AJA:159,

SONOGRAPHY

November 1992

our series were correctly classified as benign they did not have a halo (Fig. i 0).

Six malignant

tumors

(two hepatocellular

lesions

because

carcinomas,

four

lesions because et al. [8] found

a peripheral

hypoechoic

rim in approximately

hepatocellular hepatocellular

carcinomas. carcinomas

In our study, only two (33%) of six had no hypoechoic halo. Only four

(9%) of 43 liver metastases

(two colonic

50%

of all

adenocarcinomas,

one pancreatic adenocarcinoma, one renal cell carcinoma) did not have a hypoechoic halo (Fig. i 1). Thus, the hypoechoic halo is a frequent and important sonographic feature of isoechoic or hyperechoic liver metastases (Figs. 2-4). Our histopathologic analysis of the halo [4] showed that it correlated with an intratumoral rim of proliferating tumor cells.

A recent

review

[9] on imaging

of liver tumors

Even in the presence

of a characteristic

hemangioma

type of contrast enhancement, the probability that a liver lesion in a patient with a known neoplasm represents a hemangioma is only 86% [i 0]. MR imaging is said to have an 85% accuracy in distinguishing hemangiomas from metastases [9]. Several studies have shown that the quantitative data (e.g., Ti and T2 values, lesion-to-liver signal-intensity ratios) of cavernous hemangiomas as a group are significantly different from those of metastases or hepatocellular carcinomas. However, a significant overlap occurs between the values of individual benign

lesions and the values of malignant

ones; thus, it is often not

possible to identify a particular lesion [1 1 ]. In a recent trial by Itoh et al. [1 2], the lesion/liver signal-intensity ratio, the best quantitative criterion for distinguishing benign from malignant lesions, showed an overlap in 1 6 (31 %) of 51 hemangiomas and in i4 (34%) of 41 metastases. To our knowledge, no prospective randomized study has been conducted to evalu-

ate the usefulness

of sonography,

It is important

1009

to emphasize

that we evaluated

the useful-

benign from malignant

isoechoic and hyperechoic lesions. Hypoechoic liver lesions, which generally are considered to be highly suggestive of a malignant tumor, were not included. One shortcoming in our study is the retrospective analysis of selected liver lesions with confirmed diagnoses. For this reason, our preliminary results on the accuracy of the halo sign for distinguishing malignant from benign lesions should be validated in a pro-

spective

randomized

study.

In our study,

the proportion

of

hemangiomas with atypical sonographic features was unusually high (45%). Thus, under normal conditions with a nonselected patient group, the halo sign may actually provide better results for distinguishing hemangiomas from metastases.

concludes

with a statement that sonography is of limited value as it cannot be used to distinguish benign from malignant lesions. This can, however, also be said for other imaging techniques such as CT and MR. In a CT study of liver lesions by Freeny et al. [i 0], only 54% of hemangiomas had a typical pattern of contrast enhancement; the remaining 46% of hemangiomas had the same pattern of enhancement as 77% of liver metas-

tases.

TUMORS

ness of the halo sign in distinguishing

‘

metastases) were falsely classified as benign they did not have a halo (Fig. 1 1). Shinagawa

Downloaded from www.ajronline.org by 222.185.140.142 on 10/25/15 from IP address 222.185.140.142. Copyright ARRS. For personal use only; all rights reserved

OF LIVER

CT, and MR in the differ-

ential diagnosis of hepatic lesions. Thus, no objective scientific proof of the advantages of MR or CT over sonography in the differential diagnosis of hepatic tumors currently exists.

