Downloaded from www.ajronline.org by 173.255.180.22 on 10/25/15 from IP address 173.255.180.22. Copyright ARRS. For personal use only; all rights reserved
751
Pictorial I
S.
.
MR Imaging Correlation Fumikazu Jun Aoki,1
of Thymoma:
pathologic
Radiologic-Pathologic
specimens.
with surgically proved thymomas were reviewed and compared
In
nine
cases,
MR
images
of
(12 with
Materials The
excised and
specimens also were made and evaluated. On T2-weighted images, I 1 of 12 malignant thymomas had an inhomogeneous signal intensity, half with and half without a lobulated internal architecture. None of the five benign thymomas had a lobulated internal architecture, and they all had a moderately or slightly inhomo-
geneous
signal.
pathologically,
Cystic and
regions
and/or
corresponded
to
hemorrhage areas
of
were noted
inhomogeneous
four
Department Department
AJR 158:751-756,
had
showed
group
patients
comprised with
benign
microscopic
capsular
no cytologic
1992 0361-803X/92/1584-0751
0 American
with The
malignant
six women
thymoma and
1 1 men
invasion
in the
pathologic
specimens.
at surgery. Whereas two maligcytologic atypia, the other 10
atypia.
patients were examined with a 1 .5-T superconductive MR unit (Signa, General Electric, Milwaukee, WI) or a 0.5-T superconductive unit (Resona, VMS, Tokyo, Japan). MR images were obtained with spin-echo Ti -weighted imaging sequences, 500-i 000/
20-25 (TRITE), and proton density-weighted and T2-weighted multiecho imaging sequences (1 500-2500/20-25,50-i 00) by using ECG or peripheral pulse gating and respiratory compensation. The slice thickness ranged from 5 to 7 mm with a 2- to 5-mm interslice gap. MR and CT examinations were performed within 2 weeks before surgery. In nine thymomas (five malignant and four benign), Ti weighted (600/20) and T2-weighted (2000-3000/80-1 00) axial MR images of the excised specimens were obtained with the Signa unit
Matsumoto, 390, Japan. Address Asahi, Matsumoto, 390, Japan.
of Surgery, School of Medicine, Shinshu University, Asahi, Matsumoto, 390, Japan. of Laboratory Medicine, School of Medicine, Shinshu University, Asahi, Matsumoto, 390, Japan. April
1 2 patients thymoma.
Seventeen
,
4
study
Ueda,1
Methods
None showed pleural dissemination nant thymomas showed obvious
Although several articles have described the MR appearance of thymic tumors [1 2], detailed studies correlating the MR and pathologic appearances of these lesions have not yet been reported. Accordingly, we reviewed the MR images of 1 7 surgically proved thymomas and compared the findings with gross and microscopic findings noted in surgical specimens.
Received July 15, 1991 ; accepted after revision November 5, 1991. 1 Department of Radiology, School of Medicine, Shinshu University, Asahi, 2 Department of Internal Medicine, School of Medicine, Shinshu University,
five
and
Hitoshi
were 36-75 years old (average, 55 years). All thymomas were oxcised, and a pathologic diagnosis was established. Thymomas were classified as malignant when extracapsular invasion and/or pleural dissemination were seen at surgery, or when there was invasion of the capsule and/or cytologic atypia in microScopic specimens. The diagnosis of benign thymoma was made when features of malignancy were not found. Eight of 1 2 malignant thymomas showed macroscopic extracapsular invasion at surgery, and
high signal intensity seen on T2-weighted images. Examination of the excised specimens in malignant thymomas showed that the lobulated configuration seen in the tumors was caused by thick fibrous septa. Our experience suggests that, afthough calcification cannot be identified, MR is helpful in making a differential diagnosis of mediastinal tumors and in determining malignancy of thymoma.
3
.‘Y
,.
Sakai,1 Shusuke Sone,1 Kunihiro Kiyono,1 Takashi Kawai,1 Atsunori Maruyama,1 Takayuki Honda,2 Masami Monrnoto,3 Keiko lshii,4 and Shu-ichi lkeda2
MR images in 17 patients malignant and five benign)
Essay
Roentgen
Ray Society
reprint
requests
to F. Sakal.
Downloaded from www.ajronline.org by 173.255.180.22 on 10/25/15 from IP address 173.255.180.22. Copyright ARRS. For personal use only; all rights reserved
752
SAKAI
ET AL.
AJR:158,
April
1992
“-‘.4.
