1221

Giant Cavernous Hemangioma of

the Liver: CT and MR Imaging in 10
Cases

Byung Ihn Choi’ Ten giant cavernous hemangiomas of the liver in eight patients were examined with
Man Chung Han both MR imaging and dynamic bolus CT. The maximal diameters of the tumors were
Jae Hyung Park 6.5-19 cm (mean, 10.8 cm). MR imaging was done with a 2.0-T superconducting magnet
and spin-echo imaging. CT was done with single-bolus dynamic scans. On MR images,
American Journal of Roentgenology 1989.152:1221-1226.

Seung Hyup Kim

all 10 hemangiomas had a heterogeneous appearance. The main part of the tumor
Moon Hee Han
comprised uniform, well-defined, high-intensity areas on T2-weighted images, with
Chu-Wan Kim increasing intensity ratios with prolongation of TR and TE. Other parts of the tumor were
cleftlike and were of lower intensity than the remainder of the tumor on TI-weighted
images and of higher intensity on heavily T2-weighted images. These parts cone-
sponded to the areas of the tumor that were of lower density on dynamic bolus CT
scans. Internal septa in the tumor of low intensity were also noted on all MR pulse
sequences. These parts corresponded to low-density areas on delayed contrast-
enhanced CT.
Familiarity with the characteristics of the internal architecture of giant cavernous
hemangiomas on MR imaging or dynamic bolus CT might be useful in making the correct
diagnosis of this tumor.

In contrast to small, incidentally discovered hemangiomas, giant cavernous

hemangiomas of the liver are less common, more likely to produce symptoms [1,
2], and more often confused with primary or metastatic malignancy [3]. Definitive,
noninvasive imaging of giant cavernous hemangioma is important to avoid angiog-
raphy, biopsy, or exploratory laparotomy.
Dynamic CT has been considered a good imaging technique for evaluation of
hemangioma. However, a review of previous reports [4-9] reveals that CT features
of hemangiomas include a broad spectrum of morphologic patterns, thus raising
questions about the specificity of the CT diagnosis.
Recently, the usefulness of MR imaging in identifying cavernous hemangiomas
of the liver has been discussed [1 0-i 4]. These reports mainly describe MR findings
of small hemangiomas. The MR findings in giant cavernous hemangiomas have not
been documented. In this study, we describe the characteristics of giant heman-
giomas on MR images and compare these characteristics with those noted on
dynamic bolus CT scans.
Received December 13, 1988; accepted after
revision January 25, 1989.
Materials and Methods
This workwas supported in part by the Special
Fund of the Radiological Research Foundation of During a 2-year period, 1 0 giant cavernous hemangiomas in eight patients were studied
Korea (1989).
by MR imaging and dynamic bolus CT. A giant hemangioma was defined as one in which at
All authors: Department of Radiology, College
least one dimension of the tumor exceeded 6 cm. The series included six women and two
of Medicine, Seoul National University, 28, Yeon-
men (age range, 38-51 years; mean, 45). The diagnosis of giant cavernous hemangioma had
gun-Dong, Chongro-Ku, Seoul 110-744, Korea. Ad-
dress reprint requests to B. I. Choi. been established previously by angiography in six patients and/or unchanging sonographic
or CT findings over 1-4 years in two patients. Seven patients had right upper abdominal pain
AJR 152:1221-1226, June 1989
0361 -8o3x/89/1 526-1 221 or epigastric discomfort. Three patients had a palpable upper abdominal mass. In one patient,
C American Roentgen Ray Society the hemangioma was found incidentally on sonography. Liver function studies were normal
 1222 CHOI ET AL. AJR:152, June 1989

