Valerie Vilgrain, MD Dominique Najmark, Marie-Pierre
Vullierme,
#{149} Jean-Francois Fl#{233}jou,MD #{149}Lionel MD #{149}Yves Menu, MD #{149}Marc Zins, MD #{149}Henri Nahum, MD
Focal Nodular MR Imaging in 37 Patients’
Hyperplasia and Pathologic
patients with 48 lesions of focal nodular hyperplasia (FNH) underwent preoperative magnetic resonance (MR) examination and surgical resection. Sixteen lesions were imaged at 0.5 T with Ti- and T2weighted spin-echo sequences; 32 lesions were imaged at 2 T with Tiand T2-weighted spin-echo and gradient-recalled-echo sequences. Contrast material-enhanced MR imaging was performed in 20 lesions. MR finaging failed to depict six tumors that were less than 3 cm in diameter. Typical appearance was present in 18 of the 42 (43%) lesions seen at MR. Atypical lesion features included no scar (n = 15), hypointense scar on T2-weighted images (n 7), pseudocapsule (n = 6), strong hyperintense lesion on T2-weighted images (n = 3), diffuse hyperintensity on Tiweighted images (n 3), and heterogeneous lesion (n = 1). Comparison between findings at MR imaging and at histopathologic examination was performed in 38 lesions: There was good correlation between presence and size of the scar on both examinations. In 13 of 20 (65%) of the hyperintense scars on T2-weighted images, edema was prominent, whereas in five of the seven (7i%) hypointense scars on T2-weighted images, edema was absent or low. Thirty-seven
M
of the pearance
perplasia
terms:
761.3119
Liver, focal nodular hyperplasia, Magnetic resonance (MR), tissue
#{149}
characterization
Radiology
1992;
articles about the apof focal nodular hy-
(FNH)
at magnetic
reso-
nance (MR) imaging include a small number of patients (1-5). Recently, two large series have been published; they made use of different magnetic fields and pulse sequences (6,7). The results showed a marked discrepancy, considering the prevalence of MR imaging characteristics. Furthermore in these series, findings at MR imaging were not correlated with findings at histopathologic examination. The purpose of our study was twofold: to study the characteristics of FNH in a surgical series and to correlate MR findings with histopathologic findings to explain the wide range of appearances of FNH on MR images. MATERIALS Between
AND
METHODS
underwent
1985 and 1991, all the patients
surgical
resection
of the liver
lesion. Multiple reasons led to the decision to perform surgical resection in these 37 patients. Abdominal pain or discomfort and uncertain diagnosis were reasons in some cases, and pregnancy was antici-
in some
women.
The purpose
this study was to compare findings imaging and pathologic examination;
therefore,
only
the patients
who
(mean
age, 35 years;
range,
21-57
one man (aged 24 years, with tory of steroid use). From 1985 to 1987, MR imaging performed From the Departments of Radiology (V.V., L.A., D.N., Y.M., M.Z., M.P.V., H.N.), Pathology (1FF.), and Surgery U.B.), H#{244}pitalBeaujon, 100 bd du G#{233}n#{233}ral Leclerc, 92110, Clichy, France. I
ceived
the 1991 RSNA scientific
assembly.
Re-
11, 1991; revision requested February 10, 1992; revision received April 2; accepted April 7. Address reprint requests to V.V. ©
December
RSNA,
1992
of at MR
had un-
dergone resection were included in this series. These patients induded 36 women
184:699-703
with
in
13 patients
with
a 0.5-T superconducting
tem
(Magniscan;
Medical
Belghiti,
MD
Systems,
General Paris).
years) no his-
16 lesions
Electric the
with 2-mm gap. T2were obtained with TR TE of 50, 100, and 150 matrix, two acquisitions, with 1.6-mm gap. From 1987 to 1991, 24 patients with 32 besions underwent imaging on a 2-T superconducting system (Gyrex; Ebscint, Haifa, Israel). Ti-weighted spin-echo imaging was performed with TR of 450 msec, TE of 15.4 msec, a 140 x 256 matrix, four acquisitions, and 10-mm thickness with 1-mm gap. Gradient-recalled-echo images were
obtained msec,
with
sys-
CGR 0.5-T
TR of 58 msec,
a flip angle
TE of 12
of 40#{176}, a 120 x 220 ma-
tnx, and 10-mm T2-weighted
thickness.
