1991, The British Journal of Radiology, 64, 1019-1022

Computed tomography attenuation measurements for the characterization of hepatic haemangiomas By R. W. Whitehouse, BSc, M B , ChB, FRCR Department of Diagnostic Radiology, Stopford Building, Manchester University, Oxford Road, Manchester M13 9PT, UK {Received January 1991 and in revised form May 1991) Keywords: X-ray computed tomography, Haemangioma, Liver neoplasms Abstract. The attenuation values of 21 hepatic haemangiomas in 19 patients were measured on non-enhanced computed tomographic (CT) scans and compared with the attenuation of adjacent liver and the inferior vena cava (IVC). The attenuation of hepatic haemangiomas was lower than that of the surrounding liver, but there was no correlation between these two measurements. There was a highly significant correlation between the attenuation of haemangiomas and blood in the vena cava (r = 0.905, p < 0.001). All the haemangiomas had attenuations within 7 HU of caval blood. By comparison, in 34 hypodense hepatic lesions that did not show contrast-enhanced appearances characteristic of haemangioma, there was no significant correlation between the attenuation of the lesions and the IVC. Nineteen (56%) of these lesions had attenuations differing more than 7 HU from that of caval blood. The influence of this observation on the requirement for dynamic contrast-medium-enhanced and delayed post-contrast CT in the assessment of hepatic lesions is discussed.

Several characteristic computed tomographic (CT) features of hepatic cavernous haemangiomas have been recognized for over a decade (Freeny et al, 1979; Barnett et al, 1980; Itai et al, 1980). Typically, they are hypodense compared with surrounding liver on the unenhanced CT scan. Significant peripheral enhancement occurs during dynamic contrast-medium-enhanced scanning. Later they "fill in" centripetally with contrast medium to become isodense with liver on delayed postcontrast scanning, though this may take up to an hour. Clefts of non-enhancement within the lesion may be present, particularly in the larger lesions described as giant cavernous haemangiomas (Scatarige et al, 1987; Choi et al, 1989). The time required for isodense "fillin" to occur should not be less than 3 min from the injection of contrast medium. Whilst this description has a 100% positive predictive value for haemangiomas (Ashida et al, 1987), Freeny and Marks (1986a) found only 55% of haemangiomas fulfilled these criteria and also calculated that in patients with known malignant disease, only 86% of lesions with this appearance would be haemangiomas (Freeny & Marks, 1986b). Two previous publications describe hepatic haemangiomas as having similar pre-contrast attenuation to that of blood (Itai et al, 1980, 1983), but these studies did not substantiate this with attenuation measurements. The present study was undertaken to evaluate this observation quantitatively and to assess its application in practice. Patients and methods

In a retrospective study of the 4-year period from March 1986 to May 1990, 19 patients with 21 hepatic haemangiomas had been examined by CT in the Department of Diagnostic Radiology. The diagnosis Vol. 64, No. 767

was based on the CT demonstration of a lesion of lower attenuation than surrounding liver, which showed peripheral enhancement on dynamic contrast-mediumenhanced CT and became isodense with surrounding liver on delayed post-contrast scanning. These features were present in all patients. Fifteen patients also had ultrasound examinations, which revealed well defined hyperechoic lesions. Biopsy confirmation was available in five patients. Clinical follow-up of between 6 months and 4 years confirmed the benign nature of all the hepatic haemangiomas. There were six men and 13 women, with ages between 31 and 77 years (mean 52 years). In a further 13 patients who were scanned for assessment of histologically proven malignant disease, 34 other hepatic lesions were identified. These patients had pre- and post-contrast studies and were diagnosed as having multiple hepatic metastases on the basis of clinical findings, follow-up, liver ultrasound scan and abdominal CT. All patients were scanned on a GE 9800 generalpurpose CT scanner. The pre-contrast-medium scans from these patients were interrogated on an independent console; a section through the middle of each hepatic lesion was chosen and an oval region of interest (ROI) was placed to include the largest possible area within each lesion, allowing at least a 5 mm margin to prevent partial volume averaging from surrounding liver (Fig. 1). The area and attenuation of this ROI was recorded. A similar region within the inferior vena cava (IVC) was chosen, with a smaller margin (approximately 1-2 mm) as the orientation of the IVC makes partial volume averaging with surrounding tissue unlikely. The attenuation of a representative region of adjacent hepatic parenchyma was also measured. 1019

R. W. Whitehouse Results

Figure 1. Typical pre-contrast CT scan of hepatic haemangioma demonstrating ROI choice for haemangioma (1) and IVC (2).

