isoechoic and hyperechoic liver lesions on sonography. MATERIALS. AND METHODS. Sonograms of the liver in 50 patients with proved benign liver tumors.
1005
The Distinction Between Benign and Malignant Liver Tumors on Sonography: Value of a Hypoechoic Halo
r
OBJECTIVE.
Karl Wernecke1 Pierre
The purpose
sonographic
Vassallo
Ulrich Bick Stefan Diederich Peter E. Peters
halo
distinguishing on sonography. MATERIALS
sign
of this study
(defined
between
‘,:,
benign
was to determine the diagnostic value of the rim in the periphery of a lesion) in and malignant isoechoic and hyperechoic liver lesions as
any
hypoechoic
AND METHODS. Sonograms of the liver in 50 patients with proved benign in 50 patients with proved malignant liver tumors (seven primary liver neoplasms, 43 metastases) selected during a 13-month period were retrospectively analyzed by four radiologists who had no knowledge of the patients’ clinical findings or the final diagnoses. Only a single sonogram was studied in each case. The presence or absence of a hypoechoic halo on the sonogram was the only criterion for distinguishing malignant from benign hepatic lesions. RESULTS. For 95 of 100 hepatlc lesions, the four radiologists were almost (three vs one) or completely (four vs zero) in agreement about the presence or absence of a hypoechoic halo In the five cases where there were conflicting decisions (two vs two), a final decision (four vs zero) was achieved by reviewing the entire series of sonographic images. A halo could be detected in 44 malignant tumors (88%) and in only seven benign tumors (14%) (sensitivity, 88%; specificity, 86%; posItive and negative predictive values, 86% and 88%, respectively). The sonographic halo sign was particularly helpful in distinguishing hemangiomas (n = 29) from metastases (n = 43)(positive and negative predictive values, 95% and 87%, respectively). CONCLUSION. The results of this study suggest that the halo sign on sonograms is useful to distinguish benign from malignant isoechoic or hyperechoic tumors. liver
tumors
AJR
and
November
159:1005-1009,
1992
Despite continues
the increasing availability of CT and MR imaging units, sonography to play an important role as a primary screening and surveillance
technique
in hepatic
imaging.
Liver
tumors
have a broad
spectrum
of sonographic
features, which, in experienced hands, may help reduce the number of entities considered in the differential diagnosis [1 2]. However, hyperechoic and isoechoic ,
liver lesions
are difficult
to classify
because
both
metastases
and hemangiomas,
the most common malignant and benign liver lesions, respectively, may show these two types of echogenicity. Unlike hemangiomas, isoechoic and hyperechoic metastases are often surrounded by a hypoechoic halo [3]. In a sonographic-histopathologic correlation study [41, we observed that this hypoechoic halo corresponded primarily to an intratumoral rim of proliferating tumor cells. However, the value of Received February vision May 22, 1992. I
6, 1992; accepted
after re-
All authors: Institute of Clinical Aadiology,
1ki-
versity of MUnster Medical School, AlbertSchweitzer-Str. 33, 4400 MUnster, Germany. Address reprint requests to K. Wemecke.
0361-803X/92/1 C American
595-1005
Roentgen
Ray Society
this sonographic feature in distinguishing benign from malignant lesions is as yet uncertain, as its exact frequency in malignant liver tumors is unknown. In addition,
benign lesions sometimes In order
to determine
differentiating
benign
sonograms
diagnostic
from malignant
of 50 proved
tively analyzed
have a hypoechoic the
malignant
by four radiologists.
halo [i
value
isoechoic and 50 proved
,
2, 5, 6].
of the
sonographic
or hyperechoic benign
lesions
halo
sign
liver lesions, were
in
the
retrospec-
WERNECKE
1006
Materials
and
ET
AL.
radiologist.
