Vulvar Angiomyofibroblastoma is Molecularly Defined by Recurrent MTG1-CYP2E1 Fusions

Abstract

Angiomyofibroblastoma (AMF), a rare benign vulvovaginal mesenchymal tumor, poses a diagnostic challenge due to histologic and immunohistochemical overlap with other vulvar mesenchymal tumors. Recently, MTG1-CYP2E1 fusion transcripts were reported in 5/5 AMFs; no other genetic alterations have been described to date. Herein, we sought to investigate the frequency of the MTG1-CYP2E1 fusion and the presence of other potential genetic alterations in a cohort of AMFs (n=7, patient age range: 28-49 years). Tumors demonstrated classic morphologic features including alternating hypo/hypercellular areas, capillary channels surrounded by epithelioid/spindled tumor cells, and variable amounts of mature adipose tissue. RT-PCR for MTG1-CYP2E1 fusion, performed in all 7 cases, showed the fusion transcript in 5 of 6 cases (one case with technical failure). Two tumors, including the one lacking the fusion, were subjected to targeted next-generation sequencing (104 genes) and a sarcoma fusion assay (28 genes); the fusion negative AMF also underwent RNA sequencing. No additional mutations, copy number alterations, or fusion genes were identified with the assays employed. We conclude that the majority of AMFs harbor recurrent MTG1-CYP2E1 fusion transcripts and identification of this fusion may aid in the diagnosis.

Introduction

Vulvovaginal mesenchymal tumors including angiomyofibroblastoma (AMF), cellular angiofibroma, superficial angiomyxoma, and deep angiomyxoma may demonstrate considerable clinical, histologic, and immunohistochemical overlap (1-3). Distinction between these entities is critical as behavior ranges from benign, as in AMF, to locally aggressive with a high frequency of recurrence, as in deep angiomyxoma. There has thus been increased interest in detecting specific molecular markers for these neoplasms to aid in diagnosis. HMGA2 rearrangements have been found in a subset of deep angiomyxomas, (4) and monoallelic loss of chromosome 13q (which contains RB1) has been documented in cellular angiofibromas (5, 6), findings which have led to the use of, respectively, HMGA2 and Rb immunohistochemistry in the diagnosis of these tumors (7).

HMGA2 rearrangements are particularly helpful in the differential diagnosis between deep angiomyxoma and AMF (4, 8-10), diagnoses which may show foci of significant morphologic overlap (10). Of potential further assistance, Tajiri et al. recently identified a novel fusion transcript, MTG1-CYP2E1, in five AMFs (11). Given the identification of this novel fusion transcript, herein we investigate the frequency of MTG1-CYP2E1 fusion in a second series of AMF and perform additional molecular studies to identify other potential alterations that may underlie the pathogenesis of these tumors.

Materials and Methods

Seven AMFs were identified from the archives of the Massachusetts General Hospital, Boston, MA, and University of Iowa, Iowa City, IA, or from the personal consultations of Dr. Robert H. Young. All cases were reviewed by at least two gynecological pathologists; a subset was additionally reviewed by two bone and soft tissue pathologists (GPN and YPH). The clinical and gross features were obtained from the consultation correspondence, longitudinal medical record, and/or pathology reports. Two to 12 (average=4) hematoxylin and eosin-stained slides of tumor were available for review. Pathologic features evaluated included size, gross appearance, architecture, presence and extent of adipocytic differentiation, cytologic features, presence of perivascular cuffing, presence of multinucleated cells, and mitotic activity. Results of any immunohistochemistry performed during routine workup were recorded.

All AMFs were submitted for reverse transcription-polymerase chain reaction (RT-PCR) for MTG-CYP2E1 as previously described (11). Two AMFs were subjected to SNaPshot genotyping for analysis of mutations and copy number alterations in 104 cancer-related genes as well as a sarcoma fusion assay (28 genes); both tests were performed at the Center for Integrated Diagnostics at Massachusetts General Hospital using previously described methods (12). The AMF that was negative for the MTG1-CYP2E1 fusion transcript by RT-PCR was subjected to RNA-sequencing at Memorial Sloan Kettering Cancer Center’s Integrated Genomic Operation as described elsewhere (13).

Results

Seven AMFs were identified for inclusion in the study. Patients ranged from 28 to 49 years (median 40; Table 1). Further clinical information was available for five. All patients presented with a painless vulvar mass and underwent complete excision.

Table 1:

Clinical, pathologic and molecular features of AMFs.