in all eight patients. The giant hemangiomas were 6.5-19 cm (mean, differentiated from the cleftlike areas. The contours of the
1 0.8 cm) in the largest diameter. Seven tumors were located in the internal septa were linear or curvilinear in all five cases.
right lobe, two in the left lobe, and one in the caudate lobe. On T2-weighted MR images, the main part of the tumor
MR spin-echo (SE) images were obtained by using a supercon-
was of high intensity with an increasing intensity ratio with
ducting Spectro-20000 (Goldstar, Seoul, Korea) operated at 2.0 T.
prolongation of TR and TE in all 1 0 cases (Fig. 2B). On heavily
Images were constructed by using two-dimensional Fourier transform
T2-weighted images, 2000/i 20, seven of 1 0 tumors were of
technique. A body coil with a field of view of 35 cm was used. The
image matrix of 1 80 x 256 elements yielded an in-plane spatial higher intensity than fat (Fig. 2C), while all 1 0 tumors were of
resolution of 1 .9 x 1 .4 mm. Two or four data acquisitions were higher intensity on SE 2000/i 50, 1 80 images (Figs. 1 B and
averaged for each image. A multisection/multiecho technique was 2D). Intensity ratios of the main tumor were 3.00 ± 1 .02 (SE
used to make axial sections with a section thickness of 8 mm and 2000/60), 7.1 3 ± 2.55 (SE 2000/i 20), i 1 .67 ± 3.91 (SE
intersection gap of 2 mm. Relatively Ti -weighted images, 500/30 2000/i 50), and 1 2.71 ± 3.3i (SE 2000/i 80). Cleftlike areas
(TR/TE), were obtained, as were T2-weighted images, 2000/60, i20, were of high intensity on T2-weighted images, 2000/60, in all
1 50, 1 80 (TR/TE). Neither respiratory nor ECG gating was used.
1 0 cases and could not be differentiated from the main part
CT scans were obtained by using a CT 9800 scanner (General
ofthe tumor. However, on heavily T2-weighted images, 2000/
Electric, Milwaukee, WI) with a scan speed of 2 sec. In all eight
1 50, 1 80, cleftlike areas showed higher intensity than the
patients, single-level bolus dynamic scans were obtained. Nonen-
hanced CT scans of the liver were obtained at 1 0-mm intervals to
main tumor did, and the two could be differentiated from each
localize the lesions. Single-level bolus dynamic CT scans were then other in all i 0 cases (Figs. 1 B and 2D). Internal septa had low
obtained at the level that contained the largest lesion. Contrast- intensity on all T2-weighted images (Figs. 2B-2D) and could
enhanced CT scans of the liver were obtained after IV bolus injection be differentiated from cleftlike areas, which showed high
of 60 ml of 60% iodinated contrast agent (Telebrix 30, Guerbet, intensity in all five cases.
Aulnay-Sous-Bois, France). The contrast agent was injected by hand
American Journal of Roentgenology 1989.152:1221-1226.

over a period of 15 sec by using a butterfly needle placed in a

peripheral vein. Scans at the preselected level were then acquired at Dynamic Bolus CT
5 sec, i5 sec, 30 sec, 1 mm, 2 mm, 3 mm, 5 mm, io mm, and iS mm
after the end of the injection. Delayed scans were obtained until no On nonenhanced CT scans, all 1 0 giant hemangiomas
further increase in lesion attenuation or degree of filling was discern- showed low-density areas compared with surrounding liver,
ible on serial delayed scans. and all tumors contained cleftlike areas of lower density than
MR images and dynamic bolus CT scans were interpreted at the the main tumor. The attenuation number of the cleftlike areas
same time for each patient, with emphasis on comparing the MR varied from +8 to +29 H. The cleftlike areas were round or
images and CT scans for evaluation of the internal architecture of the ovoid in five cases (Fig. 1 C), linear in three, and irregular in
tumor. For analysis of MR imaging, visual estimates of the intensity two (Fig. 2E). Spotty calcification was seen in the tumor in
of the hemangioma were made as follows: low intensity (intensity one case.
lower than that of surrounding liver), isointensity (intensity equal to
On dynamic, contrast-enhanced scans, contrast enhance-
that of surrounding liver), high intensity (intensity higher than that of
ment at the periphery ofthe tumor was visible in every patient.
surrounding liver), and very high intensity (intensity higher than that
of fat). In addition, measurements of the ratio of intensity between
Delayed imaging at the preselected reference levels showed
tumor and surrounding liver were obtained. A region of interest was further centripetal enhancement in each of the tumors (Figs.
circumscribed with a cursor on the displayed image. When possible, 1 D and 2F). However, complete filling of the tumor with
all intensity measurements were made with the same x coordinate contrast material was never observed. The cleftlike areas of
(i.e. , same y axis) to minimize artifacts. Intensity was recorded in lower density did not enhance completely on delayed scans
absolute numbers (mean intensity value). Intensity ratios of heman- (Figs. 1 D and 2H). The time interval from the end of injection
giomas to surrounding liver were calculated for all pulse sequences. to maximal observed filling was i 0-i 20 mm (mean, 32).
Dynamic CT scans were reviewed to determine the size, location,
and appearance of the hemangiomas on nonenhanced CT scans, as
well as changes that occurred after bolus contrast enhancement. Comparison of MR Images and Dynamic Bolus CT Scans

