Hepatic angiomyolipoma-misdiagnosis as hepatocellular carcinoma: A report of 14 cases

Abstract

Key Words: angiomyolipoma, liver neoplasms/ diagnosis, immunohistochemistry, microscopy, electron, carcinoma, hepatocellular/pathology, case-control studies

INTRODUCTION

Angiomyolipoma (AML) is a rare benign mesenchymal tumor of the liver, composed o f a varying heterogeneous mixture of three tissue components: blood vessels, smooth muscle, and adipose cells. It has recently been proposed that the perivascula r epithelial cell (PEC) is the common progenitor[1,2]. Since its first d escription by Ishak in 1976 [3], there have been more than 100 cases repo rted in the English literature[4-6]. With the advance of radiological techniques, many more tumors are being diagnosed by the means. But radiological findings of AML may only be suggestive of the lesion; its definitive diagnosis r equires histological confirmation[9-19]. Some authors regard renal and hepatic AMLs, pulmonary and soft tissue lymphangiomyomatosis[2], pulmon ary and pancreatic clear cell “sugar” tumor, and cardiac rhabdomyoma as closely related groups of tumors, based on their morphologic overlap and common immuno reactivity for HMB-45[1]. They show different microscopic appearances, however, according to their organ of origin. The goals of this study were to hig hlight more subtle morphology and to gain possible insights into the differential diagnosis that could provide important information about this disease.

MATERIALS AND METHODS

Fourteen cases of AML were identified in the pathology files at the Department of Path ology, Chinese Military General Hospital, four of which were consulted cases. All the cases were independently reviewed by two pathologists, and the most import ant diagnostic criterion was the presence of HMB-45-positive cells. The clinical data and follow-up information were obtained in each case (Table 1). All the tumor tissues had been fixed in neutral buffered formalin and were routinelyem bedded in paraffin. Hematoxylin and eosin-stained sections were examined. Accor ding to the morphological aspect, one block was selected for each of these cases. Immunohistochemical study was performed on representative blocks by using an a vidin-biotin peroxidase complex technique. Selective cases were also examined w ith antibodies to CD68, CD31, CD34, factor VIII-related antigen.

Table 1 The clinical findings in all 14 patients.

NO.	Sex	Age(years)	Tumor size (CM)	Location (LOBE)	Primary diagnosis	Follow-up
1	F	38	4.5	R	Angiosarcoma	Well 44 mo after excision
2	F	33	2.5	R	Hepatic carcinoma	Well 35 mo after excision
3	F	33	6.0	R	Hepatic carcinoma	Well 33 mo after excision
4	F	49	26.0	R,L	Hemangioma	Well 153 mo after excision
5	F	48	3.5	R	Hepatic carcinoma	Well 31 mo after excision
6	F	38	12.5	L	Hepatic carcinoma	Well 12 mo after excision
7	F	35	14.5	L	Hepatic carcinoma	Well 9 mo after excision
8	M	63	17.0	R	Cystic adenoma or cystic adenocarcinoma	Well 9 mo after excision
9	M	37	3.5	R	Hepatic carcinoma	Well 16 mo after excision
10	M	34	10.0	R	Hepatic carcinoma	Well 11 mo after excision
11	M	30	7.5	L	Hepatic carcinoma	Well 16 mo after excision
12	F	43	6	L	Hepatic carcinoma	Well 3 mo after excision
13	M	37	12	L	Hepatic carcinoma	Well 2 mo after excision
14	F	43	6	L	Angiosarcoma	Well 1 mo after excision

RESULTS

Clinical findings

The clinical findings in all 14 patients are summarized in Table 1.There was am arked female predominance (women: men = 9:5) and the average age of patients at diagnosis was 40.07 years (range 30-63 years). Only four cases (case 4,7,10,13) had symptoms of space - occupying lesions while others were incidentally discovered on imaging studies (Figure 1). Two cases had two masses each (the other m ass was diagnosed as a hemangioma by a pathologist) and one of them had calcifi ed nodes (confirmed by a pathologist). None of 14 cases had tuberous sclerosis syndrome, angiomyolipoma of kidney, lymphangiomyoma or clear cell tumor of lung or other parts. For patients with complete follow-up data,all were well with no tumor recurrence and metastasis after excision.