.--
0
Fig. 1.-Malignant thymoma in a 41-year-old woman. A, Ti-weighted MR image (600/20) shows anterior medlastinal mass with muftilobular border. Signal intensity of mass is higher than that of skeletal muscle but less than that of fat. Low-intensity lines (arrowheads) separate mass lesion into lobules, and high-intensity regions are present (arrow). B, 12-weighted axial MR image (1800/80) shows increase in signal intensity. Intensity of mass is inhomogeneous and equal to or higher than that of fat. Lobulated internal architecture, separated by thin low-intensity lines, is visible (arrowheads). High-intensity regions are present (arrows). C, T2-welghted MR image (2000/80) of excised specimen shows low-intensity lines (arrowheads) separating tumor into lobules. High-intensfty regions (arrows) are scattered throughout mass. D and E, Gross (D) and microscopic (E) specimens show lobulated configuration, separated by fibrous septa (arrowheads). Regions of very high intensity on 12-weighted images correspond to cystic areas (arrows), some of which are accompanied by hemorrhage.
or a 2.0-T chemical-shift imaging system (General Electric, Fremont, CA). Gross and microscopic specimens (hematoxylin and eosin stain) were
prepared
in the
same
planes
as
MR
images
in
these
nine
tumors. Preoperative ings
of the
MR resected
images
were
specimens.
made of the excised specimens.
compared In nine
with
thymomas,
The homogeneity
the
pathologic MR
find-
images
were
of signal intensity
of tumors on T2-weighted images was determined by the consensus of two radiologists and dassified into one of four grades: inhomogeneous, moderately inhomogeneous, slightly inhomogeneous, or homogeneous.
-
areas of high intensity
Results
Malignant
thymomas
ranged
from 4.5 to 18 cm in maximum thymomas had
diameter (average, 8.2 cm). All 12 malignant higher signal intensity than did skeletal weighted weighted
Eleven of the 12 malignant thymomas had inhomogeneous or moderately inhomogeneous intensities (Figs. 1 B, 2B, and 3B) with scattered high-intensity regions (Figs. 1 B and 2B) on T2weighted images. The remaining one had homogeneously high intensity with several small low-intensity areas in the marginal portion of the tumor on T2-weighted images (Fig. 4B). Six of the 1 2 malignant thymomas had a lobulated internal architecture on T2-weighted images, with round or irregularly shaped high-intensity areas of various sizes, separated by 1 to 2-mm-thick low-intensity lines (Figs. 1 B and 2B). Calcification could not be identified on the MR images. MR images of the excised specimens revealed the scattered
muscle
on Ti
images and increased signal intensity on T2images (Figs. 1 A, 1 B, 2A, 2B, 3A, 3B, 4A, and 4B).
-
and lobulated
internal architecture
seen
on preoperative MR images. Scattered regions of high intensity on T2-weighted images corresponded to cystic regions with or without hemorrhage on pathologic specimens (Figs. 1 E, 2D, 4D, and 4E). Marginal low-intensity areas seen on T2-weighted images of one malignant thymoma corresponded to small mural nodules attached
Downloaded from www.ajronline.org by 173.255.180.22 on 10/25/15 from IP address 173.255.180.22. Copyright ARRS. For personal use only; all rights reserved
AJR:158,
MR
April 1992
753
OF THYMOMA
Fig. 2.-Malignant thymoma in a 75-year-old man. A, TI-weighted axial MR image (1000/20) shows mass with intermediate signal intensity in anterior mediastinum. calcification is not apparent. B, T2-welghted axial MR image (2000/80) shows that mass has a lobulated internal architecture separated by low-intensity lines (arrowheads). Signal intensity of mass is inhomogeneous and high. Regions of high intensity are present within mass (arrows). calcification is not apparent. C, T2-weighted MR image (2000/100) of cxcised specimen shows a lobulated internal architecture with inhomogeneously high intensity separated by low-intensity lines (arrowheads). Regions of very high intensity are noted (arrows). 0, Pathologic specimen shows lobulated configuration, caused by fibrous septa (arrowheads). Cystic areas (arrows) correspond to regions of very high intensity on T2-welghted
images.
r
‘-
-
-
.-.
.
.
L
- ,
. (‘
I
,-
.
‘
of a large
specimens
(Figs. 4D and 4E). The mural
unilocular
several small cystic regions. on MR images corresponded
cystic
mass
on pathologic
nodules
The lobulated architecture to the lobulated internal
included seen archi-
tecture
seen
sected
specimens
lobulated composed
‘
-
.
‘
macroscopically
architecture of mixed
.
-
-
-
.
