Assessment of Tissue Iron Overload by
`Nuclear Magnetic Resonance Imaging
`
`DONALD L. JOHNSTON. M.D.. LAWRENCE RICE. M.D., G. WESLEY VICK, III, M.D., Ph.D., THOMAS D.
`HEDRICK, M.D., ROXA~N RO~EY, M.D. HOE&,
`T&
`
`iron
`PURPOSE The ability of stored intracellular
`to enhance magnetic susceptibility forms the basis
`by which tissue iron can be detected by nuclear
`magnetic resonance (NMR)
`imaging. We used this
`technique to assess myocardial, spleen, and liver
`iron content in patients with known or suspected
`iron overload disorders.
`SpinechoNMRimages
`PATIENTSANDMJ~THODS
`were obtained in 30 patients; 20 had chronic ane-
`mias treated by multiple blood transfusions, five
`had idiopathic hemochromatosis, and five had non-
`hemochromatotic liver disease with elevated serum
`ferritin
`levels and no stainable iron on liver biopsy.
`The acquisition of oblique images through the
`short axis of the left ventricle permitted assess-
`ment of left ventricular
`functioni while demon-
`strating the liver and spleen on the same image.
`Iron content was assessed using a signal intensity
`ratio of organ (spleen, liver, or myocardium) to
`skeletal muscle.
`RFZXJLTS: In patients with multiple blood transfu-
`sions, iron content was highest in liver, followed by
`the spleen. Signific,ant iron overload was detected
`.in the myocardium of only one patient. Left ven-
`tricular systolic wail thickening was normal in pa-
`tients receiving multiple blood transfusions. Two
`patients with treated idiopathic hemochromatosis
`had norma.l signal intensity ratios, and three un-
`treated patients had evidence of significant depos-
`its of iron in the liver and spleen as indicated by a
`reduction in signal intensity ratios (0.2 f 9.01 and
`0.9 f 0.01, respectively). Rive patients with non-he-
`mochromatotic liver disease and high serum ferri-
`tin levels had normal signal intensity ratios by
`NlllR imaging.
`CONCLUSION: NMR imaging is a useful method of
`detecting tissue iron and distinguishing disease due
`to iron overload. Myocardial
`iron deposition is a
`late event, occurring after accumulation of iron in
`the spleen and liver.
`
`of Medicine,
`Department
`and Hematology.
`of Cardiology
`the Sections
`From
`and The
`of Radiology,
`Baylor
`College
`of Medicine
`the Department
`and
`be ad-
`Houston,
`Texas. Requests
`for
`reprints
`should
`Methodist
`Hospital,
`L. Johnston,
`M.D., Section
`of Cardiology,
`Baylor College
`dressed
`to Donald
`of Medicine,
`The Methodist
`Hospital,
`6565 Fannin. MS 041. Houston,
`Texas
`77030. Manuscript
`submitted
`November
`1, 1988,
`and accepted
`in revised
`form April 17, 1989.
`
`40
`
`July
`
`1989
`
`The American
`
`Journal
`
`of Medicine
`
`Volume
`
`87
`
`P atients with iron overload syndromes often exhib-
`
`it clinical evidence of myocardial and hepatic in-
`volvement that, if untreated, terminates in fulminant
`cardiac or hepatic failure [l]. To prevent the develop-
`ment of iron-induced organ failure, it is necessary to
`make an early diagnosis of tissue involvement. How-
`ever, it is not currently practical to assess iron content
`noninvasively, and tissue biopsy is required.
`Nuclear magnetic resonance (NMR)
`imaging is a
`totally noninvasive technique that has been previously
`used to detect iron iri the liver and brain [2-71. Iron-
`containing substances like ferritin and hemosiderin
`become strongly magnetized when placed in a magnet-
`ic field [8-lo]. This magnetization is quantified by the
`magnetic susceptibility, the ratio of the induced over
`the applied magnetic field. Localized regions of in-
`creased magnetic susceptibility selectively shorten re-
`laxation time Ts by creating regions of magnetic field
`non-uniformity. As water molecules diffuse through
`these regions, there is irreversible spin dephasing, and
`enhancement of the Ts relaxation rate. Thus, follow-
`ing a radiofrequency pulse, recovery of transverse.
`magnetization results in decreased signal intensity on
`spin echo images.
`The aim of this study was to assess iron content of
`myocardium, spleen, and liver in patients with known
`or suspected iron overload disorders. A ratio of organ-
`to-skeletal muscle signal intensity was derived from
`NMR images to assess tissue iron content. Since exces-
`sive myocardial iron has a detrimental effect on cardi-
`ac function, myocardial thickening fractions were also
`derived from the NMR images.
