Clinical, Histopathological, and Immunophenotypic Spectrum of Hepatic Epithelioid Hemangioendothelioma: Eight Years’ Data of a Tertiary Care Center from North India

Abstract

Background/Aims

Epithelioid hemangioendothelioma (EHE) is an uncommon vascular tumor that commonly affects the liver. Hepatic EHE (HEHE) presents with variable clinical and histopathological features. We describe detailed clinico-histopathological features, differential diagnosis, and treatment outcomes of the cases of HEHE diagnosed in our center.

Methods

All cases of HEHE diagnosed in our institute in the last eight years (2016–2023) were reviewed (n = 8; 11 samples) (total 36 cases of EHE; 22.2%). The clinical features, radiology, histopathology, immunophenotype, molecular features, and treatment outcomes of all cases were evaluated.

Results

The median age of presentation was 49.5 years with a female: male ratio of 7:1. Abdominal pain was the commonest presentation. Approximately two-thirds of the patients had multifocal lesions. Histopathology showed purely epithelioid, predominantly epithelioid, and predominantly spindle-cell morphology in 50%, 25%, and 25%, respectively. All cases showed typical myxohyaline/fibrous stroma and organized thrombi of the portal/central veins. CD31 was the most commonly used immunostain with positivity in all cases. CAMTA1 break-apart fluorescence in situ hybridization was positive in 75% of cases, while none showed TFE3 immunopositivity. Chemotherapy was the most commonly employed therapy (n = 5) followed by surgery (n = 2). The median duration of follow-up was 26 months. Five patients were alive with disease (two patients ≥3 years), one patient died of sudden cardiac death, and two patients were lost to follow-up. Two patients developed metastatic disease at follow-up.

Conclusions

We describe the clinico-histopathological features and differential diagnosis of HEHE. This appears to be the largest case series of HEHE from India.

Epithelioid hemangioendothelioma (EHE) is an uncommon vascular malignancy characterized by classical histomorphological features in the form of epithelioid/spindle cells in a myxohyaline/fibrous stroma, characteristic immunophenotypic expression in the form of expression of vascular markers, and disease-defining molecular alteration in the form of WWTR1::CAMTA1 or YAP1::TFE3 fusion.1, 2, 3, 4, 5, 6 It commonly involves the lung, liver, bone, and soft tissue.⁷ The hepatic EHE (HEHE) is a challenging diagnosis owing to its rarity, variable presentations, and clinical, radiological, and histopathological mimickers. The response to different treatment modalities is variable. We describe the clinical, histopathological, immunophenotypic, and molecular features of HEHE diagnosed in our institute in the last eight years along with the treatment outcome and differential diagnosis.

Materials and methods

The study included all histopathology-proven cases of hepatic epithelioid hemangioendothelioma (HEHE) diagnosed in our center in the last eight years (2016–2023). This included 11 samples from eight cases (core biopsy: 10; surgical resection (left hepatectomy): 1). Five cases were received for consultation from other hospitals. The histopathological characteristics of HEHE were i) a neoplastic population comprising of predominantly epithelioid cells, ii) embedding of the neoplastic cells within a myxohyaline or fibrohyaline stroma, iii) the endothelial origin of the neoplastic cells as documented by the endothelial immunophenotype (expression of endothelial markers by the neoplastic cells).

Baseline characteristics and clinical information were retrieved from the medical records, wherever available, including age, sex, clinical presentation, extrahepatic involvement, radiological findings, treatment received, and follow-up. Hematoxylin and eosin-stained sections were available for review in all cases. Paraffin tissue blocks were also retrieved for additional ancillary techniques, including immunohistochemistry and fluorescence in situ hybridization (FISH). The gross morphology could be evaluated in the surgical resection specimen. Various histomorphological features were evaluated, including the pattern of infiltration, pattern of infiltration at the advancing edge of the tumor, predominant cytomorphology, degree of nuclear atypia, portal/central vein involvement, other portal/central venous changes, and alterations in the background hepatic parenchyma.