MR imaging was equal to dynamic bolus CT in delineating

Results the extent of the tumor and in diagnosing giant hemangiomas.
MR Imaging However, MR imaging was superior to dynamic bolus CT in
characterizing the internal architecture. Cleftlike areas of
All 1 0 hemangiomas had a heterogeneous appearance on lower density on dynamic bolus CT scans corresponded to
Ti - and T2-weighted images. All of the tumors comprised a the areas of lower intensity on Ti -weighted images and of
main tumor with cleftlike areas. Five also had internal septa. higher intensity on heavily T2-weighted images (Fig. 1) in all
On Ti -weighted MR images, the main part of the tumor 10 cases. Low-density areas of the tumor showing the most
was of high intensity in two cases (Fig. 1 A), isointensity in recent enhancement on delayed dynamic CT scans corre-
four cases (Fig. 2A), and low intensity in four cases. The sponded to low-intensity areas on Ti - and T2-weighted im-
intensity ratio of the main part of the tumor was 0.90 ± 0.i 5. ages that were thick internal septa of giant hemangiomas
Cleftlike areas were well defined and were of lower intensity (Figs. 2A, 2D, 2G, and 2H) in three cases. However, dynamic
than the main tumor. They were round or ovoid in five cases, bolus CT scans could not disclose fine internal septa of giant
linear in three, and irregular in two. Internal septa also were hemangiomas that were visualized clearly on MR images in
of lower intensity than the main tumor and could not be five cases.
 AJR:152, June 1989 GIANT CAVERNOUS LIVER HEMANGIOMAS 1223

Fig. 1.-Giant cavernous heman-

gioma of liver in 41-year-old woman
with indigestion.
A, Axial SE 500/30 MR image shows
main part of tumor with slightly high
intensity in entire right hepatic lobe.
Lower-intensity areas (arrows) are
seen in tumor.
B, Axial SE 2000/180 MR image
shows heterogeneous appearance of
tumor with smooth, well-defined mar-
gin. Main part of tumor shows very high
intensity. Round and ovoid cleftilke
areas (arrows) show higher intensity
than main part does.
C, Nonenhanced CT scan shows
ovoid tumor with low attenuation in
right hepatic lobe. Cleftlike areas of
lower attenuation (arrows) are seen
also.
D, Delayed CT scan 15 mm after
injection of contrast material shows
nearly complete filling of tumor. Cleft-
like areas (arrows), which are not en-
A B
hanced, correspond to cleftlike, low-
intensity areas in tumor on Ti-weighted . .; -;
MR image and higher-intensity areas in
tumor on heavily T2-weighted image.
American Journal of Roentgenology 1989.152:1221-1226.