`
`PATIENTS AND METHODS
`Patient Population
`We studied 20 patients (nine females, 11 males) un-
`dergoing long-term blood transfusions for a variety of
`hematologic diseases. These included: sickle cell ane-
`mia (seven patients), end-stage renal disease (seven
`patierits), myeloproliferative disorders (four patients),
`aplastic anemia (one patient), and sideroblastic ane-
`mia (one patient); three patients had prior splenecto-
`my. Patients had received 82 f 58 (mean f SD) units
`of blood (total iron burden approximately 20 g) over
`one to nine years. Seven patients received less than 50
`units (33 f 12; range, 17 to 49), and 1.3 received 50 or
`more units (118 f 55; range, 50 to 220). The mean age
`was 46 years and 48 years for patients receiving less
`than 50 units and 50 or more units, respectively. Ten
`patients who did not receive transfusions were stud-
`ied. Five had idiopathic hemochromatosis, two of
`whom recently completed a course of therapeutic
`phlebotomies. Five other patients had hepatic disease
`with an elevated serum ferritin
`level, but no increase
`in stainable iron on liver biopsy. Eighteen normal sub-
`jects (mean age, 43 f 18 years) had NMR images ob-
`
`Apotex Tech.
`Ex. 2031
`
`

`

`NMR
`
`IMAGING
`
`OF IRON OVERLOAD
`
`/ JOHNSTON
`
`ET AL
`
`for
`ratios
`intensity
`to determine normal signal
`tained
`heart, liver, and spleen. Fourteen of these subjects had
`images of the heart obtained at end-systole
`and end-
`diastole
`to derive values for normal cardiac
`function.
`One transfusion
`recipient died eight weeks
`following
`NMR
`imaging, and the heart,
`liver, spleen, and skele-
`tal muscle were excised and imaged post-mortem.
`Tis-
`sue iron was measured
`in this patient quantitatively
`by flameless atomic absorption
`spectrometry.
`In all
`other patients undergoing
`tissue biopsy,
`the material
`was stained
`for iron and qualitatively
`assessed.
`
`NMR Imaging Technique
`Gated, spin echo images were obtained using a Sie-
`mens magnet operating at 0.5 tesla. All data were ac-
`quired
`in a 128 X 128 matrix using two signal acquisi-
`tions. Time
`to echo (TE) was 35 msec, and the pulse
`repetition
`time (TR) was equal to the heart rate (mean
`heart
`rate = 71 beats/minute).
`This echo time pro-
`duced a reasonable signal-to-noise
`ratio, but reduced
`the amount of blood flow signal normally associated
`with shorter echo times. To obtain short-axis
`images, a
`line was placed along the longitudinal
`axis of the left
`ventricle passing
`from
`the apex to the aortic outflow
`tract, and an angle was measured
`relative
`to the hori-
`zontal plane
`[ll]. Since the Siemens
`imaging system
`did not permit
`imaging of more than one angle, a sec-
`ond angle was chosen by placing
`the patient
`in a 30”
`right anterior oblique position. A distance was mea-
`sured between
`the base of the left ventricle and the
`apex to determine
`the length of the ventricle. With
`knowledge of ventricular
`length, it was possible
`to dis-
`tribute basilar, mid, and distal slices evenly over the
`left ventricle.
`Images were obtained at end-diastole
`and end-systole
`using the imaging protocol described
`in Figure
`1.
`
`Signal Intensity Measurements
`were made from a
`Signal
`intensity measurements
`single image using a region of interest placed over the
`heart, spleen,
`liver, and skeletal muscle
`(latissimus
`dorsi). The slice chosen
`for these measurements was
`the mid-ventricular,
`end-systolic,
`short-axis
`image. In
`two studies,
`the latissimus
`dorsi was not well visual-
`ized on this slice, and measurements
`of muscle signal
`intensity were made
`from
`the basilar
`image of the
`same acquisition. To measure myocardial signal inten-
`sity, a transparent
`plastic sheet with six segments
`(an-
`terior, anterolateral,
`posterolateral,
`inferior,
`inferior
`septum, and anterior
`septum) equally separated by
`60” was placed over a magnified
`image of the heart.
`The mid-septum
`was
`identified as falling equal dis-
`tance between
`the insertion of the superior and inferi-
`or right ventricular
`walls and was considered
`0”. A
`circular
`region of interest was chosen to measure signal
`intensity
`for each of the six segments, and a single
`signal intensity
`value was obtained by averaging
`the
`segment values. For liver and spleen, the circular
`re-
`gion of interest covered
`the largest possible area lying
`within
`the imaged organ. For skeletal muscle,
`three
`circular
`regions encompassing as much of the tissue as
`possible were averaged. The skeletal muscle was used
`as an internal standard
`to form a ratio with myocardi-
`urn (myocardium/skeletal
`muscle),
`liver (liver/skeletal
`muscle), and spleen (spleen/skeletal muscle). A typi-
`cal image demonstrating
`the location of the organs
`in
`the oblique plane is shown
`in Figure
`2.