, .
p7
._‘1,
Fig. 3.-Malignant thymoma in a 50-year-old woman. A, Ti-weighted coronal MR image (500/25) shows mass of inhomogeneous signal intensity without lobulation. B, T2-welghted MR image (1500/75) shows signal intensity of mass has increased but is less than that of fat. Signal Inhomogeneous. No lobulation is evident. C, Histologic specimen shows that thin fibrous septa separate tumor into small lobules. Incomplete capsular invasion (arrow) Because of thin fibrous septa, lesion is not lobulated.
to the wall
:
-:..
I
. . -
(Figs.
on the cut
intensity is evident
surface
is moderately in specimen.
of the re-
i D and 2D). Microscopically,
consisted lymphocytes
of round or irregular and epithelial cells
the areas sepa-
SAKAI
Downloaded from www.ajronline.org by 173.255.180.22 on 10/25/15 from IP address 173.255.180.22. Copyright ARRS. For personal use only; all rights reserved
754
Fig. 4.-Malignant
thymoma
in a 50-year-old
ET AL.
AJR:158,
April 1992
man.
A, Ti-weighted coronal MR image (800/20) shows large anterior mediastinal mass with intermediate signal intensity, which is slightly higher than that of skeletal muscle. Small low-intensity area is noted in marginal portion of tumor (arrow). B, T2-weighted axial MR image (1600/80) shows that signal intensity of mass is high. Area of inhomogeneously low intensity is seen in marginal portion of mass (arrow). C, T2-weighted MR image (2500/100) of excised specimen shows that signal intensity of mass lesion is very high. Area of low intensity in marginal portion of mass Includes small high-intensity regions (arrow). 0, At surgery, a large unilocular cystic mass was excised. Although no local invasion was noted at surgery, pathologic specimen showed microscopic capsular invasion. Cysts were filled with bloody fluid. Three mural nodules (arrows) on inner surface of capsule are visible. E, Histologic specimen of central mural nodule in 0 shows several small cystic lesions within mural nodule.
a slightly
inhomogeneous
intensity
was seen on MA, areas
rated by relatively thick fibrous septa (Figs. i 0 and i E). Six thymomas that showed no distinct lobulated internal architecture had no lobulated configuration or a small lobulation separated by thin fibrous septa on pathologic specimens (Fig. 3C). Calcification was noted on pathologic specimens of six malignant thymomas. Benign thymomas ranged from 1 .5 to 8 cm in maximum diameter (average, 3.1 cm). All five showed higher signal intensity than skeletal muscle on Ti -weighted images. Signal intensity was increased on T2-weighted images. Four of the
areas where epithelial cells predominated (Figs. SC and SD). The moderately inhomogeneous intensity seen in a small
five benign thymomas
low-intensity
equal
to or slightly
The remaining higher
had slightly inhomogeneous lower
than
that of fat (Figs.
one had moderately
inhomogeneous
than that of fat (Fig. 6A). None of the benign
intensities, 5A and SB).
intensity, thymomas
had a lobulated internal architecture. No calcification was seen in any of the five benign thymomas on MR images. On T2-weighted images of the excised specimens, four benign thymomas showed slightly inhomogeneous intensity (Fig. SC). On pathologic specimens of four benign thymomas in which
with lymphocytes and epithelial cells were mingled, without cystic regions (Fig. SD). When compared, areas where lymphocytes predominated had the same signal intensities as
benign thymoma on T2-weighted images corresponded to multiple small cystic foci in the pathologic specimens (Fig. 6). In none of the five benign thymomas was a lobulated configuration seen in pathologic specimens (Figs. SD and 6B). A
band on the preoperative
T2-weighted
image of
a benign thymoma corresponded to a thick fibrous septum in the pathologic specimen (Figs. SB and SD). No calcification was seen in pathologic specimens of benign thymomas.
Discussion
According to Rosai and Levine [3], all thymomas are composed of a mixture of lymphocytes and epithelial cells in
AJR:158,
April
MR
1992
Fig. 5.-Benign woman. A, Ti-weighted
thymoma axial
Downloaded from www.ajronline.org by 173.255.180.22 on 10/25/15 from IP address 173.255.180.22. Copyright ARRS. For personal use only; all rights reserved
shows a mass lesion
MR
7S5
OF THYMOMA
in a 51-year-old image
in anterior
(750/20)
mediastinum
with signal intensity that is higher than that of skeletal muscle but lower than that of fat. B, T2-weighted axial MR image (2250/80) shows that signal intensity of mass is high and slightly inhomogeneous. Low-Intensity band is Identified in medial portion of mass (arrow). C, 12-weighted MR image (2500/100) of cxdeed specimen shows relatively homogeneous pattern of signal Intensity. Thick low-Intensity band is seen on medial aspect of tumor (arrow). 0, Histologic specimen shows mixed lymphecyte and epithelial-cell components without cystic regions. Lymphocyte-predominant areas appear dark, while eplthelial-cell-predeminant areas appear light. Comparison wIth MR images of specimen shows that there is no distinct difference in signal Intensity between lymphocyte
and epithelial-celI-predominant areas. Low-intensity band seen on MR corresponds to relatively thick fibrous septum (arrow).