Figure 1 MRI image of case 7.

Gross findings

Thirteen cases were solid and one was cystic (the content of cyst was a brown liquid). All tumors were well circumscribed, but not encapsulated (Figure 2). T he average size of all thirteen cases was 9.4 cm (range 2.5 cm-26 cm). The surfaces of tumor sections were yellowish but not uniform and some parts of tumors were fish-like.

Figure 2 Macroscopic appearance of AML with yellowish fatty areas (case 7).

Histological features

The most common pattern was that of solid sheets of myoid cells intermixed with areas of adipose cells and vessles,most of which were thin-walled (Figure 3) and few of which were thick-walled but had to be carefully observed. In 12 out of-14-cases there were clusters of hematopoietic cells and in 1 of 14 cases there were lots of multinu clear cells (Figure 4). The myoid cells usually predominated and their morphology varied from epithelioid to intermediate (ovoid or short s pindle) spindle. The epithelioid cells in our 12 cases were main cell patter, cytoplasm of which varied from clear to vacuolated and eosinophilic granular, nuclei of which were partly normalchromatic with moderate pleomorphism, with delicate chromatin,and a single distinct eosinophilic nucleolus. Furthermore, there were some large bizarre cells in 10 cases out of 14 cases but without nucleus mitoses (Figure 5). Long spindle cells with elongated eosinophilic cytoplasm could be seen but very rarely lipoblast-like cells were observed in sheets in local areas. Only two cases had multiple local necrotic areas but their total size was smaller than 5 percent of the tumor (Figure 6).

Figure 3 Thin-walled vessles and trabecular tumor cells of AML. IH: CD31 × 100

Figure 4 Multinucleus cells in a large number in AML of case 9. HE × 100

Figure 5 Pleomorphistic and large bizarre cells. HE × 200

Figure 6 Local necrotic area of case 9. HE × 50

Immunohistochemical findings (Summarized in Table 2) The tumor cells were positive for HMB-45, but negative for EMA and AFP in all cases.Local tumor cells were positive for Actin and S-100. HMB-45 staining was intense, granular, and concentrated in the perinuclear pink cytoplasm. Although the epithelioid cell s were most consistently stained, the spindle cells were also weakly positive. A ctin staining in spindle tumor cells was more intense than in other cells. The l ipoblast-like cells were positive for both HMB-45 and S-100 protein. Lots of multinucleus cells presenting in 1 of 12 cases were positive for HMB45, Actin, S-100 and CD68. Furthermore, positive rate for Ki-67 of tumor cells in all 12 cases was no more than 1 percent.

Table 2 Immunohistochemical findings in 12 cases.

Serial number	HMB45	Actin	S-100	EMA	AFP	Ki-67%
1	++	++	local+	-	-	< 1
2	++	++	Local+	-	-	< 1
3	++	++	Local+	-	-	< 1
4	++	+	Local+	-	-	< 1
5	++	+	Local+	-	-	< 1
6	++	+	Local+	-	-	< 1
7	++	+	Local+	-	-	< 1
8	+	+	Local+	-	-	< 1
11	++	+	Local+	-	-	< 1
12	++	++	Local+	-	-	< 1
13	++	++	Local+	-	-	< 1
14	++	++	+	-	-	< 1

Ultrastructural finding Six cases were examined under H-7000 electron microscopy. Neoplastic cells gene rally were polygonal and closely arranged with minimum intercellular material. In the cytoplasm, glycogen and round-to-oval mitochondria were common features, together with a characteristic finding of lots of electron-dense, membrane-bound granules (these granules were 60-100 nm in diameter with some filament-like structures in them which could be seen under high magnification. These were de emed to be either premelanosomes or atypical lysosomal bodies) (Figure 7).

