Professional Documents
Culture Documents
Hepatocellular Carcinoma:
Correlation ofMR Imaging
and Histopathologic Findings’
820 Radiology
#{149} June 1992
crotic. Two of these three lesions were
hypointense and one was hyperin-
tense on Ti-weighted images. Two of
the three lesions were hypointense on
T2-weighted images (Fig 3) and one
(of liquefied necrosis) was hyperin-
tense.
Intracellularfat deposits and tumor
grade versus signal intensity.-Signal
intensity was analyzed in the 67 non-
necrotic tumors. Six of the 67 tumors
(9%) contained fat in all or most of the
tumor cells. All six tumors with fat
deposits were hyperintense on Ti-
weighted images (Fig 4). On T2-
weighted images, five lesions were
hyperintense and one lesion was
isointense (Table 4).
Figure 2. HCC with mosaic pattern. (a) Histologic macrosection (H-E stain) reveals intratu- Six of the HCCs were classified as
moral septa as well as capsule. (b) T2-weighted MR image (SE 2,500/80) shows two compart- grade 1 tumors (well-differentiated
ments with different degrees of hyperintensity. Arrow = tumor. HCCs) with the system used by Ed-
mondson and Steiner (6). All six grade
i tumors, regardless of whether fat
deposits were present in the cells,
Table 2 were hyperintense on Ti-weighted
Mosaic Patterns images. Classical HCCs defined as
Presence of Mosaic Presence of Mosaic grade 2 or above had various signal
Pattern at Pattern at intensities on Ti-weighted images.
Tumor Ti-weighted Imaging T2-weighted Imaging
Diameter No. of
Frequency of hyperintensity on Ti-
(cm) Lesions Positive Negative Positive Negative weighted images was significantly
higher (x2 P < .05) in grade 1 tumors
<2.0 4 1 3 0 4 (iOO%; six of six tumors) than in
2.Oto<3.0 7 3 4 4 3
3.0 25 8 17 23 2 grades 2 and 3 tumors (28%; i7 of 6i
tumors). On T2-weighted images,
Total 36 12 24 27 9
four (67%) of the six well-differenti-
ated HCCs appeared isointense and
two (33%) appeared hyperintense
(Fig 5), whereas all grades 2-4 tumors
(4.6 mm ± 1.0) were demonstrated as at least 3.0 cm in diameter (McNemar were hyperintense on T2-weighted
double-layered rings (hypointense test, P < .OOi; 95% confidence inter- images. Frequency of hyperintensity
inner layer and hyperintense outer val, 0.40-0.80). on T2-weighted images was signifi-
layer) (Fig ic). cantly higher (x2 P < .05) in grades
2-4 tumors (100%; 6i of 61 tumors)
than in grade i tumors (33%; two of
Mosaic Pattern Correlation between Signal
six tumors).
Intensity and Histopathologic
A mosaic pattern was grossly iden- Structural pattern versus signal inten-
Findings
tified in 36 of 72 tumors (50%) (Fig sity.-The correlation between the
2a), including 25 of the 30 tumors 3.0 Two of the 72 tumors were ex- structural pattern and signal intensity
cm or more in diameter (83%), seven cluded from the assessment of signal of tumors was assessed in 6i lesions
of the i6 tumors between 2.0 and 3.0 intensity (one was associated with (excluding the tumors associated with
cm in diameter (44%), and four of 26 diffuse fatty liver and the other with necrosis or fatty metamorphosis) (Ta-
tumors smaller than 2.0 cm (i5%) underlying hemosiderosis). ble 5). A trabecular pattern was domi-
(Table 2). Intratumoral linear-like struc- Tumor size versus signal intensity.- nant in 59 of the 6i tumors (97%).
