John M. Gomoni, MD Liora Komnreich, MD Joseph Ben-Meir, PhD Eliezer A. Rachmilewitz,
Hepatic Quantitative
a a
Gadi Homey, MD a Hannah Tamary, MD a Judith Zandback, MD Rina Zaizov, MD #{149} Enrique Freud, MD a Oliver Krief, MD a #{149} Haim Rotem, MSc #{149} Mario Kuspet, PhD a Phillip Rosen, PhD MD #{149}Elia Loewenthal, PhD #{149} Raphael Gorodetsky, PhD
Iron
terms:
Anemia,
eases,
761.594,
761.652
761.652
Liver,
a
ic resonance
#{149} Liver,
MR studies,
(MR),
Radiology
761.652
1991;
Liver,
a
iron
T
HE regular and frequent administration of med blood cell transfusions to patients with /3-thalassemia major has prolonged survival and eliminated the consequences of the severe anemia. However, the price of this therapeutic gain has been the exacerbation of the difficult problem of tissue iron overload. Excessive iron deposition occurs in sideroblastic and other anemias, idiopathic hemochnomatosis, and dietary iron overload. In many of these cases, the iron is metamed mainly in the parenchymal tissues. In /3-thalassemia and in other cases of transfusional iron load, a greaten proportion of iron is retained in the neticuloendothelial system, concurrently with the more hazardous parenchymal deposits. The transition from iron deposition in the melatively benign neticuloendothelial system to the parenchyma has not been thoroughly studied. The parenchymal iron deposits lead to damage of the heart, endocrine glands, and liven (1). The ability to determine accurately the iron stones of an individual organ and the total iron burden in each patient would help to relate the iron deposition to the clinical manifestations and to evaluate chelation therapy regimens. The detection of iron overload has been difficult. Hemosidenin and fermitin, the iron storage proteins, are mostly intracellular. An increased serum femnitin concentration is usually regarded as a crude parametem indicating increased total body iron. However, it may also me-
dis-
content,
761.1214
comparative
a
Magnet-
studies
179:367-369
flect inflammation or hepatic damage (2). Presently, the most accepted way to estimate total body iron overload is the direct evaluation of hepatic iron concentration by needle biopsy. These measurements correlate with the total amount of blood transfused, and the pathologic examination of the biopsy specimens indicates the severity of the subsequent hepatic fibrosis (3). However, this invasive procedure cannot be repeated for the routine follow-up of the iron ovenload in these patients. Other methods proposed for the assessment of body iron load include hepatic computed tomogmaphic (CT) attenuation measurements (attenuation coefficient) (4-6), bulk magnetic susceptibility (7), nuclear resonance scattering of gamma nays (8,9), and noninvasive evaluation of skin iron by means of diagnostic x-ray fluonescence spectrometry (iO,1 1). It has been shown that in vitro hepatic 1 /T2 measurements made by means of magnetic resonance (MR) imaging or nuclear MR spectroscopy correlate well with liver iron content in iron-overloaded experimental animals (12,13). The purpose of the present study was to assess in vivo 1/ T2 measurements obtained by means of MR imaging as a measure of liver iron content in hemosiderotic patients and compare them with hepatic CT attenuation measurements.
MATERIALS Ten mia
I
From
the
Departments
of Radiology
(EAR.),
Hadassah
University
Pediatric Surgery
Radiology (G.H., L.K.), (E.F.), Beilinson Medical
Hospital, Pediatric Center,
(J.M.G.,
O.K.),
Oncology
(EL.,
P0
12000,
Jerusalem
91120,
Box
Hematology-Oncology Petach Tikva, Israel;
HR., M.K., P.R.). Received September 28, 1990; revision February 7, 1991; accepted February 12. R.G. supported grant. Address reprint requests to J.M.G. (. RSNA, 1991 See
also
this issue.
