•
`
`EXPERIMENT AL CARDIOLOOY
`
`Quantification of cardiac and
`tissue iron by nuclear
`magnetic resonance
`relaxometry in a novel murine
`thalassemia~cardiac iron
`overload model
`
`PETER Lfu MD, MARK HENKELMAN PhD, J ]OSHl PhD, PETER HARDY PhD, JACK BUT ANY MD, MARK lWANOCHKO MD,
`MARTIN Ci.AUBERG PhD, MADHU DHAR PhD, DARYL MAI MD, SoHAIL W AIEN MD, NANCY OLIVIERI MD
`
`Cenrre far C..ardiovascular Research, The Tor onto Hospital, Department of Medical Biophysics, Sunny brook Heahh Sciences Centre, and
`Division of Hematofcgy, Hospital for Sick Chikben, University of Toronto, Toronto, Ontario; and Department of Biochemistry , Uniw:rsicy of
`T~rme.ssee , Kmn.ale, Tennwee, USA
`Corres{xlTUknce and rt1Jrinr.s: Dr Peter Uu, Drrectm-, Cardiac NMR Re.'learch, The Toronto Hos.Pital, Toronto General Division, Room 1-508,
`Gerrard Wing, Tmmiw, Oruario MSG 2C4. Telephone 416-340-3035, fax416-340-4753, e-mailpeter.liu@utLJTrJnW.ca
`Received for publication M:t 4, I 994 . Accepted June 2, 1995
`
`CAN J CARDIOL VOL 12 No 2 FEBRUARY 1996
`
`~t (g I ~9
`Rich Gennosen CCR ~CR. CRR. RMR
`
`155
`
`TAR00073381
`
`1 of 10
`
`Taro Pharmaceuticals, Ltd.
`Exhibit 1062
`
`

`

`Liu et al
`
`~\
`
`I ron is a trace metallic element essential for cell metabo(cid:173)
`
`lism and enzyme function, and its tissue levels are precisely
`regulated. However, pathological conditions do occur where
`tissue iron overload either arises congt.'flitally or is acquired
`through excessive oral or rransfusional sources ( 1-3 ). Iron
`overload can he detrimental for cell survival and tissue func- ·
`tion, leading to free radical-mediated and other toxic dam(cid:173)
`age to cell organelles (3,4). Indeed iron overload due to
`chronic blood transfusion for congenital anemias such as
`tbalassemia:; produces iron-induced heart failure, and is the
`most common cause of cardiovascular deaths worldwide in
`the second and third decades of life (.3,5).
`The available techniques for the assessment of iron in
`vivo rely on serum chemistry. However, scrum iron and
`ferritin levels do not accurately reflect differential iron stor(cid:173)
`age levels in different organs and are limited by low repro(cid:173)
`ducibility. The only alternative at present is direct tissue
`biopsy to quantify iron chemically, which unfortunately is
`invasive and impractical for long term follow-up of chronic
`conditions.
`Nuclear magnetic resonance (NMR) rclaxometry is a new
`diagnostic technique that is sensitive to the presence of iron
`in tissues, and has the potential to assess iron distribution in
`patients and to quantify the extent of iron deposition (6, 7).
`This is especially important in the heart, where iron can be
`preferentially deposited and detennine the patient's prol,>nO·
`sis. However, the ability to quantify iron requires the estab(cid:173)
`lishment of a predictable relationship between tissue iron
`and unique NMR parameters such as relaxation times. This
`
`was limited so far by the lack of either an appropriate animal
`model of significant multiorgan iron overload or of a large
`bank of fresh human tissues for correlative studies.
`Previous animal models of iron overload have been suc(cid:173)
`cessful in loading iron into the liver and ceticuloendothelial
`~y.;tem, but very little in the heart. We have recently devel(cid:173)
`oped an improved murine mode of iron overload by parenter(cid:173)
`ally administering iron to strains of inbred mice harbouring a
`potential beta·thalassemia mutation, leading to differential
`degrees of tissue iron deposition in major organs, including
`the heart. By measuring in an NMR spectrometer the Ti and
`T2 relaxation parameters of the freshly removed organs such
`as the heart, and comparing these values with direct bio(cid:173)
`chemical analyses of tissue iron contents, we can evaluate
`whether NMR relaxometry represents a new and useful in
`vivo tool to assess tissue iron stores.