`
`1st Acqkition
`
`c
`
`SLICE
`
`&
`
`Short Axis
`A
`
`(A) Basilar
`
`(B) Mid
`
`(C) Distal
`
`Longitudinal Axis
`
`the
`from
`identified
`images was
`of the short-axis
`1. Position
`Figure
`For
`the acquisition
`the
`figure.
`image
`shown
`at
`the
`top of
`coronal
`the R wave
`of
`the
`here,
`the basilar
`slice was obtained
`on
`shown
`electrocardiogram
`(end-diastole),
`and
`the distal slice was obtained
`on
`the downslope
`of the T wave
`(end-systole).
`The mid-ventricular
`slice was obtained midway
`between
`diastole
`and systole. The order of
`the second
`and
`third acquisitions
`was
`then
`rotated
`(2. 3, 1; 3, 1. 2).
`The mid, end-systole
`slice was subdivided
`into six segments
`for
`the
`purpose
`of assessing myocardial
`signal
`intensity.
`The basilar, mid,
`and distal slices were similarly
`subdivided
`to measure wall
`thickening.
`When
`time permitted,
`images were obtained
`at end-systole
`and end-
`diastole
`at
`three
`levels
`in the
`longitudinal
`axis of the
`left ventricle
`(bottom).
`A = anterior;
`AL = anterolateral;
`PL = posterolateral;
`I =
`inferior;
`IS =
`inferior
`septum; AS = anterior
`septum.
`
`Assessment of Cardiac Function
`the
`from
`Left ventricular
`function was determined
`three NMR
`images passing
`through
`the basilar, mid,
`and distal portion of the left ventricle. The epicardial
`contour,
`including
`the right side of the interventricu-
`lar septum and site of right ventricular
`insertions, and
`the endocardial contour, excluding
`the papillary mus-
`cles of the end-systolic
`and end-diastolic
`images, were
`traced onto a digitizing board. The center of geometry
`was a computer-derived,
`floating endocardial centroid
`from end-diastole
`to end-systole
`[12]. Each of the
`three slices was subdivided
`into six segments
`corre-
`sponding
`to the six segments used to derive the signal
`intensity measurements. The mid septum was labelled
`0”, and for each segment,
`five radii were placed
`in
`systole and diastole
`for a total of 30 radii per slice. The
`percent myocardial
`thickening
`fraction was deter-
`mined as change in wall thickness
`from end-diastole
`to
`
`July 1989
`
`The American
`
`Journal
`
`of Medicine
`
`Volume
`
`87
`
`41
`
`Apotex Tech.
`Ex. 2031
`
`

`

`NMR
`
`IMAGING
`
`OF IRON OVERLOAD
`
`/ JOHNSTON
`
`ET AL
`
`-”
`
`ron ovc
`the
`showing
`iron overload
`in a patient with
`image
`FlgurPP:Short-axis
`and skeletal muscle
`in the
`spleen, myocardium.
`liver,
`location
`of the
`As demonstrated
`by this
`image,
`signal
`intensity
`of
`oblique orientation.
`reduced more
`than
`that of the spleen or myocar-
`the
`liver was usually
`dium. This was observed
`in patients with both hemochromatosis
`and
`multiple
`blood
`transfusions.
`Since
`iron deposition
`is minimal
`in skele-
`tal muscle
`in iron overload
`conditions.
`the
`latissimus
`dorsi was used
`as an internal
`standard
`for assessing organ
`signal
`intensity.
`RV = right
`ventricle;
`LV = left ventricle.
`
`I
`TABLE
`Signal Intensity Ratios of Patients with Idiopathic Hemo-
`chromatosis or Liver Disease
`
`Cardiac/
`Skeletal
`Muscle
`
`Signal
`
`Intensity Ratios
`Liver/
`Skeletal
`Muscle
`
`Spleen/
`Skeletal
`Muscle
`
`1.40f0.24
`
`1.5f0.14
`
`1.9 f 0.41
`
`1.3 f 0.02
`
`0.2f0.01’
`
`0.9f0.01’
`
`1.3f0.7
`
`l.Of0.6
`
`1.9fO.l
`
`1.4f0.2
`
`1.4f0.5
`
`1.8 f 0.7
`
`Normal subjects
`(n = 18)
`Hemochromatosis-
`untreated
`(n = 3)
`Hemochromatosis-
`treated
`(n = 2)
`Liver disease not
`due to iron
`overload
`(n = 5)
`
`compared wrlh normal subjects.