Immunohistochemical (IHC) staining was performed on formalin-fixed paraffin-embedded tissue. Representative paraffin blocks from all cases were stained with at least one endothelial marker such as CD31, CD34, FLI-1, von Willebrand factor (vWF), or ERG. Other immunohistochemical stains were deployed as per the histomorphology and the differential diagnosis, considered for each case at the time of reporting. All stains were performed on Ventana Benchmark XT. The antibodies used, their sources, and their dilutions are listed in Table 1.

Table 1.

Details of the Immunohistochemical Stains (IHC) Used.

S.No.	IHC	Clone	Dilution	Site of positivity	Internal control
1.	CD34	Cell marque (QBEnd/10) mouse monoclonal antibody	1:200	Membranocytoplasmic	Native endothelium
2.	CD31	Cell marque (JC70) mouse monoclonal antibody	1:100	Membranocytoplasmic	Native endothelium
3.	FLI-1	Cell marque (MRQ-1) mouse monoclonal antibody	1:50	Nuclear	Native endothelium
4.	vWF	Dako A0082 von Willebrand factor polyclonal rabbit antibody	1:100	Cytoplasmic	Native endothelium
5.	ERG	Cell marque (EP111) rabbit monoclonal antibody.	1:200	Nuclear	Native endothelium
6.	Pan-CK	Cell marque Cytokeratin cocktail (AE1/AE3) mouse monoclonal antibody	1:300	Membranocytoplasmic	Native hepatocytes and bile ducts
7.	CK7	Cell marque (OV-TL 12/30) mouse monoclonal antibody	1:300	Membranocytoplasmic	Native bile ducts
8.	CK20	Cell marque (Ks20.8) mouse monoclonal antibody	1:300	Membranocytoplasmic	None; External control
9.	EMA	Cell marque (E29) mouse monoclonal antibody	1:100	Apical membranous	Native bile ducts
10.	Arginase-1	Cell marque (SP156) rabbit monoclonal antibody	1:100	Nucleocytoplasmic	Native hepatocytes
11.	CD117	Dako A4502 polyclonal rabbit antibody	1:500	Membranocytoplasmic	Mast cells
12.	TFE3	Cell marque (MRQ-37) rabbit monoclonal antibody	1:100	Nuclear	None; External control

IHC, Immunohistochemistry; TFE3, Transcription factor binding to IGHM enhancer 3; vWF, von Willebrand factor; ERG: Erythroblast transformation-specific (ETS)-related gene; EMA: Epithelial membrane antigen.

The CytoTest CAMTA1 break-apart fluorescence in situ hybridization (FISH) probe, labeled with CytoOrange (Locus specific probe CAMTA1 3′) and CytoGreen (Locus specific probe CAMTA1 5’) was used for interphase FISH.

Results

A total number of 36 cases of EHE were diagnosed in our institute from 2016 to 2023. This encompasses 14 cases of bone and soft tissue EHE (n = 14/36; 38.9%), 11 cases of EHE occurring in the skin (n = 11/36; 30.5%), eight cases of hepatic EHE (HEHE) (n = 8/36; 22.2%), two cases of pulmonary EHE (n = 2/36; 5.6%), and one case of EHE presenting as an omental nodule (n = 1/36; 2.8%). Thus, a little less than one-fourth of cases of EHE in our institute occur in the liver.

Clinical findings

The median age of presentation was 49.5 years (IQR = 39.5–57.5 years) (range 35–64 years). The female: male ratio was 7:1. The most common presenting feature was abdominal pain (n = 7/8; 87.5%), followed by loss of weight (n = 2/8; 25%), loss of appetite (n = 1/8; 12.5%), vomiting (n = 1/8; 12.5%), and melena (n = 1/8; 12.5%) (Table 2). The duration of symptoms prior to the diagnosis ranged from 2 to 6 months.

Table 2.

Clinicoradiological Features of Patients With Hepatic Epithelioid Hemangioendothelioma (HEHE).