the
article
by
Siegelman
a
Overload: MR Imaging’
Iron deposits demonstrate characteristically shortened T2 relaxation times. Several previously published studies reported poor correlation between the in vivo hepatic 1/T2 measurements made by means of midfield magnetic resonance (MR) units and the hepatic iron content of iron-overloaded patients. In this study, the authors assessed the use of in vivo 1/T2 measurements obtamed by means of MR imaging at 0.5 T using short echo times (13.4 and 30 msec) and single-echo sequences as well as computed tomographic (CT) attenuation as a measure of liver iron concentration in 10 severely iron-overloaded patients with $-thalassemia major. The iron concentrations in surgical wedge biopsy samples of the liver, which varied between 3 and 9 mg/g of wet weight (normal, 0.5 mg/g), comelated well (r .93, P .0001) with the preoperative in vivo hepatic 1/ T2 measurements. The CT attenuation did not correlate with liver iron concentration. Quantitative MR imaging is a readily available noninvasive method for the assessment of hepatic iron concentration in ironoverloaded patients, reducing the need for needle biopsies of the liver. Index
a
et a! (pp
361-366)
and
R.G.), Israel;
(H.T., Elscint
R.Z.), Ltd.
requested November by a Hadassah Research
and
the
editorial
by
Stark
and
Hematology
the
Departments
of
Pathology (J.Z.). and Haifa, Israel (J.B.M., 9; revision received and Development
(pp
333-335)
AND
METHODS
transfusion-dependent
major
patients
scheduled
for
creased
transfusion
mL/kg/y), All the
were patients
aged
/3-thalasse9-20
splenectomy
years, because
requirements included in were imaged
of in-
(>250 this by
study. means
MR and CT preoperatively. Since 1982, we have given subcutaneously administemed deferoxamine (Desfemioxamine, Ciba-Geigy, Basel, Switzerland) (140 mg/
in
Abbreviation:
TE
=
echo
time.
of
kg per night) age of 4 years. piiance
to all patients reaching However, the overall
with
treatment
case,
a surgical
from
the
of the
inferior
liver
is poor.
wedge
the
after
surgery.
liver
The
unprocessed
the
direct
their examiwithin 24
content
was
assayed
measurement
concentrations
sues and in vitro,
for the analysis as previously
tissue matic,
The
samples gauze,
of major
trace-
in superficial
tis-
of tissue described
elements
are excited low-energy
samples in detail
examined
by a focused, x-ray beam.
in the monochmoThe char-
actemistic x-ray fluorescence are detected concurrently
energy with the
of the
incident
beam,
serves
as an internal
normal
excitation
mg/g of wet weight or less. MR imaging was performed Gyrex (Elscint, Haifa, Israel) images
repetition
were
time
times separate tion,
images
msec with echo after 18.4 msec dient
and
were
after
the
a C
with
and
a
0;
echo
a
in two In addi-
echo.
of
The
sinc lobes
suppression
pulse of the was
complete
tance from the section section thickness. The tamed
with
phase
14.4
Fe
1 /T2 and assuming upper hepatic chymal CT
regions
(E!scint) 10-mm kV,
35-cm
in the
and
breathhold).
sumements
were
(normal
the
liver
is 40-80
sections
that
HU)
to
regions
of
lobe away hila and were
normalized for each 0.9% sodium chlo-
was
simultaneously
concentrations were compared
Radiology
mea-
at three
homogeneous hepatic
X 340
time
attenuation
They were with a standard
lated 1 /T2 and CT attenuation
a
acquisition
in the right nonparenchymal
ride solution scanned. The iron specimens
340
CT
in three
1-cm-square
averaged. section
of view,
2-second
made
interest from
The parameters thickness, 350 mAs,
field
(single
368
decay,
of interest
imager. section
matrix,
five
calculated,
1 /T2* values measurements.
of the surgical with the ca!cu-
and
with
the
analysis-that significance with
.05 for the upper .66 and P .04 for
is, correlations tests-was
SPSS/PC+
software
10000
Biopsy
(igIg)
liver
iron
content
the CT
(r
.62 and
segment the lower
and seg-
r
ment).
Chicago).