`p <O.OOl.
`diastole and systole by making a single measurement
`from
`the anterior
`segment of the mid-left
`ventricular
`slice. Regional wall motion abnormalities were deter-
`mined by visually assessing wall
`thickening
`at end-
`systole on the basilar, mid, and distal NMR
`images.
`
`Statistics
`chamber di-
`left ventricular
`ratios,
`Signal intensity
`mensions, and wall thickening
`for normai subjects and
`patients with
`iron overload were compared using
`the
`unpaired Student’s
`t-test. The
`level of significance
`was set at 5%. Intraobserver
`and interobserver
`variabi-
`lities for signal intensity
`ratio measurements were de-
`termined
`from
`images of 10 patients with
`iron over-
`load. The intraobserver measurements were made one
`week apart. Variability was considered
`the mean dif-
`ference and standard deviation of the differences
`be-
`tween paired measurements
`of signal intensity
`ratios.
`
`(C, 1 /
`
`C/SK
`
`SPlSK
`
`L/SK
`
`for heart
`ratios
`intensity
`signal
`between
`gure 3. Relationship
`the number
`of blood
`and
`(L/SK)
`and
`liver
`SK),
`spleen
`(SP/SK).
`in normal
`subjects,
`a
`ratios
`with
`Compared
`transfusions
`received.
`in
`liver and spleen
`signal
`intensity
`ratios
`oc-
`significant
`reduction
`receiving multiple
`blood
`transfusions.
`Signal
`inten-
`curred
`for patients
`sity
`ratios
`decreased
`as the number
`of transfusions
`increased
`(less
`than 50 units versus
`50 units or more).
`C = cardiac;
`SK = skeletal
`muscle; SP = spleen; L = liver; TX = transfusions.
`* = p <0.05;
`t = p
`<O.Ol;
`* = p <O.OOl,
`compared
`with normal
`values.
`
`RESULTS
`Normal Signal Intensity Measurements
`Signal
`intensity
`ratios obtained
`for heart, spleen,
`and liver were 1.41 f 0.24,1.92
`f 0.41, and 1.54 f 0.14,
`respectively.
`Intraobserver
`variability
`for heart,
`spleen, and liver signal
`intensity
`ratios were 0.08 f
`0.05, 0.07 f 0.08, and 0.08 f 0.06, respectively.
`Inter-
`observer
`variability
`for heart, spleen, and liver signal
`intensity
`ratios were 0.14 f 0.12,0.13
`f 0.12, and 0.13
`f 0.13, respectively.
`
`Patients Receiving Multiple Transfusions
`Signal
`intensity
`ratios obtained
`for heart, spleen,
`and liver were 1.23 f 0.22,0.97
`f 0.5, and 0.39 f 0.24,
`respectively
`(p = NS, p <O.OOl; p CO.001; respectively,
`compared
`to normal values). These
`ratios
`indicated
`that the largest amount of iron was present
`in the liver.
`When
`the patients were divided
`into
`those who
`re-
`ceived less than 50 units of blood (n = 7) and those who
`received more than 50 units of blood (n = 13), myocar-
`dial, spleen, and liver signal
`intensity
`ratios were all
`lower
`for the latter group (Figure
`3). Individual
`val-
`ues lower
`than 2 SD below normal
`for heart, spleen,
`and liver (0.92, 1.08, and 1.22, respectively) were con-
`sidered abnormal. For patients
`receiving
`less than 50
`units of blood, myocardial,
`spleen, and liver signal in-
`tensity
`ratios were abnormal
`in zero, three, and seven
`
`Apotex Tech.
`Ex. 2031
`
`X 100. Myo-
`thickness
`wall
`end-systole/end-diastolic
`cardial
`thickening
`fractions were summed
`for the 30
`radii of each slice. The three slices were
`then summed
`(90 radii)
`to give a mean value for each patient. Left
`ventricular
`shortening
`fraction was determined
`as
`end-systolic
`chamber dimension/end-diastolic
`cham-
`ber dimension X 100. Wall thickness was measured
`in
`
`42
`
`July 1989
`
`The American
`
`Journal
`
`of Medicine
`
`Volume 87
`
`

`

`NMR
`
`IMAGING
`
`OF IRON OVERLOAD
`
`/ JOHNSTON
`
`ET AL
`
`liver disease and elevated
`with
`in a patient
`images
`5. NMR
`Figure
`ratios were within normal
`limits
`ferritin
`level. All signal
`intensity
`serum
`iron was visualized
`on
`tissue
`on
`the NMR
`scan,
`and no stainable
`obtained
`from
`liver biopsy. See Figure 4 for abbreviations.