Case no.	Age/Sex	Clinical presentation	Radiological findings (Unifocal/multifocal)	Extrahepatic involvement	Treatment received	Outcome	Additional information
1.	43/F	Pain abdomen	Multifocal hypodense lesions in right and left lobe	Multiple sub centimetric, nonhypermetabolic nodules seen in B/L lung fields	Chemotherapy (pazopanib 400 mg BID)	Stable disease; On chemotherapy and regular FU. (AWD: 8 mths)	–
2.	56/M	Pain abdomen	NA	NA	NA	(AWD: 16 mths)	–
3.	42/F	Pain abdomen, Melena, Vomiting, Loss of weight	Hepatomegaly with multiple, peripherally enhancing, hypodense lesions in B/L lobes of liver, largest measuring 30 × 28 mm in segment VII	FDG-avid bilateral lung nodules, mesenteric lymph nodes and omental deposit	Chemotherapy (Adriamycin), planned for immunotherapy	Persistent disease; On chemotherapy and regular FU; recent appearance of omental deposit (AWD: 13 mths)	FU liver biopsies at 4 and 12 months confirm persistent tumor
4.	37/F	NA	NA	NA	NA	LFU	–
5.	58/F	Pain abdomen	Multiple, discrete and confluent hypodense lesions in segments V, VI and VII/VIII	–	Chemotherapy (Lenvatinib and Everolimus)	Stable disease; On chemotherapy and regular FU (AWD: 36 mths)	–
6.	35/F	Pain abdomen	Multiple, heterogeneously enhancing, hypodense lesions of variable size in segment VII and VIII	–	Discontinued chemotherapy and currently on CAM	Stable disease; On irregular FU (AWD: 36 mths)	–
7.	57/F	Pain abdomen	31 × 25 mm single, well defined hypodense lesion in segment III	–	Underwent resection (Left Hepatectomy)	LFU	–
8.	64/F	Pain abdomen, Loss of weight & Appetite	NA	NA	Underwent surgery from outside. Chemotherapy started, however discontinued, followed by CAM intake	Patient died after 72 months (due to SCD) (DOC)	–

B/L, Bilateral; FDG, 18-Fluoro-deoxyglucose; CAM, Complementary and Alternative Medicine; FU, Follow-Up; LFU, Lost to Follow-Up; SCD, Sudden Cardiac Death, AWD, Alive with disease, DOC, Died of other cause, DOD, Died of disease; HEHE, Hepatic Epithelioid Hemangioendothelioma.

Four patients (Case 1, 3, 5, 6) (n = 4/6; 66.7%) had multifocal lesions (Case 1 and Case 3 showed lesions in both lobes; Case 5 and Case 6 showed multiple lesions in the right lobe). One patient (Case 7) had unifocal lesion measuring 31 mm situated in the left lobe. The size of the largest lesion ranged from 30 to 31 mm. On CT scan, these lesions were hypodense, heterogeneously enhancing. PET-CT findings were available for three patients and showed intensely hypermetabolic, peripherally enhancing lesions (Table 2).

Histopathology

The gross morphology could be appreciated in the left hepatectomy specimen (Case 7). Grossly, the tumor in the left hepatectomy specimen measured 31x25 mm. The mass had ill-defined margins with a gray-white, glistening, firm-cut surface (Figure 1a).

Figure 1.

Gross pictures of Case 7 (left hepatectomy; 1a) and the histopathological features of the tumor (1b-1e): The left hepatectomy specimen highlighting an ill-defined gray-white, glistening mass (thick white arrow). The tumor cells were arranged in cord formation and were large polygonal with abundant eosinophilic cytoplasm, vesicular chromatin, and small nucleoli (Hematoxylin and eosin, 200x, 1b). Blister cells are highlighted in black arrows (hematoxylin and eosin, 400x, 1b-1c). Myxohyaline stroma (Hematoxylin and eosin, 100x, 1d) highlighted by AB-PAS stain (100x, 1e).