. . :1-’l

Discussion hyalinization, liquefaction, and fibrosis [2, 4, i 4, 1 6, 19]. These

features cause a heterogeneous appearance of the tumor on
Although hemangiomas are the most common benign tu- MR imaging. Stark et al. [i i ] reported that none of the
mors of the liver, giant hemangiomas with clinical manifesta- hemangiomas in their series had heterogeneity on Ti -
tions are rare. Because of the relative infrequency of large weighted images. In our series, however, all hemangiomas
symptomatic hepatic hemangiomas, few reports have ap- had a heterogeneous appearance on Ti -weighted images as
peared in the radiologic literature [i 5, i 6]. The definition of well as on T2-weighted images. Ros et al. [i 4] also reported
giant hemangiomas is still uncertain. Edmondson and Peters that 80% of hemangiomas in their series were mnhomoge-
[i 7] described giant hemangiomas as tumors more than i 0 neous on T2-weighted images and all hemangiomas were
or i2 cm in greatest dimension. Adam et al. [i8] defined them inhomogeneous on cut sections of pathologic specimens.
as tumors measuring more than 4 cm in diameter, because The characteristic appearance of cavernous hemangiomas
no symptomatic tumor was encountered that was smaller on CT scans after bolus contrast enhancement has been
than this. However, all 22 tumors in the series of Adam et al. reported previously [4-9]. However, controversy still exists
exceeded 6 cm in at least one dimension. Thus, we defined concerning which CT criteria must be met for reliable diag-
giant hemangiomas as tumors that exceeded 6 cm in at least nosis of a hemangioma. The i 0 giant hemangiomas we report
one dimension. behaved similarly on dynamic CT scans. All tumors showed
In our series, all i 0 hemangiomas showed heterogeneous low-density areas on nonenhanced CT scans, and each con-
intensity on Ti - and T2-weighted MR images. The larger the tamed one or more cleftlike zones of lower density. All tumors
tumor, the more diverse were the internal components of the displayed early, peripheral enhancement and some degree of
tumor. Giant hemangiomas have a spectrum of histopatho- centnpetal filling. However, none became completely iso-
logic changes, including hemorrhage, thrombosis, extensive dense on delayed scans. Because of this, none of our patients
 i 224 CHOI ET AL. AJR:152, June 1989

Fig. 2.-Giant cavernous heman-

gioma of liver in 43-year-old woman
with eplgastric discomfort. Axial MR im-
ages (A-D) are on this page; postcon-
trast CT findings (E-H) are on facing
page.
A, SE 500/30 image shows main part
of tumor to be Isointense in entire right
hepatic lobe. Lower-intensity areas (ar-
rows, arrowheads, asterIsk) are seen In
tumor.
B, SE 2000/60 image shows hetero-
geneous appearance of tumor with
well-defined, lobulated margins. Irreg-
ular cleftlike areas (arrows) are of high
IntensIty. Thin (arrowheads) and thick
(asterisk) septa are of low intensity.
C, SE 2000/120 image shows very
high Intensity of main part of tumor and
irregular cIeftilke areas (arrows). Inter-
nal septa (arrowheads, asterisk) still
show low intensity.
D, SE 2000/180 image shows three
A B different parts of hemangloma: main
part of tumor, cleftlike part (arrows),
and internal septa (arrowheads, aster-
isk).
American Journal of Roentgenology 1989.152:1221-1226.