`
`levels (mean = 1,348 f 743
`ferritin
`had elevated serum
`ng/mL) with evidence of liver disease biochemically as
`well as on liver biopsy. None of these patients
`had
`abnormal
`liver, spleen, or myocardial
`signal
`intensity
`ratios.
`
`Patients Undergoing Tissue Biopsy
`un-
`Five patients with multiple blood transfusions
`derwent
`liver biopsy and all had
`large amounts of
`stainable
`iron visualized. Consistent with
`the biopsy
`findings, mean liver signal intensity was markedly
`re-
`duced
`(0.26 f 0.15) for
`these
`five patients. All five
`patients with non-hemochromatotic
`liver disease and
`elevated serum
`ferritin
`levels underwent
`liver biopsy,
`and none had iron detected
`(Figure
`5). One patient
`with decreased spleen and liver signal intensity
`ratios
`and a normal myocardial
`signal intensity
`ratio under-
`went
`right
`ventricular
`endocardial
`biopsy, and no
`stainable
`iron was present.
`In vitro signal intensity
`ratios obtained from the organs of the patient who
`died following NMR imaging correlated well with in
`uiuo values (Figure 6).
`
`Left Ventricular Function
`To further document that the hearts of patients
`receiving multiple blood transfusions were not signifi-
`cantly involved by the process of iron loading, left
`ventricular
`function was examined using the NMR
`images obtained in the short axis (Figure 7). If mea-
`surement of signal intensity ratios underestimated the
`degree of iron loading, left ventricular
`function might
`be impaired and compensatory muscle hypertrophy
`might be expected [13]. There were no significant dif-
`ferences between patients with multiple transfusions
`and normal subjects in any functional parameters
`(Table II). When these parameters were compared in
`patients receiving a transfusion with either less than
`50 units or 50 units or more, a trend emerged towards
`worsened function in the latter group; however, this
`
`July 1989
`
`The American
`
`Journal
`
`of Medicine
`
`Volume
`
`87
`
`43
`
`Apotex Tech.
`Ex. 2031
`
`hemo-
`idiopathic
`with untreated
`in a patient
`images
`4. NMR
`Figure
`Top, end-diastole;
`bottom,
`end-systole.
`The end-dia-
`chromatosis.
`wall
`thickness measured
`15 mm,
`indicating
`left ven-
`stolic, anterior
`tricular
`hypertrophy.
`The
`inferoposterior
`wall was
`thinned
`in diastole
`and
`failed
`to thicken
`normally
`in systole,
`consistent
`with an old myo-
`cardial
`infarction.
`A small posterior
`pericardial
`effusion
`was present
`(solid arrow).
`The right ventricle
`was surrounded
`by fat (open arrow).
`Signal
`intensity
`values
`for myocardium,
`spleen
`(S), and
`liver (L) were
`1.3. 0.94,
`and 0.25,
`respectively.
`
`receiving 50 or
`For patients
`respectively.
`patients,
`more units,
`ratios were abnormal
`in one, 11, and 13
`patients,
`respectively.
`Serum
`ferritin measurements
`for patients with multiple
`transfusions
`did not corre-
`late significantly
`with
`liver signal
`intensity
`ratios.
`
`or Other Liver Disease
`Patients with Hemochromatosis
`The three patients with untreated
`idiopathic hemo-
`chromatosis
`had decreased spleen and liver, but nor-
`mal myocardial
`signal
`intensity
`ratios
`(Table
`‘I; Fig-
`ure
`4). For
`the
`two
`patients
`who underwent
`in-
`therapeutic
`phlebotomies
`prior
`to imaging, signal
`tensity
`ratios were normal
`for all organs. Five patients
`
`

`

`NMR
`
`IMAGING
`
`OF
`
`IRON OVERLOAD
`
`/JOHNSTON
`
`ET AL
`
`left) and in vitro
`(top
`transfusions
`blood
`in a patient who received multiple
`image of iron overload
`the in viva NMR
`between
`Figure 6. Correlation
`was placed
`in
`(arrow)
`Skeletal muscle
`(top
`right),
`spleen
`(bottom
`left), and
`liver (bottom
`right).
`NMR
`images at autopsy
`showing myocardium
`were as follows:
`(values
`in parentheses)
`the image plane and served
`as an internal
`reference.
`Signal
`intensity
`ratios obtained
`in vivo and in vitro
`myocardium
`= 0.84
`(0.81);
`spleen = 0.74
`(0.89);
`liver = 0.16
`(0.21).
`All values were 2 SDS below normal.