Microscopically, the tumors showed sheets of cells with ill-defined, infiltrative growth patterns as cords and single cells. The neoplastic cells showed epithelioid and spindled/dendritic morphology. The epithelioid cells were arranged in a single-cell pattern with focal cord formation and were large polygonal with abundant pale eosinophilic cytoplasm, vesicular chromatin, and small nucleoli (Figure 1b). The epithelioid morphology was noted in all cases, although in a variable proportion. Purely epithelioid morphology was seen in five patients (n = 4/8; 50%), whereas two patients (n = 2/8; 25%) showed a predominant epithelioid component (>70%) with a minor spindle/dendritic cell component. A predominantly spindled/dendritic cell (>70%) component with a minor epithelioid cell component was noted in two patients (n = 2/8; 25%). Blister cells/signet ring-like cells (epithelioid/spindled cells with intracytoplasmic vacuoles/lumina) (Figure 1c) were seen in all cases, although both cases with a predominant-spindle/dendritic cell morphology showed very occasional blister cells. Most of the cases showed moderate (n = 5/8; 62.5%) nuclear atypia followed by mild nuclear atypia (n = 2/8; 25%). Only a single case showed marked cytological atypia (n = 1/8; 12.5%). The mitotic count was variable from case to case. Four cases (50%) showed no mitoses; 3 cases (37.5%) had ≤2 mitosis/2 mm² and a single case (12.5%) had 2–4 mitosis/2 mm² (Table 3).

Table 3.

Histopathological Features of Patients with Hepatic Epithelioid Hemangioendothelioma (HEHE).

Case no.	Tissue	Cell Cyto-morphology	Atypia	Portal vein changes	Native liver parenchyma	Sinusoidal infiltration	Synchronous meta-stasis	Mitotic count (/2 mm2)	IHC	FISH for WWTR1-CAMTA1 rearrangement
1.	Biopsy, Blocks for review	Spindled (70%) to epithelioid (30%)	Mild	Tumor emboli seen (Single cells)	Only occasional incomplete portal tracts seen	Absent	–	Not seen	CD34+ CD31+ FLI-1+ CK -	Tissue NA
2.	Biopsy	Epithelioid	Moderate	Obliteration	Tumor advancing edge shows entrapped atrophic hepatocytes and an occasional portal tract	Absent	–	Not seen	CD31+ FLI-1+ ERG + Arginase - CD117-	Present
3.	Biopsy	Epithelioid (70%) to spindled (30%)	Moderate	Tumor emboli + (Tufted pattern)	Azonal macrovesicular steatosis (<33%) along with lobular inflammation (2–4/20x field), and regenerative nodules	Present	–	Not seen	CD34+ ERG+ CK-	Present
4.	Biopsy, blocks for review	Epithelioid (80%) to spindled (20%)	Moderate	Obliteration +	Occasional entrapped portal tracts seen at the advancing edge of the tumor	Present	–	≤2	CD31+ CK-	Tissue NA
5.	Biopsy, blocks for review	Epithelioid	Marked	Obliteration +, Tumor emboli +	Tumor advancing edge shows entrapped portal tracts. However, no native hepatic lobule present	Absent	–	≤2	CD34+, CD31 diffuse +, CK7 focal + TFE3-	Non interpretable
6.	Biopsy, blocks for review	Epithelioid	Moderate	Obliteration +, Tumor emboli +	Sinusoidal and central venous dilatation. Portal tracts show dilated inlet venules	Present	–	Not seen	CD31+, vWF + TFE3-	Absent
7.	Left hepatectomy	Epithelioid	Mild	Obliteration +, Tumor emboli + (Tufted pattern)	Centrizonal macrovesicular steatosis (6–33%), arterial hyalinosis, surgical hepatitis and F2 fibrosis	Present	Lymph node	Not seen	CD31+, vWF+, CK20-, EMA -	Present
8.	Biopsy,blocks for review	Spindled (90%) to Epithelioid (10%)	Moderate	Obliteration +	Sinusoidal dilatation and arterial hyalinosis	Absent	–	≤2	CD31+ CK-	Tissue NA

HEHE, Hepatic Epithelioid Hemangioendothelioma; IHC, Immunohistochemistry; NA, Not Available; TFE3, Transcription factor binding to IGHM enhancer 3; vWF, von Willebrand factor; FISH, Fluorescence in situ hybridization; CAMTA1, Calmodulin-binding transcription activator 1; WWTR1, WW domain containing transcription regulator 1; ERG: Erythroblast transformation-specific (ETS)-related gene; EMA: Epithelial membrane antigen.