images
TR of 2,000 msec,
were
Double-echo obtained
with
TE of 26 and 80 msec,
a
128 x 256 matrix, two acquisitions, and 10-mm thickness with 1-mm gap. Respiratory artifacts were suppressed by means of respiratory compensation. In addition to these sequences, a bobus of 0.1 mmol of gadopentetate dimeglumine (Magnevist;
Berlin)
per kilogram
was in-
jected through an antecubital vein in the last 17 patients (20 lesions). After the bolus injection, gradient-recalled-echo imaging was started in the plane of the lesion and basted 2 minutes. Then, 3 minutes after the bolus injection a delayed contrast material-enhanced TI-weighted sequence was started. The detection rate of the lesions with
MR imaging
was compared
with
the de-
tection rate with intraoperative US. Surgicab resection of at least one lesion per patient was performed in all patients and in 43 of the 48 lesions. In the remaining five lesions the pathologic proof was obtained
with
histologic
after
surgical
Qualitative
examination
topathobogic
of the tissue
biopsy.
analysis
made in conference (V.V., L.A.)-with diagnosis of FNH
of liver lesions
regarding
was
by two radiologists the knowledge of the but not of the his-
findings-who
consensus
was
imaging With
10-mm thickness weighted images of 1,800 msec and msec, a 256 x 256 and 8-mm thickness
Schenng,
referred to our department with suspected FNH underwent MR imaging. All the besions had been previously detected with ultrasonography (US) or computed tomography (CT). During this period 59 patients were examined; of these patients, 37
and
From
Jacques
#{149}
ofthe Liver: Correlation
OST
pated Index
Arriv#{233},MD MD
the
reached following:
a (a) sig-
nab intensity of liver lesions compared with that of normal liver on each Ti- and T2-weighted and gradient-recalled-echo image
and
Ti- and T2-weighted spin-echo imwere acquired. TI-weighted images were obtained with repetition time (TR) of 300 msec, echo time (TE) of 26 msec, a
Abbreviations:
256
plasia,
of hyperintense
lesions
on T2-
unit, ages
x 256
matrix,
four
acquisitions,
and
TE
=
FNH = focal nodular echo time, TR = repetition
hypertime.
699
weighted images compared with the signab intensity of normal spleen; (b) signal homogeneity; (c) lesion enhancement after bolus injection of gadopentetate dimeglumine; and (d) presence, size, and signal intensity
of the
lesion
contrast-enhanced sity
of the
on
unenhanced
images.
lesion,
signal
sion enhancement, scar were considered
and
Signal
inten-
homogeneity,
le-
and presence secondary
of the features.
The size of the scar was measured. lesions
were
classified
cal. Typical
appearance
according
to the
The
as typical
was
three
or atypi-
determined
criteria
of Mattison
et al (2): (a) isointensity on TI- and T2weighted sequences, (b) a central hyperintense scar on T2-weighted images, and (c) homogeneous signal intensity. The analysis of the resected specimen was performed by a pathologist Q.F.F.)
with
no knowledge
macroscopic items were (b) presence (c) presence
of the MR findings.
b.
At
examination the following assessed: (a) homogeneity, and size of the scar, and of a pseudocapsule. At micro-
scopic examination ysis was performed
semiquantitative of slides stained
hematoxylin-eosin,
chromotrope
lin, and picrosirius.
The two latter
ings
were
considered
The
following
titatively
a.
graded
R, reticu-
specific
conditions on
for
were
a scale
analwith
staincollagen.
semiquan-
from
absent
to
prominent within the scar. No statistical analysis was performed. Occluded vessels were examined. Comparison between findings at MR imaging and at histopathologic examination was made in 38 lesions by a radiologist (V.V.) and the pathologist Q.F.F.) together concerning lesion homogeneity and size, location, and composition of the scar.
d.
C.
Figure 1. Typical age (TR msec/TE
lobe. The lesion
MR msec
(2-T =
contains
unit)
and
450/15.4)
a central
macroscopic shows
findings
a large
of FNH.
isointense
hypointense
mass
scar (arrow).