`Quantitative
`iron measurements
`were as follows
`(normal
`for
`liver = 530 to 900 pg/g
`dry weight): myocardium
`= 2,272 pg/g
`dry weight;
`spleen = 9,617 pg/g
`dry weight;
`liver =
`28,541
`pg/g
`dry weight.
`The smaller
`decrease
`in myocardial
`signal
`intensity
`corresponded
`to the smaller
`amount
`of iron measured
`quantita-
`tively
`in the heart.
`In the excised
`heart
`image,
`endocardial
`signal
`intensity
`was greater
`than
`in the epicardium.
`This may have been due
`to the
`preferential
`iron deposition
`in the epicardium
`(decreased
`T2). or myocardial
`edema
`in the endocardium
`(increased
`T2) in association
`with
`the
`terminal
`illness, cardiogenic
`shock. Due to the effects of motion
`on image quality,
`it was not possible
`to clearly delineate
`differences
`in signal
`in-
`tensity
`between
`the epicardium
`and endocardium
`in vivo.
`
`II). Two pa-
`(Table
`significant
`was not statistically
`had regional wall
`transfusions
`tients with multiple
`detected visually, one of whom
`motion abnormalities
`had a previous history of myocardial
`infarction.
`
`COMMENTS
`Iron by NMR imaging
`Detection of Tissue
`The present study demonstrates
`the-value of spin
`echo NMR
`imaging
`for assessing
`iron overload
`in the
`myocardium,
`liver, and spleen of humans.
`It also
`shows
`that an approximate
`estimate of iron stores can
`be made by comparing
`the signal intensity of the tissue
`with
`that of skeletal muscle, a structure
`that contains
`very little iron in iron overload states
`[13]. For patients
`
`the deposition
`receiving multiple blood transfusions,
`compared with
`of myocardial
`iron was
`insignificant
`the amount of iron present
`in the liver and spleen.
`Myocardial
`iron deposition
`is most likely a late event
`in such patients. This was not unexpected
`since the
`reticuloendothelial
`system preferentially
`accumulates
`iron from breakdown
`of transfused
`red blood cells be-
`fore deposition of iron occurs
`in the parenchymal
`cells
`of organs such as the heart. The results of another
`study examining hearts
`from autopsy
`subjects with
`iron overload of diverse origin were consistent with our
`findings
`[13]. These authors
`found
`that cardiac
`iron
`was always accompanied by heavy iron deposits
`in the
`liver and spleen, indicating
`that iron deposition
`in the
`
`44
`
`July 1989
`
`The American
`
`Journal
`
`of Medicine
`
`Volume
`
`87
`
`Apotex Tech.
`Ex. 2031
`
`

`

`NMR
`
`IMAGING
`
`OF IRON OVERLOAD
`
`/ JOHNSTON
`
`ET AL
`
`transfusions.
`(50)
`due to multiple
`iron overload
`in a patient with
`function
`of left ventricular
`7. Assessment
`Figure
`(bottom
`left) and end-systolic
`longitudinal
`axis, end-diastolic
`(bottom
`(top
`right)
`images;
`left) and end-systolic
`and normal
`images
`only) was normal
`with normal
`wall
`thickening
`from
`the short-axis
`function
`(determined
`for myocardium,
`spleen, and
`liver,
`respectively.
`The ellipsoid-shaped
`intensity
`ratios were 1.19,0.48,
`and 0.25
`short-axis
`images
`is an artifact
`that appeared
`in all obliquely
`oriented
`images obtained
`the Siemens
`imaging
`
`with
`
`(top
`end-diastolic
`Short-axis
`right)
`images. Left ventricular
`chamber
`dimensions.
`Signal
`structure
`at the bottom
`of the
`system.
`
`TABLE
`
`II
`
`left Ventricular
`
`Function
`
`in Patients
`
`with
`
`Iron Overload due to Multiple Blood Transfusions
`
`I
`
`1
`
`End-diastole (mm)
`End-systole (mm)
`Wall thickening (%)
`Fractional
`shortening
`
`(%)
`
`Normal Subjects
`(n = 14)
`
`9.6 f 0.67
`14.2 f 1.3
`0.42 f 0.07
`-
`
`Wall Dimensions
`
`Patients
`
`<%I Units
`(n = 7)
`
`10.0 f 1.4
`16.1 f 2.1 0
`0.38 f 0.19
`-
`
`>50 Units
`(n = 13)
`
`Il.lf2.6
`15.2 f 2.4
`0.32f0.11
`-
`
`Ventricular Chamber Dimensions
`Patients
`
`Normal Subjects
`(n = 14)
`
`48.0f8.1
`31.5f
`
`0.35;
`
`7.8
`
`0.1
`
`<50 units
`(n = 7)
`
`54.0 f 5.2
`33.3 f 3.4
`
`0.38:
`
`0.1
`
`>50 Units
`(n = 13)
`
`50.9 f 7.9
`34.4 f 10.3
`
`0.33:
`
`0.2
`
`I
`
`I
`
`July 1989
`
`The American
`
`Journal
`
`of Medicine
`
`Volume
`
`87
`
`45
`Apotex Tech.