REFERENCES 1 . Weill

FS. Ultrasound

diagnosis

of digestive

diseases,

3rd ed. New

York:

Springer, 1989:121-148 2. Wemecke K, Peters PE. Sonographische

und computertomographische Diagnostik von Lebermetastasen. Radiologe 1985;25: 141-151 3. Schwerk WB, Schmitz-Moormann P. tJtrasonically guided fine-needle biopsies in neoplastic liver disease: cytohistologic diagnoses and echo pattern of lesions. Cancer 1981;48:1469-1477

4. Wemecke

K, Henke L, Vassallo P, et al. Pathologic explanation

for hype-

echoic halo seen on sonograms of malignant liver tumors: an in vitro correlative study. AJR 1992;159: 101 1-1 016 5. Onodera H, Ohta K, Oikawa M, et al. Correlation of the real-time ultrasonographic appearance of hepatic hemangiomas with angiography. JCU J Clin Ultrasound 1983;1 1 :421 -425 6. Takayasu K, Moriyama N, Shima Y, et al. Atypical radiographic findings in hepatic cavernous hemangioma: correlation with histologic features. AiR 1986;146:1 149-1153 7. Quinn SF, Gosink BB. Characteristic sonographic signs of hepatic fatty infiltration. AJR 1985;145:753-755

8. Shinagawa T, Ohto M, Kimura K, et al. Diagnosis and clinical features of small hepatocellular carcinoma with emphasis on the utility of real-time ultrasonography: a study of 51 patients. Gastroentenology 1984;86: 495-502 9. Ferrucci JT. Liver tumor imaging: current concepts. AJR 1990;1 55: 473-484 10. Freeny PC, Marks WM, Aayan JA, Bolen JW. Colorectal carcinoma eval-

11.

uation rence. Reinig

with CT: preoperative staging and detection of postoperative Radiology 1986;160:613-618 JW. Difterentiation of hepatic lesions with MR imaging: word? Radiology 1991;179:601 -602

recurthe

last

12. Itoh K, Saini 5, Hahn PF, Imam N, Ferrucci JT. Differentiation between small hepatic hemangiomas and metastases on MR images: importance of size-specific

quantitative

criteria.

AJR 1990;1 55:61-66

Downloaded from www.ajronline.org by 222.185.140.142 on 10/25/15 from IP address 222.185.140.142. Copyright ARRS. For personal use only; all rights reserved

Downloaded from www.ajronline.org by 222.185.140.142 on 10/25/15 from IP address 222.185.140.142. Copyright ARRS. For personal use only; all rights reserved

1 Oil

Pathologic Explanation for Hypoechoic Halo Seen on Sonograms of Malignant Liver Tumors: An In Vitro Correlative Study i.

Wemecki’

Karl

-1

,,

‘

t 1

“&L

OBJECTiVE. the hypoechoic

Ludger Henke1 Pierre Vassallo1 Dirk Barthold von Bassewitz2 Stefan Diedench1 Peter E. Peters1

Georg Edel2

The purpose of this study was to evaluate the morphologic substrate of halo seen on sonograms of malignant liver tumors. MATERIALS AND METhODS. We used sonograms and pathologic examinations to evaluate 17 cadaverlc livers with macroscopic tumors (three primary liver tumors, 14 metastases). During sonography (3.5 and 5.0 MHz), a representative section plane was marked, and the same section was examined histologically. Emphasis was placed on the architecture of the tumor and the morphology of the periphery of the tumor that could account for the hypoechoic halo seen on sonograms. RESULTS. In 13 of 17 hepatic tumors, a hypoechoic halo was detected on sonograms. Histopathologic examination showed an lntratumoral rim consisting of proliferating tumor cells In 12 cases and an extratumoral rim of compressed liver parenchyma In all 13 cases. A detailed comparison of sonographic and hlstopathologic findings showed that the hypoechoic halo corresponded to a greater concentration of tumor cells and areas of less marked fibrosis and necrosis in the periphery of the tumors. This occurred in 11 cases. In one case, histologic studies showed that the hypoechoic rim was caused by compressed liver parenchyma. In another case, the hypoechoic halo was caused by intratumoral (cellular peripheral zone of tumor) and extratumoral (compressed liver parenchyma) components. All four tumors without a halo at sonography were uniform histologically.