`Therefore. the purpose of our study was twofold. First, we
`sought to establish a novel murine model of iron overload
`pathologically and biochemically, including cardiac iron de(cid:173)
`position. Second, we wanted to determine the relationship
`between NMR relaxation parameters and tissue iron levels.
`
`ANlMALS AND METHODS
`Creation of a murine tbalassemia iron overload model: It
`has bt.>en established that a mutation at the beta-glohin locus
`of a DBA/2J male mouse can produce an absolute deficiency
`of normal beta-major polyp<.-ptides in a homozygous mutant.
`However, homozygous bet.a-r.halassemic mice produce
`enough beta-minor globin to stabilize the hemoglobin at 8 to
`
`156
`
`CAN J CAADIOL VOL 12 No 2 FEBRUARY 1996
`
`TAR00073382
`
`2 of 10
`
`Taro Pharmaceuticals, Ltd.
`Exhibit 1062
`
`

`

`9 g/dL (8). Hence, beta-~rnia in mice is a milder
`disease than that seen in humans, and homozygotes are fertile
`and are able to propagace this mucation to their progeny.
`Heteimygotic progeny from the original DBA/2) X
`C57B[J6J cross containing the beta-globin mutation were
`obtained from Jm::kson Laboratories (Maine}. Mice were.
`kept in plastic cages in a well ventilated, tempemture·con·
`trolled room. Fl generation from the breeding of the original
`heterozygotic parents provided the animals for this study.
`The principles of laboratory animal care as promulgated by
`the National Institute of Laboratory Animal Sciences were
`strictly observed.
`Animals were bled from the retro-orbital sinus, and elec(cid:173)
`trophoresis of cystamine·modified hemoglobin on cellulose
`acetate was used to<lli.1inguishamong homozy~ and hetero(cid:173)
`zygotic thalassemic mice and nounal wild controls. Thalas.(cid:173)
`semic homozygotes had only Hbdminor (a2P2dminor),
`homozygous normal mice had only Hb (azPll. and hetero-(cid:173)
`zvgotes had both (9).
`Twelve animals were used in this study, comprising two
`homozygotes, six heterozygotes and four normals. The ani(cid:173)
`mals were randomly assigned to be loaded with iron injected
`peritoneally on an altemate-Oay regimen (iron dextran in·
`jectable, 5 mg/day/animal, approximately 50 mg/week) for
`three consecutive weeks. The remaining animals were in(cid:173)
`jected with normal saline of equal volume to act as controls.
`The objective was to produce a wide heterogeneity of iron
`stores in this cohort of animals for model validation and NMR
`correlation. At the end of the three-week iron loading pe·
`riod, the animals were sacrificed by decortication. Liver,
`heart and peripheral gastrocnemius muscle were immedi,
`ately excised and divided such that one portion was submit(cid:173)
`ted for NMR spectrometry measurements, while the other
`portion was freeze clamped in liquid nitrogen for· future
`biochemical asses.mient.
`Pathological confirmation of iron deposition: The tisrue
`samples removed were~. embedded in paraffin block
`and sectioned for light microscopy. To delineate structure,
`the sections were stained with hematoxylin and eosin. The
`endothelial, reticuloendothelial, interstitial and intracellular
`components were specifically examined in each liver, heart
`and muscle sample. The same sections were also stained for
`iron content with Phl$ian blue.
`NMR spectroscopic measurements of tissue relaxation:
`After sacrifice, liver, heart and peripheral muscle samples
`we.re cleared of blood contents, gently blotted, and immedi·
`ately placed in sealed NMR spectrometer tubes to avoid desic(cid:173)
`cation and were submitted for NMR spectroscopic analysis of
`tissue relaxation parameters.
`The authors' group has previously characterized in detail
`the conditions for proper handling of the tissues for NMR
`spectrometty {10,11). The tissues were examined in a. vari~
`able field NMR spectrometer (Bruker Model 6225, Ger·
`many), at a locked field strength of 1.5 Tesla (proton
`resonance frequency of 63.9 MHz). This was done to petmit
`ready comparison between the observed relaxation parame~
`ters and image derived parameters found in the conventional
`
`CAN J CARDJoL VOL 12 No 2 FEBIUJAAY 1996
`
`NMR Iron quantification In a murine modet
`
`imaging magnet field strengths of around 1.5 Tesla. AU the
`mearurements were made at room temperarure of 20±1°C.