Figure 7 Ultrastructure of neoplastic tissue: glycogen, electron-dense granules. EM × 20000

DISCUSSION

Angiomyolipoma, which occurs frequently in kidney but rarely in other sites[7,8], previously considered as a hamartomatous growth rather than a true neoplasm, is a rare benign mixed mesenchymal clonal neoplasm of the liver. Since the first case described by Ishak[3], AML has been diagnosed with increas ing frequency with advances in MRI, CT, CD-SO and angiography. Despite the claim that the imaging features are highly characteristic[9-19], the preoperative diagnoses are erroneous in more than half of the cases in the present s eries. None of our 14 cases were correctly diagnosed before operating, furthermore, 5 cases were misdiagnosed as hepatic cell carcinoma or sarcoma by pathologist even after operating.

The diagnosis of AML is usually made on recognition of three or four components of the tumor, namely blood vessels, smooth muscle, mature fat, and hematopoietic tissue[4-6,20]. These elements, however, are variable in proportion and distribution. The adipose cells have no characteristic features in themselves to distinguish between various lipomatous tumors. Extramedullary hematopoiesis is not an integral part of AML and is seen in many primary benign and malignant hepatic tumors. Its occurrence is probably more closely related to the hepatic s inusoidal endothelium which plays an important part in hematopoiesis in the feta l liver. This also explains that hematopoietic cells are found only in hepatic but not renal AMLs. Blood vessels are present in all kinds of tumors which easily escape attention. The more distinctive features in AML are their tortuosity, thick walls (but rare in AML of live) and perivascular muscle proliferation. It th us appears that the myoid component is the only specific and diagnostic component in AML and it can exist in epithelioid, spindle, and intermediate forms. The e ssential component in this tumor appears to be PECs[1-6]. It has been speculated that the distinctive epithelioid cells are primitive mesenchymal cells that have an ability to differentiate towards both myoid and fat cells. Spindle myoid cells and lipocytes probably represent the mature derivatives of the epi thelioid cells. Immunohistochemistry is particularly useful for diagnosing, as m any authors have found. HMB-45 has been shown to be a promising marker for renal and hepatic AMLs and can even be applied to minute samples such as fine-needle aspirates[21,22].

Because of the rarity and the pleomorphism of histological features of hepatic A MLs, histological diagnosis may be difficult, especially with needle biopsy (one of our cases was diagnosed by this means). Many features in AML can mislead the unwary pathologist to a diagnosis of HCC: polygonal cells in trabecular arrange ment, peliosis, nuclear pleomorphism, prominent eosinophilic nucleoli, deficient reticulin framework, presence of glycogen, eosinophilic globules, and tumor nec rosis. Four of our cases were misdiagnosed by pathologists because of causes mentioned above. In AML with spindle cells and pleomorphic features, sarcoma is the most common incorrect diagnosis. For this reason, one case in our series was midiagnosed as angiosarcoma. Lipomatous AML has to be differentiated from true lipoma and focal fatty change. Evidently, a careful histologic survey of the entire specimen usually reveals the presence of the classic mixed pattern of the tumor. Immunohistochemical result (positive for HMB45 and Actin but negative for AFP or EMA), ultrastructural stucture and the low proliferation of tumor cells can further confirm the diagnosis[25-35]. The histogenesis of AML is unclear, but immunohistochemical and ultrastructural studies provide important insights into its cell differentiation[4-6,20,23,24]. The concept of “perivascular epithelioid cells” as the unified feat ure has been proposed[1]. Presumably, the PECs are an aberrant type of mesenchymal cells proliferating to form tumorous lesions, possibly due to chromos omal aberrations in the tuberous sclerosis (TS) complex gene. Because they are capable of dual differentiation, the argument for a hamartomatous proliferation is unsound. Recent molecular studies have shown tumor clonality and 5q deletions with a common region of deletion spanning 5q33 to 5q34, indicating a clonal neop lastic process[36-40].

Preoperative identification of AML is desirable because of differences in clinic al course and treatment between this disease and other hepatic neoplasms[41]. Although some imaging features may suggest hepatic AML, in such cases, his tological analysis must be performed. No matter whether or not the liver biopsy contains fat component, the diagnosis can be established based on myoid component positive for HMB45.