(2,000/80) shows homogeneous lesion. layed, contrast-enhanced, Ti-weighted
Central scar (arrow) MR image (450/15.4)
the scar (arrowheads). (d) Macroscopic and a large central scar.
view
of the same
(a) TI-weighted
MR
(arrowheads)
of the
(b) T2-weighted
im-
right
MR image
appears hyperintense. (c) Dedemonstrates enhancement of
lesion
shows
a homogeneous
lesion
RESULTS MR imaging depicted 42 of the 48 lesions identified at surgery and at intraoperative US. The mean size of all 48 lesions was 4.5 cm (range, 1.512.0 cm). Nineteen lesions were located in the right lobe, 27 in the left lobe, and two in the caudate lobe. Multiple lesions were present in eight patients (two lesions in six patients, three in one patient, and four in another patient). Six FNH lesions bocated in the left lobe (mean diameter, 2 cm;
range,
1.5-3.0
cm)
ered only at surgery found at MR imaging.
MR
were
and
Imaging
The
aforementioned
characteristics
(Fig
typical la,
ib)
of
Radiology
#{149}
Atypical
(a) Ti-weighted
FNH.
were
MR
foci (arrows). (arrow).
lumine was intravenously injected. On contrast-enhanced gradient-recalled-echo images, all the lesions were hyperintense to the liver 10-30 seconds
MR
present in 18 of 48 (38%) lesions and in 18 of 42 (43%) of the FNH lesions detected at MR examination. In seven of the 18 cases, gadopentetate dimeg700
2.
the lesion contains hyperintense no scar and hemorrhagic spots
not
Results are presented concerning typical and atypical characteristics FNH on MR images.
b.
Figure
discov-
were
a.
after
bolus
injection.
bayed contrast-enhanced, Ti-weighted images, but
one
were
On
de-
spin-echo, all FNH lesions
hyperintense
to the
liver, and in all seven cases, high signab intensity was seen within the scar (Fig
ic).
In four
of the
cases
the
scar
image
(450/15.4)
(2-T
(b) Macroscopic
view
unit)
shows
no
of the same
scar,
lesion
and
shows
was better depicted on contrast-enhanced Ti-weighted images than on T2-weighted images, but no statistical comparison was performed. The remaining lesions (n = 24) had
one
or more
atypical
lesions no scar was Ti- and T2-weighted
In seven tense images
on
findings. identified images
lesions
the scar
Ti- and (Fig 3a).
heavily
was
In 15 on both (Fig 2a).
hypoin-
T2-weighted
September
1992
were
observed
at 2 T. Other
malities-such
abnor-
as a hyperintense
be-
sion on Ti-weighted images (n = 3), strong hyperintensity on T2-weighted images (n = 3), and heterogeneous appearance (n = 1)-were encountered only at 2 T.
Comparison
of MR Findings
Pathologic
Findings at MR imaging and topathobogic examination were pared in 38 cases, representing
b. Figure
3.
Atypical
FNH.
(a) 12-weighted
image
pointense scar (arrow). (b) Macroscopic scar (arrow). Scale indicates centimeters.
view
versus
(1,800/100)
(0.5-T
of the same
lesion
unit)
shows
confirms
a small
the small
lesions
hy-
size of the
and
that
were
seen
on
MR
that were surgically Macroscopicfindings.-In
(71%)
lesions,
a scar
hiscomall the images
resected. 27 of 38 identified at
was
pathologic examination and at MR imaging. The size and location of the scar was well correlated between both examinations (Figs ib, id, 3). The mean size of the scars, determined at macroscopic examination, was 5.5 mm (range, 2-20 mm). In 23 cases the scar was solitary and was located at the center of the FNH, whereas multiple scars were present in two cases and scars were located at the periphery of the lesion in two other cases. In eight of 38 (21%) lesions, no scar was
b. Figure
4.
around
Pseudocapsule
the lesion
shows
fibrosis
around
A peripheral
the
rim
was observed pseudocapsube
patient
surrounding
MR image
mass
(arrow).