`Tt relaxation values were obtained by serial inversion
`recovery pulse sequences collecting 25 points spanning nve
`relaxation times. The Tz relaxation values were derived from
`a Carr-Purcell-Meiboom-Gill pulse sequence incorporating
`200 sample points. The pulse widths were tuned for each
`sample. Typically the transverse relaxation decay curves
`were adequately modelted with a monoexponential function
`and were reproducible on repeated measurements to ±3%.
`The relaxation values were repeated for each specimen and
`the results avemged. The correlation between the two meas(cid:173)
`ured values was excellent at 0.98, and the reproducibility
`of Tt and Tz relaxation times were ±2% and ±3%, res.pee·
`tively.
`Biochemical assays for tissue ferritio, iron and protein
`contents -Tissue~ mu! preparati.on: The col·
`lected tissue samples were kept fro7.en at-70°C until synchro(cid:173)
`nous analysis. The samples were thawed on ice and
`individually weighed. Tissues wete homogenized in freshly
`prepared Tris buffer containing 0.1 mg/mL phenylmethyl·
`sulphonyl fluoride by high energy sonication. Afrer micro(cid:173)
`centrifugation at 13,600 g for 25 mins, the supernatant was
`analyzed separately as the soluble fraction. The samples were
`subsequently analyzed both as the total sample for iron con·
`tent determinations and as the soluble fraction for ferritin
`content determination.
`Ti.uue fenitin determination: The tissue ferritin levels were
`calculated from solubilized fraction of the homogenized tis(cid:173)
`sue. Relative ferritin concentration was determined using mt
`ferritin enzyme immunoassay kits (Ramco Laboratories Inc,
`Texas). The assay for ferritin depends on an immunoenzy·
`matic reaction. where the primary antibody for ferritin is
`anchored ona solid phase. The second antibody is conjugated
`to allca1ine phosphatase, which then can react with a co1·
`orimetric substrate for quantification. A standard curve was
`first constructed using rat serum ferritin to characteri:e the
`relationship of ferritin concentration vetSUS light absorbance,
`and this showed an excellent linear relationship (r=0.98).
`This assay has high sensitivity and a wide response range, and
`is currently the best technique for ~ying a large range of
`quantities of ferritin. This .ay provides mainly a relative
`measure of ferritinconcentration against the ferritin standard
`used.
`Concentrations of various samples were determined in
`duplicate according to the observed absorbance values. All
`values were expressed as relative ferritin in mg/g wet weight
`of specimen (12).
`Elemental ironddenninadora: Total ironconcenttation was
`measured by flame atomic absorption on an IL 551 Atomic
`Absorption/Atomic Emission Specttophotometer Onstru·
`mentation Laboratory, Massachusetts). A high solids
`air/acetylene burner head and an injection sample cup were
`employed. A standard curve was also first devised for ca.lib.ra(cid:173)
`tion. before individual samples were submitted for analysis.
`Data analysis: The NMRspectrometer relaxation parameters
`of major organs of liver, heatt and muscle between iron
`
`157
`
`TAR00073383
`
`3 of 10
`
`Taro Pharmaceuticals, Ltd.
`Exhibit 1062
`
`

`

`Uuetal
`
`F'lgU.l'e l) Above Microscopic section of myvcarrlium in a heu:rocygous
`animal loaded wilh iron. The tissue was stained with Prussian blue, and
`black stained areas represent iron. Amuulant Prussian bIM posirive
`material (iron) is present, arranged in a linear manner. The heaviest iron
`deposits are in the endothelial cells (arrowhead). The pale myoftbrils
`contain a pau::hy fine sprinkling of stained granules diffusely within
`myocyre cywplasm. Prussian blue stain, x.50. Left top and bottom
`Microscopic seclioo of liver from 11 heteroeygous animal loakJ with i:ron.