Methods
The
to avoid Selection
of Hepatic
We reviewed whom
Lesions
reports
isoechoic
and
sonograms
or hyperechoic
liver
of the liver
lesions
were
of patients
detected
in
during
a
a malignant
the tumor
lesion
had enlarged.
or if reports
of follow-up
The diagnosis
considered proved if histologic examination surgery or percutaneous biopsy showed
indicated
of a benign of tissue
tumor
obtained
no evidence
included
six hepatocellular
eral cholangiocarcinoma, tric cancers, four
one
1 7 metastases
pancreatic
teratocarcinomas, two
cancer,
one
renal cell carcinomas, comas,
carcinomas,
testicular
cancer,
carcinoids,
cancer,
one squamous
melanomas,
of colonic
three
one
carcinoma
periph-
two gas-
one gastrinoma, breast
cell carcinoma,
one anaplastic
one
cancer,
two
five leiomyosarof unknown
Each radiologist tissue (irrespective
of a malignant
origin,
icity
The 50 benign
hepatic
lesions
included
1 .0 to 1 5 cm (mean,
29 hemangiomas,
nine
focal fatty infiltrations, three focal nodular hyperplasias, two adenomatous hyperplasias, one regenerating nodule, four abscesses, and two hematomas. The diagnoses were confirmed on the basis of biopsy reports in 24, characteristic findings on dynamic bolus CT and unaltered size and structure on follow-up examination in 22, charac-
teristic findings on sulfur colloid scintigraphy in one, and findings on follow-up examinations alone in three. The diameter of all benign lesions
ranged
from
0.5
to 15 cm (mean,
5.0
cm).
Analysis
Sonograms of all the lesions were presented in a random sequence to each of four radiologists who had no knowledge of the patients’ clinical
history
and
final
diagnoses.
In 31
metastases, an image of a single metastasis ative feature of all lesions was selected
patients
who
had
multiple
displaying a representfor presentation to the
indicated
favor
with
halo
preferred
in order
the diagnosis
multiple
if a hypoechoic was
defined
of the rim’s thickness)
center
of the
of the
of
metastases
had
lesion
and
In for
halo was present
as any peripheral
or
rim of
that had a lower echogen-
surrounding
liver
parenchyma.
Results
The radiologists’
decisions
concerning
the presence
or absence of
halo were compared. All cases with conflicting decisions were reviewed by all radiologists together in a second
session;
whole
the
series
of sonograms
was presented to facilitate decisions were subsequently presence
or absence was
scored
analyzed
predictive
or
as true-positive tumor.
The
values)
in distinguishing
benign,
when halo
sign
a hypoechoic
statistically
of each
(sensitivity,
to determine malignant
lesion
in question
a final decision. The observers’ compared with the final diagnoses.
of a halo was interpreted
malignant
lesion without
from
as indicating
respectively.
a lesion was
The
halo
with a hypoechoic scored
halo was specificity,
and
positive
hepatic
was
halo was a when
The results
the value of the sonographic benign
that the sign
as true-negative
benign.
final The
and
a
were
negative
halo sign
lesions.
Results In 95 of 1 00 hepatic lesions, the four radiologists were almost (three vs one, n = 1 6) or completely (four vs zero, n 79) in agreement about the presence (n 47) or absence (n = 48) of a hypoechoic halo (Figs. 1 -1 1). In those five cases in which they disagreed initially (two vs two), the final decision (four vs zero) was that four lesions had a hypoechoic halo and one had no halo. A hypoechoic halo was detected in 44 malignant tumors (88%) (Figs. 1-4) and in only seven benign lesions (1 4%) (Figs. 5 and 6). For distinguishing benign from =
Image
was
a hypoechoic (two vs two)
malignant
from
the
Evaluation
examinations 5.5 cm).
ranged
patients
lesion
would
1992
Because edge shadowing can simulate a hypoechoic rim at the lateral boundaries of a lesion, the radiologists mainly analyzed those regions where the edge of the lesion was more or less parallel to the sonic wave fronts (i.e. , anterior and posterior borders).
lesion
diameter
of 31
A sonographic
than
and two Hodgkin’s lymphomas. These diagnoses were confirmed on the basis of biopsy results in 12 and on the basis of follow-up in 38. Tumor
lesions
November
both hyperechoic or isoechoic and hypoechoic hepatic lesions. these cases, only hyperechoic or isoechoic lesions were selected presentation.
was during
of a single
multiple
Six
absent.
structure of the lesion over a period of at least 1 2 months. The study material consisted of sonograms of 50 proved malignant tumors (selected during the first 5 months) and 50 proved benign hepatic lesions (selected during all 1 3 months of the study). The mean age of the patients was 43 years (range 20-81 years). The tumors
disease.
that
lesion or if other imaging techniques (dynamic bolus CT, scintigraphy) showed specific morphologic findings or no change in the size and
malignant
presentation
bias, because
malignant
1 3-month period (January 1 989 to February 1990). The diagnosis of a malignant hepatic tumor was considered confirmed if biopsy results indicated
AJR:159,
=
Fig. 1.-sonogram of a 62-year-old man with peripheral cholangiocarcinoma shows a 4.0-cm tumor with a hypoechoic halo (arrowheads).