`Ex. 2031
`
`

`

`NMR
`
`IMAGING
`
`OF IRON OVERLOAD
`
`/ JOHNSTON
`
`ET AL
`
`heart does not occur until the other organs are saturat-
`ed with
`iron.
`been used to de-
`Spin echo imaging has previously
`tect iron overload of the liver [2-4]. Other studies have
`reported
`the results of NMR
`imaging following
`intra-
`cranial bleeding
`[6,7,14,15]. Signal intensity of the he-
`matoma was decreased
`to varying degrees, depending
`on the time lapse following
`the acute event, and pre-
`sumably
`the amount of deoxyhemoglobin, methemo-
`globin, or hemosiderin
`present. The use of a gradient
`echo pulse sequence has been shown
`to enhance detec-
`tion of intracranial
`hematomas
`[16]. Other studies of
`the normal brain have demonstrated
`discrete
`regions
`of decreased signal intensity on Ts-weighted
`spin echo
`images, and these regions have been noted in postmor-
`tem studies
`to correspond with normal brain
`iron con-
`tent
`[5]. Measurement
`of in vitro Tz relaxation
`times
`of liver samples of iron-overloaded
`rats
`[4] and spleen
`samples of patients with
`thalassemia
`[17] demon-
`strated a linear correlation
`between
`relaxation
`rates
`(l/T,)
`and iron content. What contribution
`hemosid-
`erin and ferritin make
`to Tz relaxation
`is un-
`in
`uiuo
`clear. In a preliminary
`report, Gomori and Grossman
`[18] have suggested
`that the coarser cellular
`field gra-
`dients caused by lysosomal distribution
`of hemosider-
`in may account
`for much of the Ts proton
`relaxation
`enhancement
`in iron overload states
`[18].
`A signal
`intensity
`ratio was used
`in the present
`study
`to provide a semi-quantitative
`measurement of
`tissue
`iron content. The validity of this measurement
`was substantiated
`by comparing organ signal intensity
`with
`the number of transfusions
`received. Thus, signal
`intensity of liver, spleen, and myocardium was most
`decreased
`in patients
`receiving 50 or more units of
`blood and less decreased
`in patients
`receiving
`less than
`50 units, compared with
`that
`in normal subjects. A
`closer relationship
`between
`tissue signal intensity
`ra-
`tios and the amount of blood transfused was not ob-
`served
`for at least two
`reasons. First,
`iron loss in pre-
`menopausal
`patients
`and
`in patients with
`occult
`gastrointestinal
`or urogenital bleeding may decrease
`tissue
`iron content
`relative
`to the amount of blood
`received. Second,
`iron may be chronically
`hyperab-
`sorbed with some types of anemia
`[19], thus altering
`the relationship
`between
`tissue signal
`intensity
`and
`iron content.
`levels do not
`ferritin
`that serum
`It is well known
`accurately
`reflect body iron content. Variations
`in se-
`rum ferritin
`correspond mainly
`to change in reticulo-
`endothelial
`storage
`iron and not parenchymal
`iron
`content
`[20]. Also, it is well known
`that serum
`ferritin
`is influenced by the activity of liver disease, inflamma-
`tion, and infection. Finally, assay methods
`for ferritin
`are subject
`to artifactual
`discrepancies
`at very high
`serum
`levels
`[21]. Thus,
`the
`lack of correlation
`be-
`tween
`liver signal
`intensity
`ratios and serum
`ferritin
`levels
`in the present study was not unexpected, and
`changes
`in NMR
`tissue signal intensity
`should prove
`more accurate
`for quantifying
`organ
`iron content.
`Ts relaxation
`times were not used
`in this study,
`since
`the calculation
`of these values
`is imprecise.
`Moreover,
`it may not be possible
`to calculate Ts values
`from liver and spleen due to very low signal intensity,
`similar
`to background
`noise. In future, however, quan-
`titative NMR
`imaging measurements
`of tissue
`iron
`content may be possible by applying
`techniques
`simi-
`lar to that described by Wismer et al [22] in which a
`
`46
`
`July 1989
`
`The American
`
`Journal
`
`of Medicine
`
`Volume 87
`
`of the chemical shift method of Dixon
`generalization
`[23] was used to measure
`resonance
`linewidths
`from
`subjects
`in ho. A monotonic
`increase
`in linewidth
`was noted as iron concentration
`increased.