`The tissue was scained with l'mssian blue, and black stained areas
`represent iron. Granular mm dep(Jsits are seen in virtually all Kupffer
`cells (dark wide arrowheads) and fmer deposits are seen in endothelial
`cells (dark &:mg aJTOOJ). The Kupffer cells show a spectrum of change ftom
`·mas~roe· deposits in dimers to sin,gk cells 'bfoo.ted' with mm. The cards
`ofhepawcytes contain a.fine sp.,.inkling of iron (light arrowhead).
`C Centt'al tiein. Prussian blue stain, x50 (lef'ttop) andxl 25 (1.e{tbottmn)
`
`1.lSed was Staristical Analysis System (SAS Institute, North
`Carolina), and P<0.0.5 was considered significant.
`
`RESULTS
`Tbalassemic murine model of iron overload: The model was
`successful as a model of overloading iron into the organs
`usually involved in transfusional iron overload. However, the
`two homozygous animals and some of the heterozygous
`animals receiving iron injections appeared lethargic. with
`a significant loss of fur colour. All homozygous and the two
`heterozygous animals with high iron loads had severe weight
`loss and evidence of 'congestive heart failure' such as paw
`edema or anasarca. Both homozygous mice died before the
`sacrifice dates, and the remaining IO animals were subjected
`to analysis.
`Pathological sections from representative animals were
`processed with hematoxylin and eosin fur structure and Prus-(cid:173)
`sUill blue stain for iron content (figure l). The noniron
`loaded animals showed normal heart and peripher.U muscle
`architecture (Figure 2). Despite the genetic bad:ground in
`the heterozygous animals, without peripher.al iron loading
`there was no evidence of iron loading in the heart.
`On the other hand, in the iron loaded animals, direct iron
`deposition wa,<; observed in the myocyte cytoplasm in the
`heart. There was also significant iron deposition in the tru:er(cid:173)
`stitium, round in endothelial and macrophage systems (Fig~
`
`CAN)CARD!OLVOL 12 Nol FEBRUARY 1996
`
`TAR00073384
`
`loaded and unloaded animals were compared with tw0~w<1y
`analysis of variance. lntragroup comty.uisons were carried out
`using Neuman-Keul's post-hoc subgroup analysis. Similar
`statistical procedure was perfonned for soluble ferritin and
`total iron concentrations. The relationship between ferricin
`and iron, and betw<.-en Tz relaxivity and iron concentration
`was detennined by line.1r regression analysis, and the corre(cid:173)
`lation coefficients were calculated. The statistical package
`
`158
`
`4 of 10
`
`Taro Pharmaceuticals, Ltd.
`Exhibit 1062
`
`

`

`NMR Iron quantification In a murlne model
`
`· 1
`··~~ I
`
`i MoarrJS~£M '
`i 'P•oOCS
`L_. ....... ---
`
`,
`
`Heart
`
`Liver
`Tis.sue of Origin
`
`Muscle
`
`100
`
`0
`
`Heart
`
`Liver
`Tissue of Origin
`
`Muscte
`
`-..-----~
`
`0 ......
`
`Figure 3) Top Typical T1 relaxation values representing animals
`treated with (Rx) or without iron, in. tissues from t1ui liver, heart and
`peripheral muscle. * P<O .05 for iron loading i:er.sus no iron loading in t1ui
`same tissue. Bottom The same T1 relaxation times fwther divided
`according to animal genotype of herero:agous ur normal with respect to
`globin gene mutation
`
`Figure 4) Top Typical Ti relaxation values representing animals be(cid:173)
`longing to different iron treamumt groups in tissues from the liver, hean
`and peripheral muscle. *P<0.05 far iron loading (Rx) versus no iron
`loading in t1ui same tissue. Bottom TM same Ti relaxation times further
`divided according to 'mimal genotype of heterozygous ar normal u.i th
`respect to globin gene mutation
`
`I
`
`1
`:.f
`' ;
`
`ure l, left bottom), and within the hepatic Kupffer, endothe(cid:173)
`lial aod macrophage cells, with less in the hepatocytes (Fig(cid:173)
`ure 1, left top and bottom). The peripheral muscle was
`relatively free from any intracellular iron, with only occa(cid:173)
`sional Prussian blue stained particles in the intersritium and
`vascular endothelium. The normal animal with iron loading
`showed mild iron deposition only in the reticuloendothel(cid:173)
`ial system, but not in the hepatocytes or myocytes. Overall,
`the liver consistently had more iron than the heart, which
`in tum contained significantly more than the peripheral
`muscle.