lures of hypointensity were demon- Tumors revealed various signal inten- These tumors had various signal in-
strated in 12 of the 36 tumors (33%) sities on Ti-weighted images (hypoin- tensities on Ti-weighted images; 56
on Ti-weighted images. On T2- tense in 29 of the 70 cases [4i%J, were hyperintense and three were
weighted images, 27 of the 36 tumors isointense in 17 [%1, and hyperin- isointense on T2-weighted images. A
(75%) showed nonuniform signal in- tense in 24 [34%]). In contrast, 64 of pseudoglandular pattern was domi-
tensity (Fig 2b). A mosaic pattern was the tumors (9i%) appeared hyperin- nant in only two lesions, one of which
depicted in 23 of the 30 tumors 3.0 cm tense on T2-weighted images. How- was hypointense and the other isoin-
or more in diameter (77%). In con- ever, the difference in signal inten- tense on Ti-weighted images. The
trast, only one of the four mosaic sity was not significant statistically difference in signal intensity be-
patterns in tumors smaller than 2.0 (x2 J > .05) for the tumor size ( < 2.0 tween the trabecular and pseudoglan-
cm (25%) was demonstrated on Ti- vs 2.0 cm) for either Ti- or T2- dular types was not statistically signif-
weighted images. The difference in weighted images (Table 3). icant (x2 P > .05) for either Ti- or T2-
the detection rate of mosaic patterns Tumor necrosis versus signal inten- weighted images. Although the two
with Ti- and T2-weighted imaging sity.-At histologic examination, three pseudoglandular-type lesions were
was statistically significant for tumors tumors were almost completely ne- hyperintense on T2-weighted images,
portion of the tumor, and the hemor- Figure 3. Necrotic HCC. (a) T2-weighted image (SE 2,500/80) shows tumor as a hypointense
lesion (arrow). (b) Histologic section shows almost complete necrosis of the tumor (T). C =
rhages did not affect the classification
capsule, L = surrounding liver parenchyma. (H-E stain; original magnification, xSO.)
of the signal intensities of the tumors
(Fig i).
Peliotic change of intratumoral sinus-
oid versus signal intensity-Twenty- perintensity on T2-weighted images with abundant small vessels and
two of the 72 tumors (3i%) showed and revealed no definite peliotic newly formed bile ducts (Fig ia). It is,
peliotic change of intratumoral sinus- change; the peliotic change was therefore, reasonable that the cap-
oid. Conversely, localized spotty or present in only one of the other four sules were demonstrated as single
tubular foci with marked hyperinten- tumors with localized hyperintense hypointense rings surrounding tu-
sity were identified in 32 of the 72 tu- foci on T2-weighted images. mors on Ti-weighted images (SE 500/
mors (44%) on T2-weighted images 20) (Fig ib) and as double-layered
(Table 6). The peliotic change was rings (hypointense inner layer and
DISCUSSION
identified at histologic examination in hyperintense outer layer) on T2-
2i of these 32 tumors (66%) (Fig 7). In One of the gross pathologic charac- weighted images (SE 2,500/80) (Fig
contrast, only one of the 40 tumors teristics of HCC is a capsule surround- ic) (4). The thin capsules were, how-
without localized hyperintense foci ing the tumor (5). Transcatheter arte- ever, demonstrated as a hypointense
on T2-weighted images revealed the rial embolization has been reported to ring on T2-weighted images even
peliotic change; signal intensity voids be more effective in encapsulated though the inner layers were thinner
were demonstrated on T2-weighted than uncapsulated HCCs (9). There- than the outer layers. This might be
images in this tumor. The correlation fore, the delineation of the capsule is due to paucity of vessels and bile
between peliotic change and localized important not only for the differential ducts in the thin outer layers relative
hyperintense foci on T2-weighted diagnosis but also from a therapeutic to the thick layers.
images was statistically significant (x2 viewpoint. In this study, as in previ- Another gross pathologic character-
P < .05). ous studies (2-4), Ti-weighted imag- istic of HCC is a mosaic appearance,
From the viewpoint of tumor size, ing was very sensitive in the detection caused by the presence of intratu-
28 of 46 tumors (6i%) at least 2.0 cm of capsules and is considered to be moral septa and/or histologic variety
in diameter had localized hyperin- useful in the differential diagnosis, within the tumor (5) (Fig 2a, Table 2).