The neoplastic cells were embedded in a pale-eosinophilic (fibrohyaline) to pale-basophilic (myxohyaline) stroma. Alcian blue-PAS stain performed in an occasional case highlighted Alcian blue-positive myxoid stroma while the hepatocytes were PAS-positive (Figures 1d-1e). Focal areas of hemorrhage and hemosiderin-laden macrophages were also seen in three cases (n = 3; 37.5%). A mild, mixed stromal infiltrate comprising lymphocytes, polymorphs, and plasma cells was seen in all cases. There was no stromal calcification in any case.

All tumors showed an infiltrative pattern. The advancing edge of the tumor showed entrapped islands of atrophic hepatocytes and portal tracts. A sinusoidal pattern of infiltration (intrasinusoidal presence of the tumor cells usually associated with the atrophy of the hepatocytes) was seen in 5 cases (n = 5/8; 62.5%), predominantly in the form of single cells. Infiltration of the portal vein branches was noted in 6 cases (n = 6/8; 75%) in the form of single cells as well as the occasional appearance of small glomeruloid, tuft-like clusters (Figure 2a). No evidence of perineural infiltration was seen in any of the cases.

Figure 2.

Histopathological features of the vessels, immunohistochemistry, and FISH: Infiltration of the portal veins by glomeruloid tufts of tumor cells (black arrowheads) (hematoxylin and eosin, 200x, 2a). Intrasinusoidal infiltration by the tufted, multinucleated tumor cells (black dotted arrows) beyond the confines of the main mass (hematoxylin and eosin, 400x, 2b). Portal veno-occlusive tumor emboli (thick black arrow) (hematoxylin and eosin, 200x, 2c). Tumor cells showed diffuse membranous CD31 (400x, 2d) and diffuse nuclear ERG positive (200x, 2e). CAMTA1 break-apart FISH highlighting separation of the red (3′ CAMTA1 locus) and green (5′ CAMTA1 locus) signals (thin white arrows) at least by two-signal distance (2f).

The background liver parenchyma was available for evaluation in two cases (one biopsy and one resection case). These cases were evaluated for the demonstration of tumor cells beyond the confines of the main mass. The resected specimen showed the presence of tumor cells beyond the confines of the main tumor in the form of multinucleated cells, tufted cells, and cerebriform single cells with hyperchromatic nuclei within the sinusoids at a distance of <1 mm from the main mass (Figure 2b).

The benign changes in the background hepatic parenchyma showed a spectrum of histological features ranging from macrovesicular steatosis (azonal and centrizonal) (n = 2/2; 100%), foci of lobular inflammation (n = 2/2; 100%), and occasional formation of regenerative nodules (n = 1/2; 50%). Dilatation of the central veins and the sinusoids, and arterial hyalinosis were noted in both cases (n = 2/2; 100%). A single case showed portal fibrosis with incomplete septa formation (n = 1/2; 50%).

The intratumoral and extratumoral portal tracts frequently showed lymphoid infiltrates (n = 4/8; 50%) and dilated inlet venules (n = 2/8; 25%). Obliteration of the portal veins and/or terminal hepatic venules by organized thrombi with/without recanalization of their lumina was noted in all cases (n = 8/8; 100%). A few vessels with the organized thrombi were associated with tumor emboli (Figure 2c).

The results of immunohistochemical staining are summarized in Table 3. All tumors showed positive membranous staining for at least one endothelial marker. Five cases were tested with more than one endothelial marker, while only one marker was employed in three cases due to less amount of available tissue. CD31 was the most commonly employed endothelial marker (n = 6), followed by CD34 (n = 3), vWF (n = 2), FLI-1 (n = 2), and ERG (n = 2) (Figure 2d–e). These markers showed diffuse, strong immunopositivity wherever employed. Pancytokeratin (AE1/AE3) was employed in four cases and was negative in all cases. CK7 immunostain was employed in one case that showed focal strong membranocytoplasmic CK7 positivity (n = 1/1; 100%). CK20, EMA, Arginase, and CD117 were employed in one case each and all were negative (Table 3).

Tissue for FISH was available in five cases and it was noncontributory in one case owing to a lack of interpretable signals. Of these, 3 (n = 3/4; 75%) cases were positive for the CAMTA1 rearrangement using the CytoTest CAMTA1 break-apart FISH probe (Figure 2f). A single case was negative for CAMTA1 rearrangement. TFE3 immunohistochemistry was employed in both cases, negative for CAMTA1 rearrangement/cases with not-interpretable FISH results. It was negative in both cases.