Scale
or pseudocapsule
in six patients. was complete
and
FNH.
on Ti-weighted
incomplete
The
signal intensity of the pseudocapsule was low on Ti-weighted images in all cases (Fig 4a). On T2-weighted images
the pseudocapsule was hyperintense in four of the six cases and not visible in the two others. The enhanced Tiweighted images obtained in three patients showed rim enhancement of the
A hypointense
indicates
ages
isointense
(including tense fusely
two
to the
lesions
scars). Three hyperintense
on Ti-weighted lesion tensity images.
had on
hyperim-
spleen
with
hyperin-
lesions were to normal
images
difliver
(Fig 5a). One
heterogeneous signal Ti- and T2-weighted
In 13 of the 24 atypical dopentetate dimegbumine venously
was
injected.
On
hanced
gradient-recalled-echo
images, tense
all the
lesions
in-
lesions, gawas intracontrast-enhyperin-
ages;
in one
hypointense
images,
remained
images.
Field 42 lesions
depicted imaged
on MR
with
middle
the
imaged
at high
27 (33%)
lesions
field strength. However, this difference was not statistically significant (P > .05). No scar was identified in three cases imaged at middle field strength and in 12 cases imaged at high field strength. Hypointense scar depicted
obtained
184
3
scar
were im(2 T). present in imaged at nine of
Volume
Number
the
and 27 strength were lesions and in
observed
#{149}
a or
field strength (0.5 T), aged with high field Typical characteristics nine of the 15 (60%) middle field strength
all FNH
were
enhanced
= =
on
at middle
T2-weighted
field
images
strength
3) and at high field strength 4). Three pseudocapsules
at 0.5 T, and
three
at either
MR
examination. (8%),
a small
imaging
or
In three
other
scar
was
present
at pathologic examination (2-, 3-, and 4-mm diameter, respectively) that was not depicted on MR images.
In 10 cases
on
case
15 were
images,
one
seen
on enhanced
Magnetic
(n (n
but
lesions
view
images. In two cases a scar was detected on images, whereas the scar was either hypointense (n = 1) or was not seen (n = 1) on T2-weighted im-
to liver. On delayed, contrastenhanced, spin-echo, Ti-weighted lesions
is seen
nonenhanced hyperintense enhanced
was were
(arrowheads)
(b) Macroscopic
to liver. not
Of the
lesions were strongly on heavily T2-weighted
and
unit).
centimeters.
scar
pseudocapsule.
Three intense
rim
(0.5-T
hyperintense
The in one
in five.
(a)
(300/26)
found
histologic
were others
Microscopic findings-Histologic analysis of the lesions was
performed
in the 38 cases. All lesions with typical MR characteristics (n = i8) were found to have characteristics typical of FNH at microscopy: homogeneous lesion, no hemorrhage, and no fatty infiltration. Lesions with atypical MR characteristics scopic features mal compression
parenchyma rounding the
had specific microin some cases. Abnorof adjacent hepatic
and mild fibrosis surthe lesion were detected
six rimmed
lesions
Foci of hemorrhage the heterogeneous ing
(Fig
2b).
No
at MR
were lesion
present
the
within
present at MR
hemorrhage
infiltration was found hyperintense lesions images, but sinusoidal the
(Fig
in 4b).
in imag-
or fatty
in the three on Ti-weighted dilatation was
tumor
in two
cases (Fig 5b). In 27 cases the different
of
three
compounds
of the
scar
histologic were
studied
and related to the signal intensity of the scar on T2-weighted images. Quantitative repartition of inflammation and vessels was not related to the signal intensity of the scar. Fibrosis was present in all scars to various degrees. A low content was noted in five
Radiology
70i
#{149}
of 20 hyperintense of seven hypointense weighted images. was
detected
scars and in three scars on T2Abundant fibrosis
in nine
of 20 hyperin-
tense scars and in two of seven hypointense scars on T2-weighted images. A prominent amount of fibrosis was observed in six of 20 hyperintense scars and in two of seven hypointense scars on T2-weighted images. In the 20 hyperintense scars on T2-weighted images, edema was absent in two cases, moderate in five cases, abundant in four cases, and prominent in nine cases. In the seven hypointense scars on T2-weighted images, edema was absent in three cases, moderate in two, abundant in one, and prominent in one. Therefore, edema content was rated as abundant or prominent in 13 of 20 (65%)
hyperintense
scars
and
a. Figure
b. 5.