`Controversy
`still exists as to the precise relationship
`between
`the quantity of iron present
`in the myocardi-
`urn and the degree of cardiac dysfunction.
`Some au-
`thors have reported no relationship between quantita-
`tive myocardia1
`iron measurements
`and the presence
`of heart failure
`[24], whereas others have found a ten-
`dency for function
`to deteriorate
`in those patients with
`the
`largest number of transfusions
`[25-281. Schafer
`and co-workers
`[27] noted
`that
`three of five patients
`undergoing heavy transfusion
`(mean = 168 units) had
`an increase in end-systolic
`left ventricular
`dimensions.
`Using echocardiography
`to evaluate
`left ventricular
`function
`in patients with
`idiopathic hemochromatosis,
`Candell-Riera
`et al [28] showed
`that therapeutic phle-
`botomies
`improved
`parameters
`of
`left ventricular
`function. Henry and co-workers
`[25] noted a signifi-
`cant increase
`in left ventricular
`end-diastolic
`dimen-
`sion, wall thickness, and mass in 56 patients at risk
`for
`myocardial
`iron deposition, most of whom had re-
`ceived repeated blood transfusions
`for congenital ane-
`mias. It was not possible, however,
`to be certain of the
`relative contribution
`of intravascular
`volume overload
`(in response
`to the chronic anemia) versus
`the effects
`of possible
`iron deposition
`in the myocardium. Our
`study,
`in which patients
`received an average of 82 f 58
`units, was consistent with
`the likelihood
`that cardiac
`function
`is generally preserved
`in patients who receive
`less than 100 units of blood.
`from
`thickness
`Except
`for one report
`[29], wall
`NMR
`images has not been examined using images ori-
`ented to the cardiac axes, and no NMR
`imaging study
`has considered
`the effects of rotational
`and transla-
`tional cardiac motion. To address
`this problem, we
`used a floating endocardial
`centroid method
`to mea-
`sure wall thickening
`from images oriented
`to the short
`axis of the left ventricle
`[12]. The results
`indicated
`that mean global left ventricular
`function of the pa-
`tients
`receiving multiple blood transfusions
`was nor-
`mal, a finding
`that was consistent with
`the absence of
`significant myocardial
`iron overload
`in this group.
`
`Clinical Applications
`imag-
`that NMR
`indicate
`The results of this study
`for the
`ing may
`serve as an alternative
`to tissue biopsy
`detection of significant
`iron overload. Although physi-
`cochemical measurements
`of iron concentration
`ob-
`tained at tissue biopsy may at present provide a more
`quantitative measure of iron content, NMR
`imaging
`has-the advantage of being a noninvasive
`procedure.
`Moreover,
`in comparison
`to techniques
`like dual ener-
`gy computed
`tomography
`and
`the superconduction
`quantum
`interference
`device (SQUID)
`[30,31], NMR
`imaging
`is a rapidly
`proliferating
`and
`increasingly
`available procedure.
`In the heart, where
`right ventric-
`ular endocardial biopsy
`is subject
`to considerable
`sam-
`pling error
`[32], NMR
`imaging methods may be better
`than direct
`tissue sampling. NMR
`imaging should be
`helpful
`in identifying
`patients who present with high
`serum
`ferritin
`levels without
`iron overload. NMR
`im-
`aging will also be useful
`for the serial assessment
`of
`tissue
`iron content
`in patients
`receiving
`repeated
`blood transfusions.
`Finally, NMR
`imaging has the po-
`tential
`for evaluating
`residual
`tissue
`iron content
`in
`
`Apotex Tech.
`Ex. 2031
`
`

`

`patients with
`go therapeutic
`
`idiopathic hemochromatosis
`phlebotomies.
`
`who under-
`
`Future Developments
`pulse
`of motion-compensated
`With
`the production
`sig-
`sequences
`to prevent
`regional losses of myocardial
`flow
`nal intensity,
`and the reduction
`in transmitted
`of
`signal
`to the image plane
`[33,34],
`the sensitivity
`NMR
`imaging
`to detect small quantities
`of cardiac
`iron should greatly
`improve. Since gradient echo pulse
`sequences are more sensitive
`to spin dephasing, small-
`er amounts of iron can be detected with
`these se-
`quences.
`
`ACKNOWLEDGMENT
`of this
`review
`for his
`to Dr. Bruce Rosen
`thanks
`We wish
`to express
`our sincere
`imaging
`the
`manuscript,
`to Dr. Richard Wendt
`for his assistance
`in developing
`protocol,
`to Kenneth
`Nash
`for his assistance
`in developing
`the
`imaging protocol,
`an