The results of the MR imaging and CT measurements and the liven iron concentrations in the 10 patients studied are presented in the table. The iron content of the hepatic biopsy specimens correlated well with the 1 /T2 and 1 /T2* of both the supenor and inferior portions of the night hepatic lobe. The correlation was best with the i/T2 of the lower segment (near the biopsy site); the correlation coefficient r was .93 (P .0001) (Figure), leading to the following equation: hepatic iron concentration = [a 1 /T2] + b, where a 46 g - sec /g ± 7, b = 400 g - sec/g ± 900, and the iron is in micrograms per gram (wet weight) and 1 /T2 in seconds ‘. Generally, the 1 /12* values were approximately four times the 1 /T2 values. But, the 1 /T2* values were noisier and had a worse correlation with
Iron
biopsy sample from the lower segment of the right heof wet weight versus the 1 /T2 (a) and 1 /T2* (b) values of the right hepatic lobe. The 1 /T2 versus iron content meof the right hepatic lobe is shown in a.
RESULTS
Elscint).
and lower segments of the right lobe, away from the nonpamenhila. images were obtained on an Excel
1800 were 120
performed
(SPSS,
5000
‘
Liver
b.
Statistical with two-tailed
of This
‘
(gIg)
a.
at a dis-
option;
values were monoexponential
in i-cm-square
Iron
Plots of liver iron concentration in patic lobe in micrograms per gram the lower and higher segments of gression line of the lower segment
cycle-dependent
(Freeze
1 /T2* simple
Liver
edge of 30% of the images were ob-
respiratory
encoding
0
10000
field
with suppression section profile.
almost
a
#{176}
I-
a
of view was 30 or 35 cm, the acquisition matrix was 200 X 256, and two excitations were performed. The section thickness was 10 mm with a 3- or 5-mm intersection gap. The sections were excited with a modified the side
500
-J
at a TE
spin
C
C
The is 0.5
a i-msec delay of the gradient the spin echo and at a TE of with a 5-msec delay of the gra-
echo
a
Lower
#{176}
on a 0.5-T imager. Axial
30 msec sequences.
obtained
i-
which
obtained
of 500 msec
(TEs) of 13.4 single-echo
1000
lines scatter
calibration (ii). liver iron content
nonheme
spin-echo
in
by means of diagnostic a method designed
element
(10,11,14).
lobe dur-
iron
samples
nondestructively x-ray spectrometmy,
right
obtained
and kept on ice until which was performed
the
for
sample
of
g) was
ing the splenectomy. The were placed on saline-soaked sealed, nation, hours
In each
biopsy
portion
(0.5-1.5
the corn-
P
There was liven iron attenuation
no correlation concentrations measurements.
between and the
DISCUSSION In the tients
follow-up
with
of thalassemic
severe
of other iron-overloaded iron level in needle of the liver has been
major
indicator
overload. sy of the not
of total
body
to
iron
routine
monitor
biop-
iron
load
that
MR
is
desirable.
In vitro
studies
imaging -
patients, the biopsy samples adopted as the
Nevertheless, liver
pa-
and
hemosidenosis
the
may
suggest
the
be
noninvasive
hepatic
iron
correlation
best
evaluation load.
We
between
imaging
of
of
in
wedge
of the liven. Reticuloendothelial nitin and its denatured
the vivo
and
surgical
for severe
examined the
measurements
content
modality
MR
the
iron
biopsy
samples
sidemin, iron.
are They
the appear
deposits form,
major
of ferhemo-
store
of body
hypointense
on
May
1991
MR
images
caused
by
due the
hemosidenin
to an
increased
magnetic
and
1 /T2
properties
fennitin
of
(12,13,15-
18). This 1 /T2 increases quadratically with field strength (below magnetic saturation) and is sensitive to the coarseness of the distribution of the hemosidenin and fennitin deposits (18). There is good in vitro comrelation between 1 /T2 measurements and iron content of hemosiderotic livers and spleens (12,13,17). Howeven, the 1 /T2 due to the iron overload in spleens is lower per equivalent iron concentration (J.M.G., H.T., and R.G., unpublished data, 1989). This may reflect a difference in the distnibution of iron deposits between the spleen and liver. Clinical evaluation of MR images
by Hemnandez
et a! (19)
showed
that
nine healthy subjects and 12 others with mild to moderate iron overload, excluding a patient with high liver iron content. They used higher magnetic field strengths and multiple
short
TEs.
Despite
the
dependence
7.