	Age (yr)	Size (cm)	Histology	Cytology	MTG1- CYP2E1 RT-PCR	SNaPshot (104 genes)	Sarcoma Fusion Assay (28 genes)	RNA sequencing
Case #1	44	2.5	WC, predominantly hypocellular with perivascular aggregates, no fat	Epithelioid	+	No mutations/copy number alterations	No oncogenic fusion	NP
Case #2	40	2	WC, homogenously hypercellular with perivascular aggregates, 30% fat	Epithelioid	+	NP	NP	NP
Case #3	28	23	WC, predominantly hypercellular with prominent perivascular aggregates, 5% fat	Spindled, epithelioid, and plasmacytoid; multinucleated cells present	+	NP	NP	NP
Case #4	39	2	WC, predominantly hypocellular with perivascular aggregates, no fat	Spindled to epithelioid	+	NP	NP	NP
Case #5	40	3.1	WC, predominantly hypocellular with perivascular aggregates with collagenous stroma, no fat	Epithelioid to spindled	+	NP	NP	NP
Case #6	49	22	WC, predominantly hypocellular with intermixed hypercellular clusters, large vessels, 20% fat	Epithelioid; multinucleated cells present	Test failed	NP	NP	NP
Case #7	42	6	WC, predominantly hypocellular with some perivascular aggregates, 60% fat	Epithelioid to plasmacytoid	−	No mutations/copy number alterations	No oncogenic fusion	No oncogenic fusion

^*WC: well-circumscribed; NP, not performed.

Gross and Histologic Appearance

Tumors ranged from 2 to 23 cm in greatest dimension (median 3.1) and displayed tan-gray, rubbery cut surfaces (Figure 1a). All were well-circumscribed without a capsule (Figure 1b). They demonstrated alternating hypocellular and hypercellular areas, best viewed at low power (Figure 1c, 1e). Five tumors were predominantly hypocellular while two were predominantly hypercellular. All displayed small to medium-sized thin-walled vessels, with six tumors demonstrating perivascular cellular cuffing (Figure 1c). Tumor cells were typically spindled to epithelioid with two neoplasms showing plasmacytoid features (one extensive) (Figure 1f). Occasional multinucleated tumor cells were present in two tumors (Figure 1d). No significant nuclear atypia was noted, and the mitotic rate was less than 1/10HPF in all tumors. Four tumors had adipocytic differentiation, with the overall percent ranging from 5 to 60% (median 25%) (Figure 1e).

Figure 1, Gross and microscopic features of AMF:

The tumor is grossly (a; case 5) and microscopically (b; case 2) well-circumscribed. Hypocellular and hypercellular areas are present. Note the tumor cells cuffing thin-walled vessels (c; case 5). There are rare multinucleated cells within the hypocellular tumor stroma (d; case 6). Fusion-negative AMF is a well-circumscribed, hypocellular tumor with clustering around vessels, plasmacytoid cells and admixed fat (e, f; case 7).

Immunohistochemical Results

Immunohistochemistry performed during initial workup was available in a subset of tumors with the following results: desmin and estrogen receptor positivity in three of four and progesterone receptor positivity in four of four tumors.

MTG1-CYP2E1 fusion transcript in AMFs

RT-PCR for the MTG1-CYP2E1 fusion transcript was performed successfully in six of seven tumors and revealed the presence of the MTG1-CYP2E1 fusion in 5 cases (83.3%). The tumor lacking the fusion (case 7, Table 1) demonstrated classic features of AMF, being hypocellular with epithelioid to plasmacytoid cells and admixed adipose tissue (Figures 1e-f). To confirm these results using an orthogonal method, the AMF lacking the MTG1-CYP2E1 fusion by RT-PCR was subjected to RNA-sequencing, which validated the absence of the MTG1-CYP2E1 transcript.

Other molecular alterations in AMFs

To assess the presence of other molecular alterations, two AMFs (cases 1 and 7, the latter negative for the MTG1-CYP2E1 fusion) (Figure 4), were subjected to the SNaPshot assay. No point mutations or copy number alterations were identified. A sarcoma fusion assay (performed on cases 1 and 7) and RNA-sequencing (performed on only case 7) demonstrated no other fusions.