tense foci on T2-weighted images, especially in tumors larger than 2.0 Ti-weighted imaging was not sensi-
and 20 of those 28 (7i%) revealed the cm in diameter (Table i). tive in the depiction of the mosaic
peliotic change. The peliotic change The visibility of a tumor capsule at pattern because the septa were too
was significantly (x2. J) < .05) corre- MR imaging is thought to be influ- thin to be demonstrated and most of
lated with localized hypermtense foci enced by its thickness and structure. the tumors appeared homogeneous
on T2-weighted images. Conversely, The capsules were composed of two despite the histologic inhomogeneity
22 of the 26 tumors smaller than 2.0 layers: an inner layer rich in fibrous of the tumors. In contrast, T2-
cm (85%) were of homogeneous hy- tissue and an outer, water-rich layer weighted imaging demonstrated the
822 Radiology
#{149} lune 1992
Table 4
Correlation between Tumor Grade, Fatty Metamorphosis, and Signal Intensity in 67 Nonnecrotic Tumors
No. of Lesions Signal Inten sity on Ti-weig hted Images Signal Inten sity on T2-weig hted Images
Tumor No. of with Fatty
Grade Lesions Metamorphosis Hypointense Isointense Hyperintense Hypointense Isointense Hyperintense
1 6 3 0 0 6* 0 4 2
land2 4 0 1 1 2 0 0 4
2 46 3 19 14 i3* 0 0 46
2and3 7 0 4 2 1 0 0 7
3 4 0 3 0 i 0 0 4
“?
... .. f
,#{149}_ ‘
*t.,,_ .4.
. .- :
:s ,..,-, ‘...
‘ .
;.;,
#{149}; .‘#{149}‘..
,,#{149} 4,., , .1
1:.#{149}
#{149}j.’.#{149} -
- ., #{149}I#{149}, - . . #{149}4
a. b. c.
Figure 5. Well-differentiated HCC without fat deposition. (a) Ti-weighted MR image (SE 500/20) shows hyperintense tumor (arrow). (b) T2-
weighted MR image (SE 2,500/80) shows isointense tumor. (c) Histologic section shows grade 1 HCC and no definite fat deposits. (H-E stain;
original magnification, x500.)
Table 5
Correlation between Structural Patterns of HCCs and Signal Intensity
Dominant Signal Inte nsity on Ti-weig hted Images Signal Inte nsity on T2-weig hted Images
Structural No. of
Pattern Lesions Hypointense Isointense Hyperintense Hypointense Isointense Hyperintense
Trabecular 59 (97) 26 16 i7 0 3 56
Pseudoglandular 2 (3) 1 1 0 0 0 2
Total 61 27 17 17 0 3 58
Note-A compact or scirrhous pattern was not dominant in any tumors. Numbers in parentheses are percentages.
824 #{149}
Radiology June 1992
4. Itoh K, Nishimura K, Togashi K, et al.
Table 6 Hepatocellular carcinoma: MR imaging.
Correlation of Peliotic Change and Localized Hypenntensity on T2-weighted Radiology 1987; 164:21-25.
SE Images 5. Kojiro M, Nakashima T. Pathology of
hepatocellular carcinoma. In: Okuda K,
Presence of Presence of Localized Ishak KG, eds. Neoplasms of the liver. To-
Tumor Peliotic Change Hyperintense Foci kyo: Springer, 1987; 81-104.
Diameter No. of 6. Edmondson HA, Steiner PE. Primary car-
(cm) Lesions Positive Negative Positive Negative cinoma of the liver: a study of 100 cases
among 48,900 necropsies. Cancer 1954;
<2.0 26 1 25 4 22 7:462-503.
2.Oto <3.0 16 3 13 7 9 7. Gibson JB. Histological typing of tumors
3.0 30 18 12 21 9 of the livers, biliary tract and pancreas. In:
Total 72 22 50 32 40 International histological classification of
tumors. Vol 20. Geneva: World Health Or-
ganization, 1978.
8. Dwyer AJ. Matchmaking and McNemar
in the comparison of diagnostic modalities.
Radiology 1991; 178:328-330.
9. Sakurai M, Okumura J, Kuroda C. Trans-
catheter chemoembolization effective for
treating hepatocellular carcinoma: a his-
topathologic study. Cancer 1984; 54:387-
392.
10. Mitchell DC, Burk D Jr, Vinitski S, et al.
The biophysical basis of tissue contrast in
extracranial MR imaging. AJR 1987; 149:
831-837.
11. Matsui 0, Kadoya M, Kameyama T, et al.
Adenomatous hyperplastic nodules in the
cirrhotic liver: differentiation from hepato-
cellular carcinoma with MR imaging. Radi-
ology 1989; i73:123-126.
12. Gabata T, Matsui 0, Kadoya M, et al. MR
imaging of hepatic adenoma. AJR 1990;
155:1009-1011.