8.
of
the 1 /T2 of iron deposits on field strength, they pooled measurements obtained at 1.0 T with those obtained at 1.5 T. They obtained a slope of the iron content of the hepatic needle bi-
opsy
samples
(dry
weight)
versus
1/
T2 six times ours (we used wet weight) with a correlation of r .98. In view of the different field strengths (ours 0.5 T, theirs 1.0 and
1.5 T) and
the
difference
between
crepancy
noisy
signal.
Our
use
for
with
field
quence,
and
particular
unit
strength, MR
im-
1.
2.
3.
4.
Number
2
noisy.
previous
5.
tissue
iron
17.
18.
19.
20.
21.
(4-6).
Computed
and
J Pathol
1975;
JL, Nienhuis
AW,
tomographic
analysis
307:1671-1675. Vartsky D, Ellis K), Hull DM, et a!. Nuclear resonant scattering of gamma rays: a new technique for in vivo measurement of body iron stores. Phys Med Biol 1979; 24:680-701. Wie!opolski L, Ancona RC, Mossey RT, et a!. Nuclear resonance scattering measurement of human iron stores. Med Phys 1985; 2:401404. Zeimer R, Be!kin M, Leitersdorff E, Rachmilewitz EA. A noninvasive method for the evaluation of tissue iron deposition in betathalassemia major. J Lab Clin Med 1978; 91:24-31. Corodetsky R, Goldfarb A, Dagan I, Rachmilewitz EA. Noninvasive analysis of skin iron and zinc levels in 3-thalassemia major and intermedia. J Lab Clin Med 1985; 105:4451. Stark DD, Moseley ME, Bacon BR, et a!. Magnetic resonance imaging and spectroscopy of hepatic iron overload. Radiology 1985; 154: 137-142. Israel J, Unger E, Buetow K, Brown T, Blum-
B, London
hepatic
WT.
Correlation
and Magn
R, Sherman
MA.
16.
The
studies
concentration
fibrosis in thalassemia. 95. Long JA, Doppman
magnetic Reson
22.
Magnetic
25.
imaging
MR.
of trans-
1990; 74:360-363.
Wismer CL, Buxton RB, Rosen BR, et a!. Susceptibility induced MR line broadening: applications to brain iron mapping. J Comput Assist Tomogr 1988; 12:259-265. Edelman RR, Johnson K, Buxton R, et a!. MR of hemorrhage: a new approach. AJNR 1986; 7:751-756. Young IR, Khenia S. Thomas DGT, et a!. Clinical magnetic susceptibility mapping of
the brain. 26.
resonance
U, Wexler
fusional hemosiderosis complicating thalassemia major. Radiology 1984; 150:767-77 1. Comori JM, Grossman RI, Goldberg HI, Zimmerman RA, Bilaniuk LT. Intracranial hematomas: imaging by high-field MR. Radiology 1985; 157:87-93. Gomori JM, Grossman RI, Drott HR. MR relaxation times and iron content of thalassemic spleens: an in-vitro study. AJR 1988; 150:567569. Gomori JM, Grossman RI. Mechanisms responsible for the MR appearance and evolution of intracranial hemorrhage. RadioGraphicy 1988; 8:427-440. Hernandez RJ, Samnaik SA, Land I, et a!. MR evaluation of liver iron overload. J Comput Assist Tomogr 1988; 12:91-94. Johnston DL, Rice L, Vick W, Hedrick TD, Rokey R. Assessment of tissue iron overload by nuclear magnetic resonance imaging. Am Med 1989; 87:40-47. Chezmar JL, Nelson RC, Malko JA, Bernardino ME. Hepatic iron overload: diagnosis and quantification by noninvasive imaging. Gastrointest Radio! 1990; 15:27-31. Kaltwasser JP, Gottschalk R, Schalk KP, Hart! W. Non-invasive quantitation of liver ironoverload by magnetic resonance imaging. Br
Haematol
24.
Y, Zagher
between resonance Imaging 1989;
11:2-6. Overmoyer Uniformity (storage)
Med
J Comput
Assist
Tomogr
1987;
BA, McLaren CE, Brittenham of liver density and nonheme iron distribution. Arch Patho!
GM. Lab
1987; 111:549-554.