Discussion

AMFs were first described in 1992 by Fletcher et al (14). They typically occur in women during the fourth to fifth decades and present as a painless mass in the labia majora, usually measuring from 0.5 to 12 cm, though our study includes two larger tumors (22 and 23 cm) (9, 14-17). Histologically, they are centered in the dermis or subcutis and demonstrate alternating hyper- and hypocellular areas composed of bland spindled to epithelioid to plasmacytoid myofibroblastic cells aggregating around thin-walled vessels (15-17). Mitotic activity is absent to low, and necrosis is absent (9, 14-17). Adipocytic differentiation and multinucleated cells may be present. Rare sarcomatous transformation has been described (18). Though a diagnosis of AMF is typically rendered based on histomorphologic features, immunohistochemistry may assist in some cases as desmin, ER, and PR are typically positive; however, this profile is non-specific (1-3). Thus, due to morphologic and immunohistochemical overlap with other vulvar mesenchymal tumors, including deep angiomyxoma, diagnosis may be challenging. Recently, in an attempt to identify a pathognomonic molecular marker of AMF, MTG1-CYP2E1 fusion was identified via RNA sequencing in an index case which led to detection of the same fusion in four other AMFs, providing a possible molecular marker for the tumor (15).

These findings led us to study a separate cohort of typical AMFs to determine if MTG-CYP2E1 fusion is the pathognomonic driver or, if as in HMGA2 fusions in deep angiomyxomas, only a subset demonstrates this alteration. The majority (5/6, 83.3%), but not all, of our tumors demonstrated MTG1-CYP2E1 fusion, supporting the prior study that at least the majority of AMFs are molecularly defined by MTG1-CYP2E1 fusion. However, there is likely a subset of these tumors that lack an identifiable alteration. The tumor in our cohort that lacked the fusion (Table 1, Case 7) demonstrated typical features of AMF, as blindly and unequivocally confirmed by two soft tissue pathologists (GPN, YPH). It was a 6 cm, well-circumscribed, hypocellular tumor with plasmacytoid ER- and PR-positive cells and an associated adipose component (Figure 4). Plasmacytoid cells have been previously reported in AMFs and were noted in one of our fusion-positive cases (Table 1, Case 3) (14, 17).

When considering the primary differential diagnosis of deep angiomyxoma, mammary-type myofibroblastoma, cellular angiofibroma and superficial myofibroblastoma, a combination of morphologic, immunohistochemical, and molecular findings aid in diagnosis. Deep (aggressive) angiomyxomas which may recur locally, are typically unencapsulated hypocellular tumors with spindle to stellate cells, myxoid stroma, and infiltrative margins; foci may demonstrate near identical overlap with AMF (10). ER, PR, and desmin are typically positive in deep angiomyxoma; however, these findings are non-specific, as AMF and other tumors in the differential express these markers to variable degrees (19). HMGA2 immunohistochemistry may be of assistance as up to 68% of deep angiomyxomas are positive, with the caveat that rare AMF may also show expression (20). Mammary-type myofibroblastoma (MTMF), which is characterized by spindle cells with relatively stubby nuclei and variable adipocytic differentiation, typically demonstrates a loss of Rb expression (92%), which is particularly helpful in diagnosis in distinguishing MTMF from AMF (21, 22). Cellular angiofibromas are typically distinguished from AMF on morphologic grounds alone; however, loss of Rb is also seen in cellular angiofibroma, which may aid in diagnosis (23, 24). For tumors with overlapping morphologic features and ambiguous immunohistochemical results, molecular studies to identify HMGA2 rearrangement (33% of deep angiomyxomas) (4), chromosome 13q14 loss (mammary-type myofibroblastoma/cellular angiofibroma), or MTG1-CYP2E1 fusion (AMF) would aid in the diagnosis (1, 11, 25). Superficial myofibroblastoma is composed of bland stellate and spindle cells and lacks adipocytic differentiation. While they are thought to occur more often in the cervix and vagina, rare tumors have been reported in the vulva (26-28). MTG1-CYP2E1 fusion has been demonstrated in a subset of superficial myofibroblastomas (11), potentially suggesting a similar cell of origin with AMFs. Further, this finding raises the possibility that superficial myofibroblastoma and AMF may not be distinct neoplasms and instead represent entities within a spectrum of genetically-related tumors.

The main limitations of our study are the sample size and the quality/quantity of tissue available for analysis. One of our cases failed RT-PCR (Case 6) presumably due to the quality of the tissue and the low tumor cellularity. Additionally, the targeted sequencing analysis was performed with a small panel in a limited number of samples.

Our results confirm that most AMFs are molecularly defined by MTG1-CYP2E1 fusion. Using targeted sequencing and RNA-sequencing approaches, no other molecular alterations were identified in our single fusion-negative AMF. Future studies are warranted to ensure that this fusion is limited to AMF and superficial myofibroblastoma and to identify other genetic alterations in rare fusion-negative cases.