Atypical
FNH.
(a)
TI-weighted
MR
image
tense mass (arrowheads) and a central hypointense the same lesion shows sinusoidal dilatation within nal magnification, x250).
(450/15.4)
(2-T
unit)
scar. (b) Microscopic the hepatic nodules
shows
a hyperin-
high-power view of (reticulin stain; origi-
as ab-
sent or bow in five of seven (71 % ) hypointense scars on T2-weighted images (Fig 6). DISCUSSION FNH
is a rare
benign
liver
tumor
that is often observed in the 3rd to 5th decades of life. The frequency of this tumor is not well known but has been reported to be 3% in the adult population studied by Karhunen (8). FNH is more
common
than
hepatic
ade-
noma. FNH is a nonencapsulated besion composed of nodules of normal hepatocytes and stellate fibrous bands,
with
edema,
vessels,
bile
Radiology
#{149}
6.
Microscopic
views
of scars
that
were
(a) hyperintense
T2-weighted images. (a) Microscopic low-power view edema (picrosinus stain; original magnification, x40). central scar (same magnification as in a) demonstrates sirius
and
(b) hypointense
of the central (b) Microscopic dense fibrosis
on
scar shows prominent low-power view of the with little edema (picro-
stain).
ducts,
and Kuppfer cells. A central scar is a characteristic feature at macroscopy. Imaging modalities such as angiography and CT demonstrate a homogeneous hypervascularized lesion, but specific diagnosis is rarely possible (9,10). To our knowledge, the first artide about MR imaging of FNH was written by Butch et al (1); they reported mass effect, with tumor nearly isointense with adjacent liver, and the presence of a hyperintense central steblate region on T2-weighted images that corresponded at pathologic examination to the colbagenous scar. Large lesions are most often easily detected on MR images; recognition of small lesions can be difficult due to the relative isointensity of the lesion to the liver. One FNH lesion, 1 cm in diameter, was not found retrospectively in the series published by Mathieu et al (7). In our series, intraoperative US and surgical examination disclosed six of 48 FNH lesions (13%) that were not found at MR imaging; it is interesting to note that all of these lesions were less than 3 cm in diameter and were located in the left 702
b.
a. Figure
lobe of the liver, where motion and vascular artifacts occur more often than in the right lobe. The frequency of the typical appearance of FNH on MR images is not clearly known and varies from 9% to
(seven tense images
of 47) of the lesions were on both Ti- and T2-weighted obtained with the 0.6-T
while
none
50%
the discrepancy was noted only on T2-weighted images. In our series, typical and atypical lesions were not statistically different at middleor high-field-strength imaging. Different TRs used with spin-echo
(2,6).
The
wide
range
in the
fre-
quency of the diagnosis of FNH can be explained in part by a bias of the population (medical vs surgical series). Especially in our surgical series, we observed a fairly high number of patients (24 of 42) with atypical appearance. This is due to the fact that patients underwent surgery when the diagnosis could not be clearly made with preoperative imaging methods. Also, others reasons may exist, including use of different magnetic fields or different TRs with spin-echo sequences, performance of contrastenhanced or nonenhanced imaging, and use of new sequences such as ultrafast dynamic snapshot flash imaging (5,7). The robe of the magnetic field in tissue contrast on Ti- and T2weighted images has been observed by Lee et al (6), who noticed that 47%
tense i.5-T
on images unit. The
sequences intensity. sequences, contrast,
sions
(6).
unit,
to be isoin-
obtained authors
lesions number In our
spin-echo
and
sequences
gave
To our
with suggest
the that
and, therefore, of isointense series,
concerning
debe-
Ti-weighted
gradient-recalled-echo similar
knowledge,
gadopentetate (5,7). cantly
found
also play a robe in lesion Heavily Ti-weighted pulse by providing better tissue increase the number of hy-
pointense crease the
articles
were
isoin-
results.
there
are few
enhancement
dimeglumine
Such enhancement improve lesion-liver
with
in FNH may significontrast
when dynamic imaging is performed. In our study, as in others, the lesions were hyperintense on enhanced imSeptember
i992
ages. Also, the scar was delayed contrast-enhanced images
obtained
enhanced on Ti-weighted
in all cases
(n
in a previous study (7) and one case in our study. Ultrafast imaging
25)
=
in all but
gadolinium-enhanced
MR
enabled evaluation of the fast hemokinetics of FNH by the liver at a frequency of 30 images per minute
(5).