`Spectrometric measurements of tissue relaxation parame(cid:173)
`ters: Representative TI and T 2 relaxation values among the
`treatment groups in the different tissues are plotted on Fig(cid:173)
`ures 3 (top) and4 (top). As expected, the tissues that showed
`the most dramatic pathological evidence of iron also had the
`most severe depression of Tz relaxation, followed by Tt
`showing the same trend. Overall, the liver showed the great·
`esc changes, with marked shortening of both T1 and Tz
`values, whereas the heart showed an intermediate amount
`of iron deposition, and peripheral muscle had minimal Level
`of iron effects. The data comparing the differences accord·
`ing to the genotype of the animals in terms of rhe glob in gene
`mutation are shown in Figures 3 (bottom) and 4 (bottom) .
`The results suggested that there was more depression of
`
`I
`I !
`.!
`l !" ;
`I
`J
`
`CAN J CARDCOL VOL 12 No 2 FEBRUARY 1996
`
`T 1 relaxation in the heterozygous group than in the normal
`wild genotypes. However, due to the small sample size in·
`volved (n=3 for the heterozygotes and n=2 for the normals),
`statistical analysis is likely not very meaningful, and the
`results must be interpreted with caution. Overall, iron
`loading had the most dramatic impact on relaxation
`changes, followed by the organ involved. The genetic predis(cid:173)
`position suggested a secondary pennissive effect, which is not
`quantifiable in this study due to the small sample size in·
`volved.
`The relaxation decay curves were sufficiently monoexpo·
`nential to allow extraction of a single relaxation time, par·
`ticularlyforT1 relaxation. However, the Ti relaxation decay
`curves were more complex, containing multiple relaxation
`components. To gain more insight into the relaxation proc(cid:173)
`ess in this situation, the Tz decay curves could be examined
`using continuous distribution analysis of relaxation times
`(13). Using this analysis technique, the specific Ti compo(cid:173)
`nents can be identified, and their relative strength of contri·
`bution to the relaxation process can be compared between
`experimental groups. Typical T 2 distribution analysis for the
`liver from normal and heterozygous animals with or without
`iron loading are shown in Figure 5. The nonnal control had
`a dominant contribution from a peak with a mean Ti value
`of 40 ms. The iron loaded normal mouse showed a bimo-
`
`159
`
`TAR00073385
`
`5 of 10
`
`Taro Pharmaceuticals, Ltd.
`Exhibit 1062
`
`

`

`Liu et al
`
`~------·---··············--····
`
`0
`
`75
`
`0
`
`75
`
`50
`25
`-ra ••cl
`Het ro + F
`
`/o
`
`50
`25
`T~ W!tel
`
`75
`
`0
`
`50
`25
`T2 !eml
`
`75
`
`Figure 5) Continuous spectral plot of Ti relaxation medianfams ftam.
`the livers af four representative animals, refiecting cont:ributions from.
`iron loading and genetic makeup. According to this analysis, the norma!
`control (Cml) had a dominant contribution from a peak wiih a mean Ti
`value of 40 ms. The iron loaded (Fe) normal mouse showed a bimodal
`distribution with a dominant peak having a mean value of 11 ms. The
`iron loaded hetcroeygrms (Hetcro) mouse showed essentially r.he same
`pattern. The noniron loaded heteroeygote mouse shows a pattern of
`relaxation components similar ro that of the iron loaded animal>. There(cid:173)
`fore, the presence of either iron or of genetic abnormality leads to
`significant alterations of the Ti relaxation envelope
`
`dal distribution with a dominant peak having a mean value
`of 11 ms. The iron loaded heterozygous mouse showed essen·
`tially the same pattern. Interestingly, the noniron loaded .
`heterozygote mouse showed a similar pattern of relaxation
`components a.; the iron loaded animals. Althclugh the origin
`of these relaxation distributions are not yer fully understood,
`the patterns are indicative of the relaxation mechanism at a
`microscopic level, suggesting that the hctcrozygote is pro·
`ducing and storing iron in a manner that is different from
`that of the normal wild-type animals, even before iron load(cid:173)
`ing. The. only difference is the absolute Ti level difference,
`which is dependent on the rota! quantity of iron deposited.
