Hepatic leiomyosarcomas: CT features with pathologic correlation

Hepatic leiomyosarcomas: CT features with pathologic correlation

EUROPEAN JOURNAL OF RADIOLOGY ELSEVIER European Journal of Radiology 19 (1995) 177-182 Hepatic leiomyosarcomas: CT features with pathologic correla...

511KB Sizes 0 Downloads 64 Views

EUROPEAN JOURNAL OF RADIOLOGY

ELSEVIER

European Journal of Radiology 19 (1995) 177-182

Hepatic leiomyosarcomas: CT features with pathologic correlation P h i l i p p e S o y e r *a, D a v i d A . B l u e m k e a, M a u r e e n R i o p e l b, R a l p h H . H r u b a n b, Elliot K. Fishman a aDepartment of Radiology. The Johns Hopkins Hospital, 600 N Wolfe Street, Baltimore, MD 21205-2180. USA bDepartment of Pathology, The Johns Hopkins Hospital, 600 N Wolfe Street, Baltimore, MD 21205-2180, USA

Received 4 October 1994; revision received 15 November 1994;accepted 18 November 1994

Abstract

The purpose of this study was to determine the computed tomographic (CT) features of hepatic leiomyosarcoma and to correlate them with pathologic findings. CT examinations of 15 patients with pathologically proven leiomyosarcoma of the liver were retrospectively reviewed. Three patients had primary leiomyosarcoma of the liver and 12 patients had 43 distinct hepatic metastases originating from a primary gastrointestinal (n = 8), uterine (n = 2) or retroperitoneal (n = 2) leiomyosarcoma. In the 11 patients who had partial hepatic resection or surgical biopsy of their tumors, a correlation was made between CT imaging and pathologic findings. Primary leiomyosarcomas showed heterogeneous enhancement; two displayed internal and peripheral enhancement, and one showed peripheral enhancement with a pseudocystic pattern (i.e. enhancing thick wall with internal nonenhancing low attenuation area). Leiomyosarcomas metastatic to the liver were homogeneous, showing no or moderate enhancement (10 of 43 metastases, 23%) or heterogeneous and predominantly peripheral enhancement (33 of 43 metastases, 77%). A pseudocystic pattern was seen in 13 of the 43 metastases (30%). Homogeneous tumors were made of smooth fibrous-like tissue without area of necrosis. Heterogeneous tumors contained varying degrees of necrosis and hemorrhage or gelatinous tissue. We found that primary and secondary hepatic leiomyosarcomas of the liver can exhibit different features. A pseudocystic pattern is uncommon. Furthermore, purely cystic tumors were not seen in our series. Keywords: Computed tomography, neoplasms; Liver, neoplasms; Neoplasms, liver; Leiomyosarcoma, liver

1. Introduction

Leiomyosarcoma of the liver is a rare tumor. Most of the time, these tumors are metastases from primary tumors arising in other organs such as the uterus and small bowel [1]. The CT features of leiomyosarcomas metastatic to the liver have previously been described [1-3], and a main conclusion of these earlier reports was that these neoplasms are often cystic. Primary leiomyosarcoma of the liver is exceedingly rare, and only a few eases have been sporadically reported in literature [4-8]. The CT features of these tumors did not receive special attention [4,6] and, to date, little is known about the CT presentation of primary leiomyosarcoma of the liver. * Corresponding author, Department of Radiology, HGpital Foch, 40 rue Worth BP 36, 92151 Suresnes cedex, France.

The aim of this study was to reassess the CT features of hepatic leiomyosarcoma, particularly in light of recent improvements in CT technology since the previous reports were published, and to correlate these features with pathologic results. 2. Materials and methods 2.1. Patients

A computer search of the records of the Division of Surgical Pathology over a 56-month period (July 1989 to February 1994) disclosed 24 patients with a pathologically proven leiomyosarcoma in the liver. Of these, 11 were patients referred from outside institutions for pathologic examination only and had no CT images available for review. The remaining 13 patients had CT

0720-048X/95/$09.50 © 1995 ElsevierScienceIreland Ltd. All rights reserved SSDI 0720-048X(94)00592-Z