had tumors that fulfilled the criteria used by Freeny and Marks septa of hemangiomas could be explained as poorly cellular
[8] for a typical hemangioma. However, all would have been fibrous tissue. The areas of internal septa were of low intensity
acceptable under the conditions proposed by Ashida et al. on MR images on all pulse sequences. Fibrosis decreases
[9]. Our results are similar to those reported by Scatarige the water content and shows low intensity. The low-density
et al. [i5]. areas of hemangiomas with the most recent enhancement on
The cleftlike areas of hemangiomas noted on MR images delayed scans corresponded exactly to the areas of low
or CT scans may be due to areas of cystic degeneration or intensity on MR images. On the basis of an MR-pathologic
liquefaction, rather than fibrosis, thrombosis, or hemorrhage, correlation in three resected hemangiomas, Ros et al. [i 4]
as was postulated by several authors [4, i 5]. The reason for reported that nodular areas and septations of decreased
this is because the attenuation numbers of the cleftlike areas intensity within a hyperintense tumor on T2-weighted images
in all cases were less than +30 H, all of the cleftlike areas did corresponded to fibrotic nodular areas and fibrotic strands,
not show contrast enhancement on delayed scans, and the respectively. Muramatsu et al. [20] described the CT-patho-
cleftlike areas with lower density on dynamic bolus CT scans logic correlation of the fibrotic part of a hepatic tumor. On
corresponded to the areas with lower intensity than the main late-enhancing CT scans obtained more than 5 mm after
part of the tumor on Ti -weighted images and with higher administration of contrast material, the fibrotic part showed
intensity than the main part on heavily T2-weighted images. contrast enhancement because fibrous stromas retained con-
Takayasu et al. [i 91 also reported one case of giant heman- trast material longer than tumor tissue did. However, on CT
gioma of the liver with a central cavity that contained trans- scans in the early phase after a bolus administration of
parent serous fluid. Ros et al. [i 4] reported that some he- contrast material, it was difficult to differentiate tumor tissue
mangiomas contained cystic cavities filled with gelatinous from fibrotic connective tissue. Further studies with histologic
material on pathologic examination. The areas of internal correlation are needed because none of the hemangiomas in
 AJR:152, June 1989 GIANT CAVERNOUS LIVER HEMANGIOMAS 1225

Fig. 2.-E-H, Postcontrast CT

scans.
E, 5 sec after bolus injectIon of con-
trast material. Large, ovoid tumor with
low attenuation Is seen in right hepatic
lobe. Cleftlike areas of lower attenue-
tion (arrows) are seen also.
F, 15 mm after injection. Peripheral
contrast enhancement is seen.
G, Delayed scan 30 mm after injec-
tion shows isodense filling that spares
cleftlike areas (arrows) and central part
(asterisk), which correspond to central
low-intensity zone on MR Images. How-
ever, delineation of fine internal septa,
which are visualized cleariy on MR lm-
ages, is poor.
H, Delayed scan 50 mm after injec-
tion shows nearly complete filling of
mass except for cieftlike areas (ar-
rows), which are not enhanced, and
central zone (asterisk), which shows
late enhancement.
E F
American Journal of Roentgenology 1989.152:1221-1226.

our series were surgically resected and available for patho- large, well-defined, heterogeneous masses that contain areas
logic study. with an increasing intensity ratio with prolongation of TA and
Giant cavernous hemangiomas have a heterogeneous ap- TE, cleftlike areas of low intensity on Ti -weighted images
pearance on MR images. Therefore, there is sometimes con- and of higher intensity on heavily T2-weighted images, and
fusion between giant hemangiomas and malignant tumors. low-intensity internal septa on all pulse sequences.
Necrotic metastatic tumor and hepatoma are included in the
differential diagnosis of giant hemangiomas [i i]. The char- ACKNOWLEDGMENTS
acteristic MR findings of giant hemangiomas, such as well-
We thank Z. H. Cho and H. W. Park for assistance in manuscript
defined, high-intensity areas on T2-weighted images with
preparation.
increasing intensity ratios with prolongation of TR and TE,
are useful in differentiating these tumors from necrotic metas-
REFERENCES
tases or hepatomas. In addition, focal hemorrhagic necrotic
areas with high intensity on Ti -weighted images [1 1 , 2i] and 1 . Ishak KG, Rabin L. Benign tumors of the liver. Med Clin North Am
i975;59:995-1013
the presence of a capsule, target, or halo sign [21 -24], which
2. Ishak KG. Mesenchymal tumors of the liver. In: Okuda K, Peters RL, eds.
are encountered often in hepatomas or metastases, are not Hepatocellular carcinoma. New York: Wiley, 1976:275-286
seen in hemangiomas. 3. Cronan JJ, Esparza AR, Dorfman GS, Ridlen MS, Paolella LP. Cavernous
In summary, we believe that careful evaluation of the inter- hemangioma of the liver: role of percutaneous biopsy. Radiology 1988;
nal architecture of giant hemangiomas with MR imaging or 166:135-138
4. Johnson CM, Sheedy PF, Stanson AW, et al. Computed tomography and
dynamic bolus CT makes it possible to facilitate the correct
angiography of cavemous hemangiomas of the liver. Radiology 1981;
diagnosis in these tumors. The diagnosis of giant cavernous 138:115-1 21
hemangioma should be considered when MR images show 5. ltd Y, Ohtomo K, Araki T, Furui 5, ho M, Atomi Y. Computed tomography
 i 226 CHOI ET AL. AJR:152, June 1989