`Biochemical assay of iron and forritin in tissue samples:
`Relative ferritin levels from the soluble fraction are plotted
`in Figure 6 (left) for the overall groups, and in Figure 6 (right)
`with the genotype classified into heterozygotes or normals. A
`large range of tissue ferritin levels was actually obtained,
`covering r.wo orders of magnitude. This attested to the success
`of the model in loading iron in the various organs, and the
`limited ability for the whole animal to excrete iron effectively
`when there is a large exogenous parenteral load. Similar
`results were found in total tissue iron concentration measure·
`ments (Figure 7). No statistical differences were found among
`the genotype backgrounds, but this is likely a limitation of
`the small sample sire.
`
`160
`
`"' 11 10
`QI
`-'
`t: ,u
`tf
`~ 6·
`·?5
`~ 4
`~ :1'
`
`l j :1-
`
`.
`
`Heart
`
`U~er
`Tissue ot Origin
`--·--------------
`
`Mon'1liHZM
`·p .... 0.-05
`
`Muscle
`
`"
`
`• !>f-C1&;~U:li N:.l!l'r.al
`
`M1J...<;clO
`
`.......__ ____ _
`
`'
`Figure 6) Top Tissue ferritin measurement..1 among animals tremed with
`(Rx) or withottt iron in the different organ d!st:ributions. * P<O .05 for
`iron loading versus no iron loading in die same tissue. Bottom The same
`measurements further di1dded acoording to animal genotype with respect
`to globin gene mutation
`
`A linear regression analysis was carried out between rela(cid:173)
`tive ferritin concentration and total tissue iron, and the
`relationship between the two values was shown to be linear
`(Figure 8), with r .. 0.92 and P<O.OOL This suggests that the
`iron stored in ferritin is a constant fraction of the total iron,
`independent of the degree of iron overload under r.he condi·
`tions of the experiment.
`Correlation of tissue relaxation and actual iron contents:
`Increasing iron content in tissue contributes additional mecha(cid:173)
`nisms for NMR relaxation. It is expected, therefore, that the
`Tz relaxation rate (l!fz) would be a linear function of iron
`concentration starting from some intrin~ic tissue relaxation
`rate when the iron concentration was zero. That is:
`
`( 11
`l
`df"'k(FeJ+'r
`l 2 ssue@{FO)m0
`2
`
`where k is the Tz relaxivity for iron and (1/rzhissue@ fFeJ .. o is
`the Tz relaxation rate of a particular tissue when the iron
`concentration is zero. Equation l can be rearranged to give a
`universal tissue iron dependence as follows:
`
`CA.."l J CARO!Ot. VOL l2 No 2FEBRUARY1996
`
`TAR00073386
`
`6 of 10
`
`Taro Pharmaceuticals, Ltd.
`Exhibit 1062
`
`

`

`NMR iron quantification in a murlne model
`
`'100 ;
`
`e
`
`~ 10
`
`aMu.oc;.et~
`ofieart
`.t.ivtit
`
`0.1
`
`0::
`
`. .
`
`?J
`e
`0
`u
`,§
`.,
`'E
`LI.
`~
`•
`~ O.o1 {
`I / "
`0.001 t---
`0.001
`
`:x
`
`00 .q.<J
`
`..
`.
`..
`
`~
`
`•b
`
`-11
`I
`I
`
`!
`'
`100
`
`0.01
`0.1
`10
`Total Iron Concen!ralion (mg/g)
`
`Figure 8) Li1'.ear regression analysis of tissue ferritin and iron levels,
`showing relatively linear relationship close to identit.:;i f01 the two values
`
`1000.,-~~~~~~~~~~~~....,,..,.~.........,
`
`• Must:!•
`'
`Hos.rt
`t
`tii
`100i •
`l.ivnr
`f
`
`10v~
`.. :.
`.
`..
`1 • .
`0.1 +-----.,-----,..--------,-,· --·--··J
`
`0.01
`
`10
`0.1
`Iron Concentration (mgtg)
`
`100
`
`Figure 9) Regression analysis of tissue Ti relaxivity ( 1 /Ti) and rota!