178

P. Soyer et al./European Journal of Radiology 19 (1995) 177-182

examination at our hospital. In addition, two patients examined with CT during this period of time by one of the authors in another institution were included in the study. The study group comprised five males and 10 females, aged 30-84 years (mean age, 57 years). Three patients had a primary leiomyosarcoma of the liver. Twelve patients had metastases to the liver from gastric (six patients), uterine (two patients), duodenal (one patient), jejunal (one patient), and retroperitoneal (two patients) primary leiomyosarcoma. Specimens for histologic confirmation of the diagnosis were obtained by surgical resection in eight patients, exploratory laparotomy in two patients, and percutaneous biopsy in five patients. On the basis of the origin of the tumor, (i.e. primary or metastatic) the patients were divided into two groups: Group 1 included three patients with primary leiomyosarcoma of the liver, and Group 2 included 12 patients with leiomyosarcomas metastatic to the liver. 2.2. CT imaging

Fifteen CT examinations were performed on various CT units (Somatom Plus, Siemens, Iselin, NJ; CT Pace Plus, General Electric-Medical System, Milwaukee, WI; Somatom DR, Siemens; Elite Plus, Elscint, Haifa, Israel). Nine CT scans were performed during intravenous injection of iodinated contrast material, using a dynamic technique with 5-10 mm collimation at 5-10 mm intervals, respectively (six examinations), or a spiral technique with 8-mm collimation and 4 mm overlapping reconstruction (three examinations). The contrast material was injected as a bolus into an antecubital vein, through a 20-G canula, either manually or using an automatic injector. Ionic contrast medium (120 ml) were injected in all cases (38 g 1/100 ml in seven cases, and 30 g 1/100 ml in two cases). Six CT examinations were obtained during arterial portography (CTAP) using either conventional incremental (one examination) or spiral CT technology (five examinations), using previously described protocols [9,10].

by the presence of a tumor with prominent internal nonenhancing hypoattenuating area with enhancing thick wall. In all cases in which the tumors were resected or sufficient histologic material was obtained by biopsy, the CT findings were correlated with the pathologic findings. Correlation between CT findings and pathologic features was made by one radiologist and a pathologist who reviewed the pathologic reports, the histologic slides, and the gross photographs. Correlation between CT features and pathologic findings was performed in the 17 tumors (three primary and 14 secondary) from which histologic material was obtained. 3. Results 3. I. Group I: primary leiomyosarcoma

The largest diameters of the tumors were 12, 15 and 19 cm. All tumors were solitary. The tumor was located in the right hemiliver in one patient, and in both hemilivers in two patients. Two tumors had smooth margins and one had irregular margins. The three tumors appeared as low attenuation masses on unenhanced CT scans; two were homogeneous, and one was heterogeneous with a central markedly hypoattenuating area corresponding to tumor necrosis on pathologic examination. One tumor displayed capsular retraction and had internal small punctate calcifications (Fig. 1). On contrast-enhanced CT scans, tumor enhancement was predominantly peripheral; two tumors displayed heterogeneous internal and peripheral enhancement and one tumor showed only peripheral

2.3. Image analysis

CT images were reviewed by two radiologists reading side-by-side and consensus opinions were obtained. On CT images, hepatic tumors were evaluated for number, size, and margins (smooth or regular). Retraction of the liver capsule adjacent to the tumor, bile duct dilatation, venous involvement (both portal and hepatic venous), and extrahepatic spread were searched for. On CT images tumor attenuation was compared to that of the adjacent liver, and characteristics of enhancement, homogeneity, and presence of a hyperintense peripheral capsule or hypointense rim were analyzed. Baseline nonenhanced CT scan was used to determine increase in tumor attenuation. A pseudocystic pattern was defined

Fig. 1. Primary leiomyosarcomaof the liver. UnenhancedCT scan (liverwindow)showsa largehepatictumor,centrallylocated,containing small calcifications(arrowheads). Retractionof the liver capsule (curved arrow) adjacentto the tumor is seen.