Fig. 6.-Benign
thymoma
in a 54-year-old
woman. A, 12-weighted axial MR image (2400/80) shows that signal intensity of mass is higher than that of fat. Mass has relatively inhomogeneous intensity without distinct IObUIatIOn. B, Histologic specimen shows many small cystIc regions (arrows). No lobulations are noted. Owing to limited spatial resolution of MR images, small cystic regions could not be visualized dIstinctly. Small cystic regions probably cause relatively inhomogoneous high-intensity appearance on T2-welghted images.
varying proportions, with scattered cystic regions of various sizes seen in 40% of thymomas pathologically. Cystic lesions in the tumor are filled with clear, proteinaceous, or bloody fluid. Larger tumors are more likely to exhibit cystic changes.
The presence into
lobules
of sharply defined fibrous septa dividing tumors is one
of the
most
characteristic
features
of
thymomas,
and the lobulated
architecture
was seen in 88%
of all thymomas pathologically. In our study, scattered high-intensity regions on T2weighted images corresponded to cystic regions on pathologic specimens (Figs. i B-i E and 2). On Ti -weighted images, the signal intensity of cystic regions was variable, probably
Downloaded from www.ajronline.org by 173.255.180.22 on 10/25/15 from IP address 173.255.180.22. Copyright ARRS. For personal use only; all rights reserved
7S6
SAKAI
depending on the protein content of the cyst fluid or the presence of hemorrhage (Figs. 1A and 4A). Inhomogeneous intensity was probably caused by multiple small cystic regions (Figs. 4 and 6). On occasion, cystic changes are so extensive that solid components of the thymoma are reduced to small mural nodules attached to the capsule (Figs. 4A-4D). The lobulated internal architecture of thymomas seen on MR images corresponded to lobulations caused by fibrous septa in pathologic specimens. Small lobulations or lobulations separated by thin fibrous septa seen in pathologic specimens could not be seen on MR images, owing to the low spatial resolution of MR. The pathologic characteristics of thymomas were well represented on MR images. It is helpful to understand the pathologic background of the abnormalities seen on MR images of thymoma when considering a differential diagnosis. The specificity of the MR findings of the thymomas described here is to be investigated further. Eleven of the i 2 malignant thymomas had an inhomogeneous high-intensity appearance and six had a lobulated internal architecture. Four of five benign thymomas had a slightly inhomogeneous intensity and none had a lobulated architecture. Larger malignant thymomas were more likely to
ET AL.
AJR:158,
April 1992
show the pathologic features of thymomas on MR images. The MR appearance of thymomas may be useful in assessing the malignancy of thymomas, but definitive differentiation between malignant and benign thymomas with MR currently is not possible.
ACKNOWLEDGMENTS
We thank Shun lmai and Mitushiro Momose, Department diology, and Tsutomu Katsuyama, Department of Laboratory cine, Shinshu University, for valuable suggestions Haniuda, Hideki Nishimura, and Osamu Kobayashi, Surgery, Shinshu University, for assistance.
of Ra-
Mediand Masayuki Department of
REFERENCES 1. Von Schulthess GK, McMurdo K, Tscholakoft D, de Geer G, Gamsu G, Higgins CB. Mediastinal masses: MR imaging. Radiology 1986:158: 289-296 2. Molina PL, Siegel MJ, Glazer HS. Thymic mass on MR imaging. MR 1990:155:495-500 3. Rosai J, Levine GD. Tumors of the thymus. In: Atlas of tumor pathology, 2nd series, fasc 13. Washington, DC: Armed Forces Institute of Pathology, 1976: 1-1 66
New Format for Abstracts Beginning in Fall 1 992, the AdA will publish abstracts of articles in a different format, much like the ones recently adopted by the New England Journal of Medicine and the Journal of the American MedicalAssociation. The purpose of the new design is to have abstracts present the essential elements of articles more clearly and concisely. The contents, organization, and length of the abstracts will be the same as before; however, rather than being all one paragraph except for the conclusion, abstracts will now be divided into four paragraphs, each with a title: objective, subjects (or materials) and methods, results, and conclusion. The AJR Guidelines for Authors, which appears at the front of every issue of the Journal, has been revised to help you write abstracts in this new format. Please familiarize yourself with these guidelines and incorporate them into all manuscripts submitted for publication. Robert N. Berk Editor-in-Chief