The
specificity
FNH
is not
of MR imaging
well
known
but
in
may
be
high. Only one false-positive case was described by Mathieu et al (7). This lesion
had
typical
MR
characteristics
but was surrounded with an enhanced rim. Pathologic examination demonstrated hepatocelbular carcinoma with a scar. Neither other studies
(1-3,11)
nor
ours
showed
false-positive case. In our series atypical observed cases, the
any
were
(57%). In 15 not seen on
MR images. This finding was observed in three of six cases (50%)
of Mattison et al (2), in i8 of 35 lesions (51%) of Lee et al (6), and in none of the cases of Mathieu et al (7). In our series, comparison of presence and
size of the pathologic
edema,
whereas
showed
absent
hypointense or low
scars
of various
primary
scar at MR imaging and examination showed a re-
hepatocellubar cases of hepatic
edema.
Similar correlation between ings at pathologic examination MR imaging was performed
findand at in central
liver
tumors
(ii). According to findings at the pathologic examination, scars were divided into three subtypes: (a) vascubar scars composed predominantly of vascular channels traversing collagenous tissue; (b) inflammatory scars with edema, necrosis, hypercellularity, and loose (c) collagenous
fibrous
tissue;
and
scars composed predominantly of dense, sclerotic cobbagen. Vascular and inflammatory scars appeared hypointense relative to liver on Ti-weighted images and hyperintense on T2-weighted as collagenous scars
findings
in 24 of 42 cases central scar was
nent scars
tense
relative
images, wherewere hypoin-
to liver
T2-weighted
on both
images.
Rummeny
In the
of the
scars.
and
The
ious typical
and
article
by
in our
1.
study,
4.
ports
the
conclusion
can depict diameter. T2-weighted
that
MR
imaging
a scar greater than 2 mm in Among the 27 scars seen on MR images in our study,
20 were hyperintense and seven were hypointense. Hypointense scar was reported in one FNH lesion (3). In
two cases, hypointense radiating linear areas were associated with hyperintense scar (7). In the latter cases, pathologic examination demonstrated prominent obliterative vascular
perintense Ti-weighted intensity (3%)
were
of the
cases
in a previous
in our
cases,
no
could
perhaps
be the
and
fibrosis,
edema,
inflammation
The
connective
were
distribution
yescarefully
of fibrosis,
inflammation, and vessels was not apparently different in the hyperand hypointense scars. However, edema differed markedly according to the signal
the
intensity
of the
hyperintense
Volume
184
scar.
scars
Number
#{149}
Most
showed
3
of
promi-
of hepatic
parenchyma,
of
nodular
characteriza-
primary
malignant
1987; 148:711-715. H’s’, Saul SF1, Yeager
BA,
MR imaging offocal nodof the liver. J Comput Assist 11:651-654.
1987;
M, Li KCP,
Urhahn
IR,
of he-
WB.
hyperplasia
El Rahman
from
AiR Kressel
of
LE, Francis
MR imaging
hyperplasia:
distinction tumors. ML,
imaging
J Comput
Ros
PR.
Hepatic
R, Klose
KC,
Drobnitzky
hyperplasia:
focal
Magn
M.
Hepatic
diagnosis
by ultra-
fast Cd-DTPA enhanced flash MR. J Comput Assist Tomor 1991; 15:848-851. Lee MJ, Saini S. Hamm B, et al. Focal nodular hyperplasia of the liver: MR findings in 35
proved cases. AJR 1991; 156:317-320. Mathieu D, Rahrnouni A, Anglade MC, et al. Focal nodular hyperplasia of the liver: assesswith
Karhunen mor-like
contrast-enhanced
TurboFLASH Radiology 1991; 180:25-30. PJ. Benin hepatic tumor and tuconditions in man. J Clin Pathol 1986;
39:183-188. 9.