`tissue concentration from the same tissue samples. The relationship
`showed r=O. 96 and standard mO'f of estimate at 3%. Ro Intrinsic tissue
`relaxivity
`
`occurs at a gene frequency of2 to .5% in the general popula·
`tion (16). There is an excessive amount of iron absorption
`through the gastrointestinal tract in this condition, with
`iron-related damage to the heart, liver and endocrine glands
`(17). Acquired iron ooerlaad occurs as a result of repeated
`transfusions in patients with chronic anemias, or increased
`absorption of dietary iron such as in alcoholic liver disease.
`The iron accumulates in the hean, pancreas and liver, lead(cid:173)
`ing to death in the second or third decade of life in
`unchelated patients (2,5,18-20). The absolute organ iron
`levels have considerable influence in the diagnosis and natu·
`ral history of conditions of iron overload. Hepatic cirrhosis
`in acquired iron overload states develops principally when
`liver iron levels exceed 22 mg/g of tissue {21 ). Patients with
`hereditary hemochromatosis may also develop hepatomas
`when iron levels exceed 5 mg/g tissue. Therefore, the detec·
`tion of iron at lower tissue levels will be important in identi·
`fying patients at risk and potentially in monitoring the effect
`of therapy.
`In monitoring iron load, serum iron levels give an impre•
`cise estimation of tissue iron stores. Alternative modalities
`such as computed tomography imaging using conventional
`techniques to assess the degree of iron overload are too
`insensitive and inaccurate for clinical application ( 22). 0th·
`
`Figure 7) Top Tissue iron measurements among animals treated with
`(Rx) or without iron in the different organ distributions. *P<0.05 f(lr
`iron loading versus na iron loading in the same tissue. Bottom The same
`measurements further divided according w animal genot-ype with respect
`to globin gene mutation
`
`i
`f t l
`l
`r I
`
`The T2 data were analyzed versus total iron concentra·
`tion using this relationship, and a best fit to the data was
`obtained with values of {1/T2)[FeJ~o of 18, 20 and 24 s-1
`for heart, liver and muscle, respectively, and a relaxivity of
`35 s-1/mgfFe]Jg {figure 9).
`Similarly, the relationship between T1 relaxation and
`tissue iron contents was derived. A linear relationship was
`obtained, but the correlation was weaker, indicating that Tz
`is a better measure of tissue iron than is T1.
`
`:1'
`
`. •
`
`DISCUSSION
`Clinical importance of iron quantification: Iron overload in
`the body results in predictable, well described toxicity. Excess
`iron is initially stored as crystalline iron oxide within ferritin
`and hemosiderin particles. Further excess iron is present in
`cytosolic transit pools, available to be taken up by various
`susceptible organs. Iron's ultimate toxicity is due to its cata*
`lytic role in free radical production, with direct peroxidation
`of lipid membranes, mitochondria and lysozymes (3,+,H).
`Excessive iron also interferes with the function of mecalloen·
`zymes, which are essential for many of the energy production
`metabolic pathways within cells.
`Several conditions of iron overload exist, frequently asso(cid:173)
`ciated with a fatal outcome (15). Idiopathic hemochromatosis
`
`CAN J CARDIOL. Vot. 12 No 2 FEBHUARY 1996
`
`161
`
`TAR00073387
`
`7 of 10
`
`Taro Pharmaceuticals, Ltd.
`Exhibit 1062
`
`

`

`Liu et al
`
`ers have exploited the paramagnetic properties of iron in
`detecting the magnetic susceptibility of liver by means of
`superconductingqmmtum interference device (23). The lat·
`ter, unfortunately, is relatively inaccessible, useful mainly
`in extreme cases of iron overload, and cannot assess cardiac
`iron.
`NMR imaging is a rapidly evolving technique that is sen·
`sitive to local magnetic field inhomogeneities. Small iron
`particles can create significant local field inhomogeneity
`where Welter molecules come into dose proximity (24). Both
`theoretical calculation and practical observations have sug(cid:173)
`gested that a predictable relationship should exist between
`Tz relaxation rate and iron levels. Preliminary clinical stud·
`ies also support the potential of NMR to quantify iron levels
`in tissues (6). This is especially advantageous for the heart,
`where no other technique exists for this purpose. We have
`also preliminary data to suggest the ability for NMR to moni(cid:173)
`tor the efficacy of chelation agents on cardiac iron contents
`(7,25,26). However, the precise relationship between car(cid:173)
`diac iron and NMR parameters has not previously been vali(cid:173)
`dated in vivo, and experimentation in an appropriate animal
`model was necessary.