and sonography of cavemous hemangioma of the liver. AiR 1983; heterogeneous appearance on T2-weighted images. AiR 1987;149: 1167-
141:315-320 1170
6. Itai V. Furui 5, Araki T, Yashiro N, Tasaka A. Computed tomography of 15. Scatarige JC, Kenny JM, Fishman EK, Herlong FH, Siegelman 55. CT of
cavernous hemangioma of the liver. Radiology 1980;1 37: 149-i 55 giant cavernous hernangiorna. AiR 1987;149:83-85
7. Freeny PC, Vimont TR, Bamett DC. Cavemous hemangioma of the liver: 16. Olmsted WW, Stocker Sr. Cavernous hernangioma of the liver. Radiology
ultrasonography, arteriography, and computed tomography. Radiology 1975;1 17:59-62
1979;132:143-148 17. Edmondson HA, Peters RL. Neoplasms of the liver. In: Schiff L, Schiff ET,
8. Freeny PC, Marks WM. Hepatic hemangioma: dynamic bolus CT. AJR eds. Diseases of the liver. Philadelphia: Lippincott, 1982:1144-1145
1986;147:711-719 18. Adam YG, Huvos AG, Fortner JG. Giant hemangiornas of the liver. Ann
9. Ashida C, Fishman EK, Zerhouni EA, Herlong FH, Siegeiman SS. Corn- Surg 1970;172:230-245
puted tomography of hepatic cavemous hemangioma. J Comput Assist 19. Takayasu K, Moriyama N, Shima Y, et aJ. Atypical radiographic findings in
Tomogr 1987;1 1:455-460 hepatic cavernous hemangioma: correlation with histologic features. AJR
10. Glazer GM, Aisen AM, Francis IR, Gyves JW, Lande I, Aider DD. Hepatic 1986;146:1149-1i53
cavernous hemangioma: magnetic resonance imaging. Radiology 1985; 20. Murarnatsu Y, Takayasu K, Morlyama N, et al. Peripheral low-density area
155:417-420 of hepatic tumors: CT-pathologic correlation. Radiology 1986;160:49-52
11. Stark DD, Felder RC, Wittenberg J, et al. Magnetic resonance imaging of 21 . Wittenberg J, Stark DD, Forman BH, et al. Differentiation of hepatic
cavernous hernangioma of the liver: tissue specific characterization. AJR metastases from hepatic hemangiomas and cysts by using MR imaging.
1985;145:213-222 AJR 1988;151:79-84
12. Itai V. OhtornoK, Furui 5, Yarnauchi T, Minami M, Yashiro N. Noninvasive 22. Ebara M, Ohto M, Watanabe Y, et al. Diagnosis of small hepatocehlular
diagnosis of small cavernous hemangioma of the liver: advantage of MRI. carcinoma: correlation of MR imaging and tumor histologic studies. Ra-
AJR 1985;145: 1195-1199 diology 1986;159:371-377
13. Ohtomo K, Itai Y, Furui 5, Yashiro N, Yoshikawa K, ho M. Hepatic tumors: 23. Itoh K, Nishimura K, Togashi K, et al. Hepatoceilular carcinoma: MR
differentiation by transverse relaxation time (T2) of magnetic resonance imaging. Radiology 1987;164:21-25
imaging. Radiology 1985;155:421 -423 24. Itai Y, Ohtomo K, Furui 5, Minarni M, Yoshikawa K, Yashiro N. MR imaging
14. Ros PR, Lubbers PR, Olmsted WW, Morillo G. Hemangioma of the liver: of hepatocellular carcinoma. J Comput Assist Tomogr 1986;10:963-968
American Journal of Roentgenology 1989.152:1221-1226.