There was no histomorphological distinction between the CAMTA1-positive and CAMTA1-negative HEHE.

Management and follow-up

The treatment history of six patients was obtained from the archives (Table 2). Five patients received chemotherapy, while two individuals opted out and chose complementary and alternative medicine (CAM) (Case 6 and Case 8). The chemotherapy regime included pazopanib monotherapy, lenvatinib and everolimus combination, and adriamycin-based therapy. The only patient with a unifocal liver lesion in segment III and no extrahepatic involvement underwent a left hepatectomy in the institute (Case 7). One other patient underwent complete resection of the tumor from outside, attained negative margins, started on chemotherapy, discontinued chemotherapy, and opted for CAM intake instead (Case 8).

Adequate follow-up data were available for six patients, and two patients were lost to follow-up. The median duration of follow-up was 26 months (range 8–72 months). One patient (Case 8) who underwent complete resection of the tumor from outside and opted for CAM intake instead of chemotherapy, survived approximately 6 years following the diagnosis and succumbed to sudden cardiac death. Five patients were alive with the disease (two patients ≥3 years). Metastatic deposits were seen in two patients (Case 1 and Case 3) in the form of bilateral pulmonary nodules (Case 1 and Case 3), intraabdominal lymph nodes (Case 3), and omental deposits (Case 3). Both Case 1 and Case 3 are currently on therapy and are alive with disease at follow-up of 8 and 13 months, respectively.

Discussion

Epithelioid hemangioendothelioma (EHE) is a malignant endothelial neoplasm that usually affects the liver, lung, bone, or soft tissue. The hepatic epithelioid hemangioendothelioma (HEHE) is an uncommon vascular malignancy of the liver composed of epithelioid cells, usually embedded in a variable myxohyaline, myxofibrous, or fibrous stroma. The neoplastic cells in HEHE express endothelial markers. The molecular signature of the tumor is characterized usually by the fusion of WWTR1 (WW domain containing transcription regulator 1)::CAMTA1 (calmodulin-binding transcription activator 1) gene.^1,8 Rarely YAP1 (yes-associated protein)::TFE3 (transcription factor binding to IGHM enhancer 3) gene fusion, or very occasionally WWTR1::TFE3 fusion can also occur.^4,9 We have analyzed the clinical, histopathological, and immunohistochemical spectrum of all retrospective cases of HEHE reported in the last eight years from our institute. This appears to be the largest series of HEHE from India to date.

EHE dominantly involves the bone and soft tissue, liver, and lung.⁷ The index series highlights a little less than one-fourth (22.2%) of reported cases of EHE to be hepatic in origin. A similar series by Flucke et al. comprising 39 cases of EHE showed nearly two-thirds of the cases to originate from the bone and soft tissue.¹⁰ However, only two cases (5.1%) of EHE in the series by Flucke et al. were hepatic in origin.¹⁰ The median age of presentation in the index study was 49.5 years, and the youngest individual was a 35-year-old female. This is comparable to the large studies conducted by Makhlouf et al. and Ishak et al. with a median age of 46 years and 49.6 years, respectively.^5,6 A literature review by Mehrabi et al. also highlighted a mean age of 41.7 years.¹¹ No individual in the index series presented at an age of <20 years while approximately 5% of individuals were <20 years of age in the series by Makhlouf et al.⁵ Our series showed a striking female preponderance with a female: male ratio of 7:1. A similar ratio was also documented by Agostini-Vulaj et al.¹² Rest of the series showed a female preponderance although it was not as striking as the current series .^5,6,11 None of our cases were incidentally detected, unlike the series by Makhlouf et al. where nearly 40% cases were incidental.⁵ Abdominal pain was the most common manifestation in the index series, akin to the previous studies.^5,6,11 Other manifestations, including loss of weight and appetite, vomiting, and melena were similar to the existing literature.^5,6,11

Radiologically, nearly two-thirds of our cases appeared to be multifocal with/without involvement of both lobes of the liver. A total of two cases showed multiple metastases involving the lungs, intraabdominal lymph nodes, and omentum. Multifocality and synchronous/metachronous intraabdominal/pulmonary metastases were commonly seen in our series, an observation similar to the previous studies.^5,6,11