The area of the haemangiomas varied from 0.61 cm2 to 15 cm2 (mean 3.74 cm2). There was no significant difference in the results of attenuation values or correlation with blood attenuation for small lesions (less than 1 cm2) compared with large lesions. The results have therefore been combined. Haemangiomas had a significantly lower mean pre-contrast attenuation (38.3 ±7.1 HU) than surrounding liver (57.5 ± 5 HU) (p < 0.0001), but showed no correlation with liver attenuation. Haemangiomas had similar mean attenuation to blood in the IVC (37.7 ± 6.2 HU) and a significant correlation was evident between these measurements (r = 0.905, p < 0.001) so that all haemangiomas lay within 7 HU of blood attenuation (Fig. 2). All the hepatic lesions in patients with malignant disease were hypodense when compared with surrounding liver (mean = 30 ±8.8 HU), but showed no significant correlation with liver or IVC attenuation (r = 0.26). Fifteen of the 34 lesions (44%) had attenuations within 7 HU of blood. The distribution of attenuation values relative to the IVC in these patients is demonstrated in Fig. 3. None of these lesions showed

tissue attenuation •f

60 >

• •

40

-



v/r-0.9

• •

y

m

m

haemangioma liver

on

20

30

40

IVC attenuation 1020

50

Figure 2. Pre-contrast CT attenuation values for liver and hepatic haemangiomas plotted against attenuation values for the inferior vena cava content on the same section. The British Journal of Radiology, November 1991

CT attenuation of hepatic haemangioma

tissue attenuation

60

4

i"

40

a

u

i

a

+ •••

a



• •

t



a

a a

20

a

a a

+ a

Figure 3. Pre-contrast CT attenuation values for liver and hypodense hepatic lesions plotted against attenuation values for the inferior vena cava contents on the same section.

n

25

the dynamic contrast-medium-enhanced CT appearances typical of haemangiomas. Discussion

Histologically, hepatic cavernous haemangiomas are composed of large vascular spaces with thin areas of stroma and occasional hepatocytes. Cystic change, haematoma and fibrosis may occur in larger haemangiomas, with consequent change in CT appearances (Takayasu et al, 1986). Most smaller haemangiomas, however, have consistent CT appearances as described in the introduction. It should not therefore be surprising that the CT attenuation of a haemangioma is similar to that of blood, as this is the largest component by volume. Despite the potential difficulties in obtaining accurate attenuation measurements from CT scans owing to beam hardening, motion artefacts and other factors, this study demonstrates that practically useful results can be obtained by using a relevant intrinsic tissue as a reference. Where CT attenuation numbers are obviously inaccurate due, for example, to streak artefacts or beam hardening, then no measurement of attenuation, either Vol. 64, No. 767

liver metastasis

1

i

35

45

IVC attenuation absolute or relative to other parts of the same scan, will be of great value. In the present study this did not pose a problem in the assessment of hepatic lesions, probably because of the large size and uniform nature of the liver. Peripheral lesions or scans of reduced quality should, however, be interpreted with care. Whilst the use of the unenhanced, post-contrast and dynamic contrast-medium-enhanced appearances of hepatic lesions as criteria for diagnosis of a haemangioma in this study places limitations on the interpretation of the results, as none of these patients have developed evidence of malignant disease, then these appearances are approximately 100% specific for haemangioma (Ashida et al, 1987; Freeny & Marks, 1986a,b). Haemangiomas with attenuation values similar to, or higher than, surrounding liver are unusual but may occur in livers that have fatty infiltration or in patients with high haematocrits. Although no such lesions were included in the present study, three haemangiomas were within 10 HU of liver attenuation. Previous publications have described haemangiomas in fatty infiltrated livers (Ashida et al, 1987; Freeny & Marks, 1986a,b) noting them to be of higher attenuation than surrounding liver. No satisfactory modifica1021