`Animal model of iron overload: Previously, good models of
`iron overload in which iron can be detected in the heart have
`not existed. Severa! previous models of iron overload using
`iron ingestion have resulted in increased iron levels in
`the reticuloendothelial system in the liver, but not in other
`organs. It has been difficult to develop an animal model
`of iron overload that closely mimics the clinical situation in
`transfused humans, in which deposition of iron within
`the myocardium determines survival in transfused thalas~
`semia major. In this study, iron was administered parentcr~
`ally, and although iron deposition was greater in the liver
`than in the heart, significant myocardial loading was ob~
`served.
`Our results suggest that iron loading in animals with
`homozygotic or heterozygotic trait for thalassemia may result
`in synergistic tissue iron overload. Both of our homozygous
`animals died following iron loading, suggesting that they arc
`most susceptible to parenteral iron effects. Furthermore,
`there was a trend for the relaxation values to be lower and
`the iron levels to be higher in the heterozygotic animals after
`iron loading. However, this study was not designed ro
`quantify the genetic contribution to tissue iron loading. The
`actual exposure to iron loading was the most instrumental
`factor in creating the large amount of iron deposition in such
`a short time span, far above the level seen wirh genetic
`pemtissiveness in the heterozygotes.
`The patrem of iron deposition is also similar to human
`disease. In human transfusional iron overload, the liver is
`frequently the most affected organ, foHowed by heart and
`bone marrow. Peripheral muscle is rarely involved in this
`process. The iron deposition generally begins in the intersti~
`tium and endothelial cell layers, followed by intracellular
`deposition (27). TI1e degree of intracellular deposition is
`likely a function of the duration of the disease. If our model
`were to extend si&'llificantly beyond three weeks, it would
`
`162
`
`likely show an even greater amount of intracellular iron
`deposition.
`Relation between T2 relaxation and tissue iron concentra~
`tion: There was a significant correlation between tissue Tz
`relaxation rate and total tissue iron concentration. As the
`tissue iron concentration increased, there was a predictable
`shortening of tissue Tz, or a linear increase in Tz relaxivity
`(l{fz). The linear relationship between Tz rdaxivity and
`iron content of the tissues was maintained over a large
`dynamic range of iron values. To detecr a real change in Ti
`values clinically, the lower limit is likely to be 0.1 mg/g.
`Similarly, as iron concentration increases to above 20 mg/g,
`relaxlvity begins to be saturated, In this situation iron is
`causing such rapid Tz decay that it is unlikely for current
`instruments t.o detect signifkant vatiation of iron concentra(cid:173)
`tion.
`This confirms rhe known behaviour of iron in biological
`tissues to alter relaxation times. The presence of iron mole·
`cules disturbs the local magnetic field and shortens T:z relaxa·
`tion through an inhomogeneiry effect. This effect
`is
`concenmnion-depcndent, as was also shown by previous
`experiments using particulace iron (28). The relaxation
`enhancement goe.s beyond the traditional dipole-dipole in·
`teraction of paramagnetic molecules. Transverse relaxa·
`tion is specially enhanced by translational diffusion of water
`through microscopic magnetic field gradients. However, this
`relationship conrinu<'~~ to be a good model for understanding
`iron behaviour in biological tissues where the iron appears to
`be packaged into ferritin and is deposited into both intersti(cid:173)
`tial and intracellular compartments. The proportion of iron
`packaged into ferritin molecules was constant in our model,
`further supporting that the use of NMR derived relaxation
`parameters can be used to monitor important iron kinetics.
`Most important, these results suggest that it is valid to use
`tissue T z relaxation parameters as a measure of total tissue
`iron content. This is a potentially useful new noninvasive
`tool to answer important biological quesdons in the future.
`Since the ability to detect iron effect reliably fa likely limited
`when the tis.sue concentrations of iron or ferritin is less than
`1 mg/g, this may not be a particularly sensitive technique to
`detect minute quantities of tissue iron. However, as the iron
`concentration increases from this level, there is a readily
`detectable decrease in tissue T2, and this increases in parallel
`t.o lffz.
`Conversely, in