P. Soyer et al./ European Journal of Radiology 19 (1995.) 177-182 ~!~

~,

z¸•¸

179

subcutaneous metastases) and ascitis were seen in one patient and confirmed at surgery. 3.2. Group 2." leiomyosarcoma metastatic to the liver

iil~

Fig. 2. Secondary leiomyosarcoma of the liver. Unenhanced CT scan shows a single homogeneous metastasis (arrow) displaying low attenuation compared to that of the liver.

enhancement with a pseudocystic pattern. On pathologic examination, the enhancing peripheral areas corresponded to solid portions of the tumor. In the three tumors, the internal aspect of the tumor consisted of fibrosis, necrosis and hemorrhage in varying degrees. No capsule was depicted on CT examinations while a true peripheral capsule was depicted in two tumors on pathologic examination. Portal encasement by tumor was seen in one patient. Extrahepatic spread (i.e. pancreatic and omental involvement, enlarged celiac lymph nodes, and

a

The number of visible metastases per patient ranged from one to more than 15. Eleven patients had a total of 42 visible metastases, with the largest diameter of the metastases ranging from 0.5 to 16 cm (mean, 4 cm). In the remaining patient, metastases were too numerous (> 15) to count accurately; in this patient, the metastases displayed similar CT features and were evaluated as a single mass. In six patients, the metastases were solitary. No hemiliver was more frequently affected than the other. Three metastases in the same patient had irregular margins; pathologic correlation was available in two of these metastases and showed infiltrative margins. The other 40 metastases had smooth margins; pathologic correlation available in 12 of these metastases showed well-defined metastases, with a true fibrous capsule in two of them, and marked reactive changes in the adjacent hepatic parenchyma in the other ten. All metastases appeared as homogeneous areas of low attenuation on unenhanced CT scans (Fig. 2). No peripheral metastasis was associated with retraction of the adjacent hepatic capsule or internal calcifications. On contrast-enhanced CT scans, 10 (23%) metastases were homogeneous, showing no or minimal enhancement; pathologic examination was available for three of these metastases and showed that they were made of smooth fibrous-like tissue without macroscopically visible areas of necrosis (Fig. 3). In the remaining 33, contrast enhancement was predominantly peripheral. Twenty of these 33 metast-

b

Fig. 3. Secondary leiomyosarcoma of the liver. (a) CT during arterial portographic scan shows a single homogeneous metastasis (arrows) displaying mild enhancement compared to that of the liver. Enhancement of the tumor was established by comparison with the pre-CTAP scan (not shown). (b) Pathologic examination of the resected specimen shows typical features consisting of cells with oval-shaped nuclei and thin delicate cytoplasm which group to form fascicles interlocking at right angles. No necrosis nor hemorrhage was found in this tumor (HE stain, original magnification x 100).

180

P. Soyer et aL /European Journal of Radiology 19 (1995) 177-182

a

b

Fig. 4. Secondaryleiomyosarcomaof the liver. (a) Contrast-enhancedCT scan shows a single heterogeneoushypodensemetastasis (arrow) displaying mild enhancementcompared to that of the liver. The enhancement is predominantly peripheral, with a hypodensecentral area. Large heterogeneous primary gastric leiomyosarcoma is seen. (b) Pathologic examination of gross specimen after surgical wedge resection of the liver shows well-circumscribedtumor (arrow) containing central gelatinous tissue.

ases (47%) displayed heterogeneous internal and peripheral enhancement; correlation between C T a n d pathologic features performed in six metastases showed

a

that they contained varying degrees of necrosis a n d hemorrhage or contained gelatinous tissue (Figs. 4, 5). Thirteen of these 33 metastases (30%) showed only

b

Fig. 5. Secondary leiomyosarcomaof the liver with a pseudocysticpattern. (a) Contrast-enhanced CT scan shows heterogeneousmetastases. One displays pseudocystic pattern (arrow) with enhancing peripheral wall and homogeneous hypodense center. A larger one (curved arrow) shows enhancing peripheral wall and heterogeneouslyenhancing center. (b) Correlation between CT features and pathologic examination available for the larger metastasis shows focal areas of typical features of leiomyosarcoma(L), with areas of necrosis (N) and myxoid degeneration (M) (HE stain, original magnification x 250).