Histomorphology in all cases showed predominantly epithelioid cells embedded in a variably myxohyaline to fibrohyaline to fibrous stroma. These epithelioid cells are usually arranged in single cells or focally as cords and are polygonal in shape with central to eccentric round nuclei, vesicular chromatin, inconspicuous to occasionally prominent nucleoli, and moderate to abundant variably eosinophilic cytoplasm. Intracytoplasmic lumina are useful clues to the diagnosis. A spindle cell/dendritic cell component is also seen admixed with the epithelioid cell component and the proportion of the spindle cell component can vary. The spindle cells, also described as dendritic cells, can have a wispy spindle, stellate, or plump ovoid shape.^5,6 An intermediate cell population with features intermediate between the epithelioid and dendritic/spindle cells is also seen.¹² Consequent to the variable histomorphology, the differential diagnosis varies from case to case. Hepatocellular carcinoma (HCC), primary/metastatic adenocarcinoma, epithelioid angiomyolipoma (E-AML), epithelioid leiomyosarcoma (E-LMS), metastatic melanoma, and rare cases of metastatic germ cell tumors are the differential diagnosis of a predominantly epithelioid tumor.¹³ Rare cases of hepatic follicular dendritic cell sarcoma can also be considered in the differential diagnosis.¹⁴ The diagnosis of HEHE in a small biopsy is usually aided by the characteristic stroma along with the epithelioid cells. The stroma, if predominantly fibrous, mimics desmoplasia. The cellularity in a small biopsy can be variable, and the tumor cells could be difficult to identify within a fibrous/myxofibrous stroma. The blister cells can mimic the cells of an adenocarcinoma with intracellular lumina. The differential diagnosis of a predominantly spindle cell tumor encompasses sarcomatoid HCC, metastatic gastrointestinal stromal tumor (GIST), primary/secondary leiomyosarcoma (LMS), metastatic melanoma, and solitary fibrous tumor.^15,16 Immunohistochemistry is helpful in this regard and often aids in clinching the diagnosis. HCC shows Heppar-1 and/or arginase positivity, although immunonegativity for these markers in a small biopsy in a case of sarcomatoid HCC is expected. HCC usually does not show a stroma akin to HEHE and loss of/fragmented reticulin stain pattern usually points toward a diagnosis of HCC. Pancytokeratin (AE1/AE3) expression in EHE is documented in approximately 30% of cases.¹⁰ Low-molecular-weight cytokeratin (CAM 5.2) positivity is demonstrated in approximately 12% of cases, and nearly 50% of EHE shows CK7 positivity.^12,17 These observations mostly corroborate our findings. The pancytokeratin positivity can be focal and could be the reason for the negative result in most of our cases, as largely our series comprised of biopsied material. Thus, pancytokeratin negativity/focal/faint positivity along with positivity for vascular markers (CD34/CD31/ERG/FLI1/vWF) aid in the distinction of HEHE from a primary/metastatic carcinoma. However, low-molecular-weight cytokeratin or pancytokeratin positivity should be followed by a panel of vascular markers if the histomorphology is suggestive of HEHE. The diagnosis of different other tumors require specific markers (HMB45/melan A for E-AML and metastatic melanoma, smooth muscle actin for LMS, and E-AML, KIT, and DOG1 for GIST, STAT6 for solitary fibrous tumor, SALL4/Oct3/4 for germ cell tumor) and needs to be judiciously employed in a case-to-case basis depending on the clinical features and histomorphology.

We have evaluated the extratumoral patterns of infiltration beyond the confines of the main tumor mass in cases where substantial hepatic parenchyma was present adjacent to the tumor. We documented an intrasinusoidal pattern of infiltration in the form of multinucleated cells, hyperchromatic cerebriform cells, and tufted cells in the single case with adequate liver parenchyma. However, we failed to demonstrate this pattern of infiltration in a biopsy, possibly owing to a limited tissue sample. This pattern, as described by Agostini-Vulaj et al. can aid in the diagnosis, especially in a limited sample obtained from the periphery of the main tumor mass.¹² In such cases, a demonstration of such atypical-looking cells or tufts within the sinusoids or the central veins with ERG positivity can be highly suggestive of a diagnosis of HEHE.