13. Johnson GA, Herfkins RJ, Brown AA. Tis-
sue relaxation time: in vivo field depen-
dence. Radiology 1985; 156:805-810.
Figure 7. HCC with peliotic change of intratumoral sinusoid. (a) T2-weighted MR image 14. Stark DD, Goldberg HI, Moss AA, et al.
(SE 2,500/80) demonstrates hyperintense foci (arrows) in HCC. (b) Histologic section reveals Chronic liver disease: evaluation by mag-
dilated sinusoids (peliotic change) in the tumor. (H-E stain; original magnification, xSOO.) netic resonance. Radiology 1984; 150:149-
151.
15. Dixon WT. Simple proton spectroscopic
imaging. Radiology 1984; 153:189-194.
tensity within low-signal-intensity cannot be correlated with peliotic 16. Wehrli FW, Perkins TG, Shimakawa A, et
nodules on clinical MR images. changes in the tumors, but they are al. Chemical shift-induced amplitude
We found a correlation between thought to be the major cause of lo- modulations in images obtained with gra-
dient refocusing. Magn Reson Imaging
peliotic changes of intratumoral si- calized hyperintense foci seen on T2- 1987; 5:157-158.
nusoids and localized hyperintense weighted images, especially in tumors 17. Ohtomo K, Itai Y, Minami M, et al. Hepa-
foci on T2-weighted images (Table 6). larger than 2.0 cm in diameter. tocellular carcinoma: MR appearance mim-
In general, vessels with fast flow icking cavernous hemangioma. J Comput
Assist Tomogr 1990; 14:650-652.
show signal intensity loss on T2- Acknowledgments: We express our deep ap-
18. Gomori JM, Grossman RI. Head and neck
weighted images (iO). In this study, preciation to our colleagues Jun Yoshikawa,
hemorrhage. In: Kressel HY, ed. Magnetic
MD, Kazunori Arai, MD, Toshifumi Gabata, MD,
loss of signal intensity was demon- resonance annual 1987. New York: Raven,
Kazuhiko Ueda, MD, Ken-ichi Kobayashi, MD,
strated in only one tumor, despite the 1987; 71-112.
Masashi Unoura, MD, Ryohei Izumi, MD, Kazuo
19. Terada T, Kadoya M, Nakanuma Y, et al.
fact that a flow compensation tech- Hashimoto, MD, Masashi Miura, MD, and Tsuy-
Iron-accumulating adenomatous hyper-
nique was not used. The flow of di- oshi Mitsui, MD, for their valuable contribu-
plastic nodule with malignant foci in the
tions. We also thank Hayumi Dejima and John
lated sinusoids in the tumors seemed cirrhotic liver: histopathologic, quantitative
Gelbium for their kind assistance in preparing
to be different from that of the sup- iron, and magnetic resonance imaging: in
the manuscript.
vitro studies. Cancer 1990; 65:1994-2000.
plying artery and was rather slow.
20. Mitchell DC, Rubin R, Siegelman ES, et al.
Because cavernous hemangiomas are References Hepatocellular carcinoma with siderotic
demonstrated as markedly hyperin- 1. Matsui 0, Kadoya M, Kameyama T, et al. regenerative nodules: appearance as a
tense foci on T2-weighted images Imaging diagnosis of hepatocellular carci- nodule within a nodule on MR images. Ra-
(2i), peliotic changes of intratumoral nomas. Jpn J Cancer Chemother 1989; 16: diology 1991; 178:101-103.
25-33. [Japanesel 21. Stark DD, Felder RC, WittenbergJ, et al.
sinusoids are believed to be hyperin-
2. Ebara M, Ohto M, Watanabe Y, et al. Di- Magnetic resonance imaging of cavernous
tense on T2-weighted images because agnosis of small hepatocellular carcinoma: hemangiomas of the liver: tissue-specific
of the slow flow of blood in them (Fig correlation of MR imaging and tumor his- characterization. AJR 1985; 145:213-222.
7). All localized hyperintense foci in tologic studies. Radiology 1986; 159:371-
377.
the tumors on T2-weighted images
3. Itai Y, Ohtomo K, Furui S, et al. MR imag-
ing of hepatocellular carcinoma. J Comput
Assist Tomogr 1986; 10:963-968.