P. Soyer et al./ European Journal of Radiology 19 (1995) 177-182

I a

181

b

Fig. 6. Secondary leiomyosarcoma of the liver. (a) Spiral c r dring arterial portography scan shows single large heterogeneous metastasis displaying central scar (arrow). (b) Spiral CT during arterial portography scan shows single large homogeneous metastasis. Tiny vessels at the periphery of the tumor are seen (arrowheads). Opacification of the vessels during portal opacification suggests a portal origin.

peripheral enhancement with a pseudocystic pattern; pathologic review of five of these metastases showed that they contained varying degrees of necrosis and hemorrhage or contained gelatinous tissue on pathologic examination (Fig. 5). In eight metastases (19%), an incomplete hypodense peripheral rim was depicted; pathologic review of specimens available for two of these metastases showed that they were surrounded by a small rim of normal compressed hepatic parenchyma, but that they lacked a true fibrous capsule. A central scar was seen in one large metastasis (2%) which arose from a retroperitoneal leiomyosarcoma (Fig. 6a). In two metastases (5%), tiny vessels were seen at the periphery of the metastases (Fig. 6b). Portal encasement by a metastasis was seen in one patient. Extrahepatic extension of the primary tumor was seen in three patients (vertebral metastasis in one patient, enlarged lymph nodes in the porta hepatis in two patients). Ascitis was seen in one patient. Biliary duct dilatation was not observed. 4. Discussion

Because primary leiomyosarcoma of the liver is rare, its diagnosis requires a careful search to ascertain that the tumor did not originate from an adjacent structure or is not a single metastasis from another primary organ. Grossly, primary leiomyosarcoma of the liver usually presents as a single large mass, firm in consistency [IlL Histopathologically, the tumor displays intersecting bundles of elongated or round cells which

demonstrate desmin immunoreactivity. Aggressive surgical resection, when possible, is the preferred treatment [4], although orthotopic liver transplantation has been proposed [12]. During the past two decades, CT features of a variety of both benign and malignant neoplasms of the liver have been extensively described. However, the CT features of primary leiomyosarcoma of the liver, probably due to its rarity, have been infrequently reported [4, 6]. Furthermore, to our knowledge, there is no description of the CT findings of this tumor in the radiologic literature. On CT, this tumor has been reported as a large, well-delineated mass, frequently with a predominantly peripheral enhancement [4,8], or less commonly displaying a mainly cystic appearance [6]. In our study, two primary tumors displayed heterogeneous internal and peripheral enhancement and one tumor showed only peripheral enhancement with a pseudocystic pattern (i.e. enhancing thick wall with internal nonenhancing low attenuation area). Whereas a true fibrous capsule rather than compressed hepatic parenchyma was present on pathologic examination in two tumors, it was not seen on CT examinations. One possible reason for the 'invisibility' of the capsule is that it was very thin (< 3 mm), whereas the capsule in other primary tumors (such as hepatocellular carcinoma) are often thicker. In our three primary leiomyosarcomas, the CT features of the tumors were not specific. Ancillary findings such as portal encasement, ascitis, bile duct dilatation, and enlarged lymph nodes can be found in a wide range of primary or secondary hepatic tumors. We found that CT features of primary leiomyosarcomas of the liver

182

P. Soyer et aL /European Journal of Radiology 19 (1995) 177-182

were not uniform and depended on the pathologic presentation o f the tumor. Leiomyosarcoma metastatic to the liver is also uncommon. The most frequent primary sites are the alimentary tract (stomach, small bowel, colon), the genitourinary tract (uterus, bladder, prostate, kidney), and the retroperitoneum (inferior vena cava) Ill. Pathologically, these metastases have the tendency to develop internal hemorrhage, necrosis and cystic changes [1]. Although leiomyosarcomas metastatic to the liver are more frequent than the primary tumors, they have received little attention in radiologic literature. In a study describing the C T features in a large series of patients with leiomyosarcomas of various sites, 44 patients were found to have hepatic metastases; 29 of these had apparent necrosis within their metastases with a cystic pattern [1]. This earlier paper did not focus on the description of the hepatic metastases, so that comparison between the CT finding observed in this series and ours is not feasible in retrospect. Furthermore, the enhancement patterns of the tumors were not described. Metastases to the liver from leiomyosarcoma have been described as frequently cystic [1,2,13,14]. These tumors were described as centrally necrotic in three cases in one report [2] and predominantly necrotic with a cystic pattern in 29 of 44 cases in another report Ill. In a report by Federle et al. [13], one patient with multiple hepatic metastases from leiomyosarcoma of small bowel was studied with sonography and CT. The metastases were described as purely cystic because they did not show internal septation. However, the tumors displayed thick walls. In our study, no metastases displayed a cystic pattern (i.e. homogeneous hypodense t u m o r with thin nonenhancing wall). We observed, rather, a 'pseudocystic' pattern in 30% of tumors; some tumors had predominantly central necrosis, but others contained an amorphous gelatinous tissue without necrosis. However, in all cases, an enhancing wall was seen. A hypointense rim seen at the periphery of some of the hepatic metastases from leiomyosarcoma was found to be made o f normal compressed hepatic parenchyma. A similar finding has been reported by M u r a m a t s u et al. [15] who described a peripheral area of low attenuation in hel~atic metastases of colorectal cancer. However, in this study, the peripheral area corresponded to viable tumor rather than compressed hepatic parenchyma. Calcifications in leiomyosarcoma metastatic to the liver are exceedingly rare and were not present in any of the secondary tumors in our series. They were also absent in the 44 patients studied by McLeod et al. Ill. However, calcification in a liver metastasis from colonic leiomyosarcoma has been reported [16]. In conclusion, no characteristic feature allowing differentiation between primary and malignant leiomyosa-