The other benign features in the background hepatic parenchyma dominantly include various porto-lobular vascular changes. The obliteration of the portal and central veins in both intratumoral and extratumoral locations, intrasinusoidal infiltration by the tumor cells, and local compression by the tumor predispose vascular changes in the form of recanalizing thrombi, herniated inlet venules, venous dilatation, and the formation of regenerative nodules. Thus, these features in the relevant clinical background can be nonspecific clues to the diagnosis of HEHE in small biopsies.

The EHE is characterized by the WWTR1::CAMTA1 fusion in more than 90% of cases, with a small subset showing YAP1::TFE3 fusion.1, 2, 3, 4^,10 In physiological conditions, WWTR1 protein is highly expressed in the vascular endothelial cells, while the expression of CAMTA1 protein is limited to the brain. The fusion causes inappropriate activation and expression of CAMTA1 protein through the promoter-switch mechanism.¹ The CAMTA1 protein expression can also be documented by immunohistochemistry other than break-apart FISH. The fusion of YAP1::TFE3 causes expression of TFE3 protein, demonstrable via strong expression of TFE3 by immunohistochemistry.⁴ YAP::TFE3 fusion produces a distinct histomorphology different from a conventional EHE in the form of i) voluminous eosinophilic cytoplasm akin to an oncocyte, ii) more vasoformative areas and pseudo-alveolar structures, iii) histiocytoid morphology, iv) occasional biphasic morphology with both pseudo-alveolar and conventional areas, and v) the absence of the typical stroma of EHE.⁴ Very rarely, a noncanonical fusion between WWTR1 and TFE3 is also documented in EHE.⁹ The index series showed the presence of disease-defining CAMTA1 rearrangement by FISH in 75% of cases. None of our cases showed TFE3-positivity by immunohistochemistry. No histological difference was noted between the CAMTA1-positive and negative HEHE.

The treatment and outcome of HEHE are variable. It ranges from surgical resection, liver transplant, and chemotherapy to even watchful waiting. The USA National Cancer Data Base (2004–2018) highlighted a nonsurgical approach in more than 65% of the patients of HEHE followed by surgical resection and liver transplant in approximately 22% and 10% of HEHE.¹⁸ In contrast, a review of the literature of all published cases and series from 1984 to 2005 showed liver transplant to be the dominant modality of treatment (∼45%), followed by no treatment (∼25%), chemotherapy/radiotherapy (21%), and surgical resection (∼9%).¹¹ The majority of our cases received chemotherapy alone or along with surgery. The overall median survival was maximum following liver transplant (111 months), followed by watchful waiting (80 months), and surgical resection (69 months), according to the USA National Cancer Data Base (2004–2018).¹⁸ The five-year survival rate was maximum following the liver resection, liver transplant, and chemo/radiation and dismal without therapy.¹¹ In the index series also, the large majority of our patients were alive with disease, although an exact correlation of the histomorphology with the type of treatment and treatment outcome cannot be established.

We have described a series of eight cases of HEHE from a tertiary care center in North India, including the clinical presentation, histopathological, immunophenotypic, and molecular features, and treatment outcome. This study highlights the differential diagnosis, histopathological clues to the diagnosis, and the importance and implication of the diagnosis of this entity. Further, this appears to be the largest series of HEHE from India to date.

Credit authorship contribution statement

GK collected and analyzed the data and prepared the first draft of the manuscript. SM, DC, UNS, and AsD were involved in the diagnosis of the cases along with the evaluation of the histomorphology and immunohistochemistry. AB was involved in the analysis and interpretation of FISH and critically reviewed the manuscript. TD, DK, UD, and AjD were involvement in the clinical diagnosis and management of the cases and critically reviewed the manuscript. LK was involved in the surgical management of the cases and critically reviewed the manuscript. SM conceptualized the idea, was involved in the final tabulation and analysis, and prepared the final draft of the manuscript.

The authors thank Prof. Dr. R. Nada, Dept of Histopathology for her help.

Funding

The authors received no funds for this study.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.