R. W. Whitehouse

tion to the CT description of hepatic haemangioma was suggested by these authors to allow for this. By extrapolation from the current findings, an altered definition of the pre-contrast appearances of haemangiomas would allow lesions to be included if they lay within 7 HU of the IVC attenuation, whatever the relationship to hepatic attenuation. Several different techniques for CT liver scanning have been described, the choice of technique being largely determined by the clinical circumstances (Ferrucci et al, 1988). Typical appearances on unenhanced scans, dynamic contrast-enhancement and delayed post-contrast-medium scans are required for a CT diagnosis of hepatic haemangioma. The usual clinical circumstance in which this series of scans is performed is in the further evaluation of a patient with a solitary hepatic mass, where the differential diagnosis includes metastatic tumour and haemangioma. Currently, any incidental hepatic lesion of lower attenuation than surrounding liver on unenhanced CT is thus potentially a haemangioma, and requires dynamic intravenous contrast enhancement for further CT evaluation, followed by delayed post-contrast scanning if peripheral enhancement is evident on the dynamic scans. The results of this study suggest that such further evaluation will only be positive (typical of a haemangioma) if the attenuation of the unenhanced lesion lies within 7 HU of the IVC. As blood and most hepatic metastases are of lower attenuation than surrounding liver, a range of 7 HU each side of blood attenuation may still incorporate a large proportion of lesions that will not show characteristic contrast enhancement (44% of those measured in this study, none of which were subsequently identified as haemangiomas). However, by the same argument, such enhancement may be confidently excluded in the 56% of cases with lesions more than 7 HU away from caval blood on the unenhanced scans, obviating the need for contrast studies in these patients. The results of these quantitative measurements made in hepatic lesions suggest that comparison of attenuation in hepatic lesions and the IVC may provide criteria

1022

by which the need for performing additional contrastmedium-enhanced scans be judged. References ASHIDA, C , FISHMAN, E.K.,

ZERHOUNI, E.A., HERLONG, F.H. &

SIEGELMAN, S.S., 1987. Computed tomography of hepatic cavernous haemangioma. Journal of Computer Assisted Tomography, 11, 455-460. BARNETT, P. H., ZERHOUNI, E. A., WHITE, R. I. & SIEGELMAN,

S. S., 1980. Computed tomography in the diagnosis of cavernous hemangioma of the liver. American Journal of Roentgenology, 134, 439-447. CHOI, B. I., HAN, M. C , PARK, J. H., KIM, S. H., HAN, M.

H.

& KIM, C , 1989. Giant cavernous haemangioma of the liver: CT and MR imaging in 10 cases. American Journal of Roentgenology, 151, 1221-1226. FERRUCCI, J. T., FREENY, P. C , STARK, D. D., FOLEY, W. D., MUELLER, P. R., MAY, G. & BURHENNE, H. J., 1988.

Advances in hepatobiliary radiology. Radiology, 168,319-338. FREENY, P. C. & MARKS, W. M., 1986a. Hepatic haemangioma: dynamic bolus CT. American Journal of Roentgenology, 147, 711-719. FREENY, P. C. & MARKS, W. M., 1986b. Patterns of contrast enhancement of benign and malignant hepatic neoplasms during bolus dynamic and delayed CT. Radiology, 160, 613-618. FREENY,

P. C ,

VIMONT,

T. R.

&

BARNETT,

D. C ,

1979.

Cavernous haemangioma of the liver: ultrasonography, arteriography, and computed tomography. Radiology, 132, 143-148. ITAI, Y., FURUI, S., ARAKI, T., YASHIRO, N. & TASAKA, A.,

1980. Computed tomography of cavernous hemangiomas of the liver. Radiology, 137, 149-155. ITAI, Y., OHTOMO, K., ARAKI, T., FURUI, S., IIO, M. & ATOMI,

Y., 1983. Computed tomography and sonography of cavernous haemangioma of the liver. American Journal of Roentgenology, 141, 315-320. SCATARIGE, J. C ,

KENNY, J. M.,

FISHMAN, E. K.,

HERLONG,

F. H. & SIEGELMAN, S. S., 1987. CT of giant cavernous haemangioma. American Journal of Roentgenology, 149, 83-85. TAKAYASU, K., MORIYAMA, N., SHIMA, Y., MURAMATSU, Y., YAMADA, T., MAKUUCHI, M., YAMASAKI, S. & HIROHASHI, S.,

1986. Atypical radiographic findings in hepatic cavernous hemangioma: correlation with histological features. American Journal of Roentgenology, 146, 1149-1153.

The British Journal of Radiology, November 1991

Computed tomography attenuation measurements for the characterization of hepatic haemangiomas.

The attenuation values of 21 hepatic haemangiomas in 19 patients were measured on non-enhanced computed tomographic (CT) scans and compared with the a...
567KB Sizes 0 Downloads 0 Views