Fig. 2.-5onogram of a 56-year-old man with carcinoid shows a homogeneous hyperechoic 6cm metastasis surrounded by a hypoechoic halo (arrowheads). A = aorta.
AJR:159,
November
SONOGRAPHY
1992
OF
LIVER
TUMORS
1007
Fig. 3.-sonogram of a 67-year-old woman with adenocarcinoma of the colon shows a 4.5-cm isoechoic metastasis that can be delineated from surrounding liver tissue only by its faint hypeechoic halo (arrowheads).
Fig. 4.-sonogram of a 30-year-old woman with melanoma shows two isoechoic metastases 3.0 and 3.5 cm in diameter with extremely broad hypoechoic rims (arrowheads) having poorly defined outer and inner borders. = inferior vena cava.
Fig. 5.-Sonogram
of a 46-year-old
woman with
..,..;
hemangioma shows a 4.5-cm hyperechoic lesion with bulging of liver contour and unequivocal hypoechoic halo (arrowheads).
Fig. 6.-sonogram
of a 47-year-old
I.
I.
I’.
._.;
.
.--
.
.:“-“‘-t
: “-#{149}-“
..--
-zs-.---.-
.1
woman with
focal nodular hyperplasia shows a 7-cm Isoechoic tumor with a faint hypoechoic halo (arrowheads). C = inferior vena cava.
lesions, the halo sign had a sensitivity of 88%, a specificity of 86%, and positive and negative predictive values of 86% and 88%, respectively. In malignant tumors, five (71%) of seven primary liver neoplasms (Fig. 1) and 39 (91 %) of 43 metastases (Figs. 2-4) showed a hypoechoic halo. The ranges of sizes of malignant lesions with and without halos were 1 .5-15.0 cm (mean, 5.5 cm) and 3.0-i 2.0 cm (mean, 7.0 cm), respectively. In benign tumors, only two (7%) of 29 hemangiomas (Fig. 5) and one (25%) of four abscesses had a hypoechoic halo. All nine focal fatty infiltrations (Fig. 1 0), all two abscesses, and one regenerating nodule had no halo. It is important to note that all three focal nodular hyperplasias (Fig. 6) and one of two adenomas had a hypoechoic halo. Thirteen (45%) of 29 hemangiomas had an atypical appearance on sonograms (focal internal hypoechoic zones or malignant
a peripheral hypoechoic rim) (Figs. 8 and 9). Twenty-three of 29 hemangiomas were larger than 3 cm in diameter (range, 1 -1 5 cm; mean, 4.8 cm). The ranges of sizes of hemangiomas with
(mean,
and without halos were 4.9 cm), respectively.
3.0-3.5
cm and
1 .0-i
5.0 cm
distinguishing isoechoic or hyperechoic metastases hemangiomas, the halo sign had a sensitivity of 88%, a specificity of 93%, and positive and negative predictive values of 95% and 87%, respectively. For
from
Discussion Liver tumors can be classified into four groups according to their echogenicity relative to surrounding liver tissue: anechoic, hypoechoic, isoechoic, or hyperechoic. The differential diagnosis of hypoechoic liver lesions includes widespread malignant tumors (such as metastases and hepatocellular carcinomas) and rare benign tumors (focal nodular hyperplasia, hepatic adenoma, adenomatous hyperplasia, and, less often, hemangiomas in a diffusely fatty liver or inflammatory granulomas). For statistical reasons, most hypoechoic liver lesions are probably malignant (with the exception of focal sparing in fatty infiltration of the liver, which normally has rather characteristic sonographic features) [2]. Hyperechoic and isoechoic liver lesions are more difficult to classify be-
I