116:83Mills
of beta-
thalassemia syndrome with hemochromatosis: pathologic findings with clinical and laboratory correlations. J Comput Assist Tomogr 1980; 4:159-165. Houang MT. Skalicka A, Arozena x, Huehns
ER, Shaw
a
more
Powell LW, Isselbacher KJ. Hemochromatosis. In: Braunwald E, Isse!bacher KJ, Petersdorf RC, Wilson JD, Martin JB, Fauci AS, eds. Harrison’s principles of internal medicine. New York: McCraw-Hil!, 1987; 1632-1635. Lifschits DA, Cook JD, Finch CA. A clinical evaluation of serum ferritin as an index of iron stores. N Eng! J Med 1974; 290:12131216. Ridson RA, Barry M, Flynn DN. Transfusional iron overload: the relationship be-
SR.
is necessary.
179
also
with
tween
se-
imaging
The accuracy of iron evaluation with MR imaging by means of 1/T2 measurement was also suggested recently in a communication by Kaltwasser et a! (22). They studied
Volume
ments
et a!. of hu-
1982;
15.
References
low 1evwhich
pulse
As expected, to
were
J Med
Gorodetsky R. Elevated concentrations of elements and abnormalities of neuromuscular functions in tongue muscles of Down’s syndrome. J Neurol Sci 1987; 79:315-326. Brasch RC, Wesbey CE, Gooding CA, Koerper
23.
have longer T2 relaxation times. This should be corrected by the acquisition of a sequence with a delayed echo, perhaps at a TE of 50 msec. Our data yield a calibration for the noninvasive quantitation of hepatic iron content by means of 1 /T2 measurement with our clinical MR imaging unit operating at 0.5 T. A study of the variation of the 1 /T2 of hemosideno-
tic liver
et a! (23).
JW,
measurements
N Eng!
Yarom
from
the 1 /T2* measurements proved be more sensitive to iron deposits (24,25); however, these measure-
DF, Harris
stores.
iron
14.
after
In conclusion, quantitative MR imaging is a readily available noninvasive method for the assessment of hepatic iron overload. U
of an earlier echo with a TE of 13.4 msec, and perhaps the use of singleecho sequences as well as respiratory cycle-dependent phase encoding, yielded less noisy measurements. However, our present MR imaging sequence is less sensitive els of hepatic iron overload,
of Wismer
echo
patterned
Farell
susceptibility
man
liver iron content imaging in rats. 7:629-634.
low-intensity,
that
echo
12.
GM,
Magnetic
berg
proved correlation with the liver iron content of the 1 /T2 value nearest the biopsy site suggests mild intrahepatic variation in iron content in iron-overloaded livers, unlike the uniformity observed in nonoverloaded livers (26). The lack of correlation between hepatic iron content and CT attenuation measurements is striking. We do not have an explanation for the dis-
the gradient was
spin
11.
13.
hepatic iron concentration did not correlate well with hepatic T2 measurements and correlated only poorly with the hepatic-to-muscle MR imaging intensity ratios. Their study was performed at 0.35 T with TEs of 28 and 56 msec. Johnston et a! (20) demonstrated only a qualitative comrelation with the liver-to-muscle MR imaging intensity ratios at 0.5 T with a TE of 35 msec. Similarly, Chezman et a! demonstrated no correlation with the liver T2 measurements and only a poor correlation with the liven-tomuscle intensity ratios at 0.5 T with a TE of 60 msec (21). It appears that their poor correlations were due to the relatively delayed first echo of greater than 30 msec, resulting in a
the
10.
the
dry weight and wet weight of liver samples, their results for moderate liver iron contents are compatible with ours for high liver iron loads. Our method of measuring 1 /T2*
by separating
9.
Brittenham
DC.
ed tomographic in thalassemia
Correlation
between
comput-
values and liver iron content major with iron overload. Lan-
cet 1979; 1:1322-1333. 6.
Olivieri
DJ, Rose
NF,
Crisaru
V. Freedman
D, Daneman
MH.
A, Martin
Computed
to-
mography scanning of the liver to determine efficacy of iron chelation therapy in thalassemia major. J Pediatrics 1989; 114:427-430.
Radiology
a
369