rcoma of the liver emerged in our study. We found that primary and secondary hepatic leiomyosarcomas of the liver can exhibit different features. A pseudocystic pattern may occur but is uncommon. Furthermore, purely cystic features o f these tumors were not seen in our series.

Acknowledgements This work has been supported in part by the Laboratoire Guerbet, Aulnay-sous-Bois, France, and by a grant from the French Radiological Society.

References [1] McLeod AJ, Zornoza J, Shirkhoda A. Leiomyosarcoma: computed tomographic findings. Radiology 1984; 152: 133-136. [2] Noon MA, Young SW, Casteilino RA. Leiomyosarcoma metastatic to the liver: CT appearance. J Comput Assist Tomogr 1980; 4: 527-530. [3] Wooten WB, Bernardino ME, Goldstein HM. Computed tomography of necrotic hepatic metastases. AJR 1978; 131: 839-842. [4] lshii H, Nakayama T, Hiyama Y. Primary hepatic ieiomyosarcoma: the investigation of domestic and foreign 34 cases [Japanese]. Nippon Shokakibyo Gakkai Zasshi 1991; 88: 1256-1263. [5] Huguet C, Harb J, Gavelli A, Riberi A. Leiomyosarcoma of the inferior vena cava with liver metastasis: complete resection with venous reconstruction [French]. Gastroenterol Clin Biol 1992; 16: 714-717. [6] Tashiro K, Matumoto T, Hirata K et al. A case report of primary leiomyosarcoma of the liver with specific growth type [Japanese]. Kanzou 1990; 31: 804-810. [7] Baur M, Potzi R, Lochs H, Neuhold N, Walgram M, Gangl A. Primary leiomyosarcoma of the liver: a case report. Z Gastroenterol 1993; 31: 20-23. [8] Paraskevopoulos JA, Stephenson TJ, Dennison AR. Primary leiomyosarcoma of the liver. HPB Surg 1991; 4: 157-162. [9] Soyer P, Lacheheb D, Levesque M. CT arterial portography of the abdomen: effect of injecting papaverine into the mesenteric artery on hepatic contrast enhancement. AJR 1993; 160: 1213-1215. [10] Bluemke DA, Fishman EK. Spiral CT arterial portography of the liver. Radiology 1993; 186: 576-579. [!1] Ishak KG. Malignant mesenchymal tumors of the liver. In: Okuda K, Ishak KG, editors. Neoplasms of the liver. New York: Springer-Verlag, 1987; 159-176. [12] Saint-Paul MC, Gugenheim J, Hofman Pet al. Leiomyosarcoma of the liver: a case treated by transplantation [French]. Gastroenterol Clin Biol 1993; 17: 218-222. [13] Federle MP, Filly RA, Moss AA. Cystic hepatic neoplasms: complementary roles of CT and sonography. AJR 1981; 136: 345-348. [14] Arriv6 L, Kraiem C, Roche A. Morphologic evaluation of pseudocystic hepatic tumors: report of 10 cases [French]. Gastroenterol Clin Biol 1988; 12: 516-520. [15] Muramatsu Y, Takayasu K, Moriyama Net ai. Peripheral lowdensity area of hepatic tumors: CT-pathologic correlation. Radiology 1986; 160: 49-52. [16] Clark RA, Alexander ES. Computed tomography of gastrointestinal leiomyosarcoma. Gastrointest Radiol 1982; 7: 127-129.