, McVcigh ER,
`vivo measurement of T2* and field inhomogencity maps in the
`human heart at 1.5T. Magn Reson Med 1998; 39: 988-98.
`[15] Villencuve JP, Bilodcau M, Lepage R, Cote J, Lefebvre M.
`Variability in hepatic iron concentration measurement from
`necdle-biopsy specimens. J Hepatol 1996; 25: 172-7.
`(16] Emond MJ, Bronner MP, Carlson TH, Lin M, Labbe RF,
`Kowdley KV. Quantitative study of the variability of hepatic
`iron concentrations. Clin Chem 1999; 45: 340-6,
`(30] Li D, Dhawale P, Rubin PJ, Haacke EM, Gropler RJ.
`(!7] Lorenz CH, Walker ES, Morgan VL, Klein 88, Graham TP.
`Myocardial signal response to dipyridamole and dobutamine:
`Normal human right and left ventricular muss, systolic
`Demonstration of the BOLD cflect using a double echo
`function, and gender differences by cine magnetic resonance
`gradient echo scquence. Magn Reson Med 1996; 36: £620.
`imaging. J Cardiavasc Magn Reson 1999;
`|: 7-23,
`[31] Wacker CM, Bock M, Hartlep AW ef af, BOLD-MRL in [0
`(18) Hershko C, Link G, Cabantchik I. Pathophysiology of iron
`patients with coronary artery disease: Evidence for imaging of
`overload. Ann NY Acad Sei 1998; 850: 191-201.
`capillary recruitment in myocardium supplied by the stenotic
`[19] Stark DD, Hepatic iron overload: paramagnetic pathology.
`artery, MAGMA 1999; 8: 48-54.
`Radiology 1991; 179: 333-5.
`and
`ltemochromatosis
`32] MacDonald RA, Mallory GK,
`[20] Hardy PA, Henkclman RM. Transverse relaxation rate
`hemosiderosis; study of 21] autopsied cases, Arch Intern Med
`enhancement caused by magnetic particulates. Magn Reson
`(Chicago) 1960; 105; 686-700.
`Imaging 1989; 7: 265-75.
`[33] Buja LM, Roberts WC.Iron in the heart. Etiology and clinical
`(21) Kaltwasser JPGR, Schalk KP, Hartl W. Non-invasive
`significance. Am J Med E971; 51: 209-21.
`quantitation of liver iron-overload by magnetic resonance
`imaging. Br J Hacmatol 1990; 74; 360-3,
`[34] Jessup M, Manno CS. Diagnosis and management of iron-
`induced heart disease in Cooley's ancmin. Ann NY Acad Sci
`22) Gomori JMHG, Tamary H, Zandback J ef af. Hepatic iron
`1998: 850: 242-50.
`overload: quantitative MR imuging. Radiology 1991; 179:
`367-9.
`(35] Kremastinos DT, Tiniakos G, Theodorakis GN, Katritsis
`DG,Toutouzas PK. Myocarditis in beta-thalassaemia major.
`{23) Jensen PDJF, Christensen T, Ellegaard J. Non-invasive
`A cause of heart failure. Circulation 1995; 91: 66-71.
`assessment of tissuc iron overload in the fiver by magnetic
`resonance imaging. Br J Hacmatol 1994; 87: 171-84.
`(36] Kremastinos DT, Flevari P, Spyropoulou M, Vrettou H,
`Tsiapras D, Stavropoulos-Giokas CG. Association of heart
`[24] Liu PP, Henkelman M, Joshi J er af. Quantification ofcardiac
`andtissue iron by nuclear magnetic resonance relaxometry in
`failure in homozygous beta-thalassacmia with the major
`a novel murine thalassaemia—cardiac iran overload model.
`histocompatibility complex. Circulation 1999; 100: 2074-8.
`Can J Cardiol 1996; 12: 155-64,
`{37] Economou-Pcterscn E, Acssopos A, Kladi A et af, Apolipo-
`(25} Johnston DL, Rice L, Vick W, Hedrick TD, Rokey R.
`protein E epsilon4 allele as a genetic risk factor for left
`Assessment of tissue iron overload by nuclear magnetic
`ventricular failure in homozygous beta-thalassemia. Blood
`1998; 92: 3455.9,
`Tesonance imaging, Am J Med 1989; &7: 40-7.
`[26] Mavrogeni SI, Gotsis ED, Markussis V et aif. T2 relaxation
`[38] Barosi G, Arbustini E, Gavazzi A, Grasso M, Pucci A.
`limestudy of iren overload in b-thalassemiu. MAGMA 1998:
`Myocardial
`iron grading by endomyocardial biopsy. A
`& 7-12
`clinico-pathologic study on iron overloaded patients. Eur J
`(27] Gandon Y, Guyader D, Heautot JF er af, Hemochromatosis:
`Hacmatol 1989; 42: 382-8,
`diagnosis and quantification ofliver iron with gradient-echo
`MR imaging. Radiology 1994; 193: 533-8,
`
`