Welch TJ, Sheedy PF, Johnson cal nodular hyperplasia and
CM, hepatic
et al. Foadenoma:
of angiography, CT, US, and scmtigraphy. Radiology 1985; 156:593-595. Mathieu D, Bruneton JN, Drouillard J, CaronPointreau C, Vasile N. Hepatic adenomas and focal nodular hyperplasia: dynamic CT study. Radiology 1986; 160:53-58. comparison
10.
11.
Rummeny
E, Weissleder R, Sironi S. et al. in primary liver tumors: MR feahires, specificity, and pathologic correlation. Radiology 1989; 171:323-326. Itoh K, Nishimura K, Togashi K, et al. Hepa-
Central
12.
scars
tocellular
with
mild fibrosis surrounding the lesion. All the atypical findings observed in our series can mimic hepatic adenoma and hepatocellular carcinoma. Hyperintense signal intensity on Tiweighted images has been reported to occur in 18 of 58 (31%) cases of
focal
hepatic Schiebler
MR
MR imaging.
sion
sels,
study,
and
WD.
ment
images.
In our
vessels
Bree RL, Ensminger
focal nodular
8.
studied. Obliterative vascular changes were noted in both hyperintense and hypointense scars on T2-weighted
in the
5.
7.
the hyperintensity of some hyperintense lesions on Ti-weighted images. It is well known that FNH is a nonencapsulated lesion. However, we observed six cases in which complete or incomplete pseudocapsules were present. In all cases, pathologic examination disclosed abnormal compres-
changes tissue.
of
MR
nodular hyperplasia: new MR findings. Reson Imaging 1989; 7:687-688.
6.
or
cause
Malt RA. hyperplasia. 10:874-877. GM, Quint
Tomogr
is generally hemorrhage,
series,
were
for diagnosis by the presence
Butch RJ, Stark DD, hepatic focal nodular Assist Tomogr 1986; Mattison CR, Glazer
ular
study
surgical
of FNH
dynamic
Axel L, Gefter
fatty infiltration was present. Furthermore, in two of the three cases, the sinusoids were dilated. Intrabesion peliosis
findings
tion and
hy-
hemorrhage
MR
patic
relative to normal liver on images. Such high signal was observed in 12 of 35
(6). The hyperintensity the result of intralesion but
diffusely
tumors.
in our
References
examination.
lesions
T2-
imaging and to compare results with contrast-enhanced and nonenhanced imaging to increase sensitivity while maintaining high specificity. U
2.
discrepancy
on
and size of the scars as demonstrated at pathologic examination. Further studies are needed to assess the role
logic
the
liver
sitivity of MR imaging FNH could be limited
3.
sup-
region
present in 18 of 42 (43%) of the lesions detected at MR. Atypical MR findings, including the absence of scar, were well correlated at histopathobogic examination. This suggests that the sen-
markable correlation in 35 of 38 besions; in the three remaining cases a small scar was detected only at patho-
finding
primary
In conclusion,
could probably be the result of studying different tumors (FNH versus hepatocellular carcinoma and giant hemangioma). In three of 42 (6%) cases in our
This
central
and in some (12,13). A
weighted images has been described in fibrolamellar hepatocellular carcinoma (14). The presence of a capsule or pseudocapsube is suggestive of var-
of contrast-enhanced
Ti-
study, scars with predominant edema had bong T2 TRs. Differences occurred in the significance of hypointense scars on T2-weighted images: low edema content or predominant collagen content
et al (ii)
hypointense
carcinoma adenoma
carcinoma:
13.
1987; Cabata aing
164:21-25. T, Matsui of hepatic
14.
Titelbaum
DS,
Kressel
HY, Burke
lar hepatocellular I Comput Assist
MR imaging.
0, Kadoya adenoma. Hatabu
Radiology
M, et al. MR imAJR 1990; 155:1009-
H, Schiebler
DR. Saul
SH.
ML,
Fibrolamel-
carcinoma: MR appearance. Tomogr 1988; 12:588-591.
Radiology
703
#{149}