CONCLUSION. The sonographic to be caused predominantly the lesion.

seems

AJR

159:1011-1016,

Sonograms rim known as malignant liver visualized with suggestive of only when the authors [3, 4] in the periphery

Aeceived February vision May 20, 1992. I

Institute

6, 1992; accepted

of Clinical

Aadiology,

after re-

University

November

halo seen on sonograms by a zone of proliferating

of malignant liver tumors tumor in the periphery of

1992

of malignant liver tumors frequently show a peripheral hypoechoic the halo sign [1]. Although the exact frequency of this finding in tumors has not been determined and a hypoechoic rim can also be benign hepatic lesions, it is generally accepted that the halo sign is a malignant liver tumor [2]. The hypoechoic halo can be detected tumor is isoechoic or hyperechoic compared with the liver. Previous have suggested that the halo may be caused by intratumoral tissue of the tumor,

whereas

others

[5] have thought

that the finding

is

due to extratumoral tissue in the adjacent liver. To clarify the pathologic basis for the sonographic findings, we compared postmortem sonograms and histopathologic sections of 18 liver tumors.

of

MUnster Medical School, Albert-Schwaltzer-Str. 33, 4400 MUnster, Germany. Address reprint requests

Materials

to K. Wemecke. 2 Institute of Pathology,

Sonograms of 35 cadaveric livers with macroscopic tumors were obtained 3-4 hr after autopsy. Before resection of the liver specimens at autopsy, the hepatic vessels and the common hepatic duct were ligated in order to keep air out of the liver. Sonograms were inadequate in 14 of the 35 livers because of intrahepatic air bubbles due to autolysis or incomplete ligation of the hepatic vessels. Four livers with multiple, dissemi-

Medical

School,

of MUnster

4400 MOnster, Germany,

0361-803X/92/1595-101 C American

Lkiiversity

Roentgen

1 Ray SOCiety

and

Methods

Downloaded from www.ajronline.org by 222.185.140.142 on 10/25/15 from IP address 222.185.140.142. Copyright ARRS. For personal use only; all rights reserved

i Oi 2

WERNECKE

ET AL.

AJR:159,

nated, confluent metastases were excluded because a detailed cornparison of sonographic and histopathologic sections of a single lesion was not possible. The study included 17 tumors in 17 liver specimens. The tumors measured 1 .0-1 0.5 cm in diameter, with 9 lesions smaller than 3 cm (Table 1). The histologic diagnoses of the tumors (three primary liver tumors, 14 hepatic metastases) are listed in Table 1.

and the internal

Sonographic

Histopathologic

Finally,

Sonograms were obtained with 3.5- and 5.0-MHz linear-array (LS 7000, Picker International, Munich, Germany). The liver specimen was placed in a bath containing physiologic (0.9%) saline. were

classified

as

hypoechoic,

isoechoic,

After

of liver tissue containing formalin for 24 hr. This preliminary fixation facilitated cutting the specimen in the selected section plane. After fixation, the slices were placed in 70% alcohol for 30 mm and then photographed. For histologic study, the slices were fixed in 5% formalin for 8 weeks, embedded in paraffin wax, and stained with hematoxylin and eosin, elastica, periodic acid-Schiff

or hyper-

depending on their echogenicity in relation to the surrounding liver parenchyma, and the presence or absence of a surrounding hypoechoic halo was recorded. A hypoechoic halo was defined as an

area in the periphery of the tumor that was hypoechoic compared the

regardless

center of the

of

the

lesion

thickness

and

of the

TABLE Hepatic

surrounding rim.

The

liver

inner

1: Correlation Tumor

and

of Sonographic

contours

and Histopathologic

Sonographic

Case No.

Histologsc Diagnosis

examination,

was excised

a block

and fixed in

5%

(PAS), Giemsa, Ladewig, and Prussian blue stains. The macroscopic, unstained sections of the liver were analyzed for the following: outer contour of the liver tumor; presence or

parenchyma, outer

Findings

in 17 Cases

of Malignant

Findings

Tumor

Thickness

Echogenicity

Diameter (cm)

of Halo (mm)

Intratumoral

Extratumoral

1

Breast carcinoma

lsoechoic

1 .3

3.0

3.0

2

Bronchogenic ade-

lsoechoic

1.3

3.0

3.0

1.0

3

nocarcinoma Hepatocellular car-

Isoechoic

2.3

8.0

8.0

1.0

Isoechoic

3.5

3.0

3.0

lsoechoic

3.8

8.0

Hyperechoic

8.0

8.0

3.0

10.5

4.0

4.0