Emerging Mesenchymal Tumors

Abstract

Two new soft tissue entities are described. First, GL1-altered epitheloid mesenchymal tumors comprise two reported molecular subtypes: one with GLI1 fusion and the other with GLI1 amplification, both of which show increased expression of GLI1 RNAm. While most tumors described are malignant, a small subset provisionally termed distinctive nested glomoid neoplasm pursue an indolent course even with regional lymph nodes metastatic spread. There is no known association between prognostication and molecular subtype. Second, the so-called pseudoendocrine sarcoma is considered to be a sarcoma of intermediary grade with a morphology reminiscent of neuroendocrine tumors and characterized by recurrent mutations of the B-catenin gene. This lesion occurs more and less in elderly patients and is most often found in paravertebral area, which makes complete resection surgically relatively challenging.

1. Introduction

Many soft tissue tumors are currently emerging as new diagnostic entities, a substantial number of them being referenced solely according to their characteristic recurrent molecular alterations, such as PRD10-, N1RD-, or KMT2A-rearrangements, to name just a few [1, 2, 3], if only because of the considerable morphological overlap that is observed with already classified lesions in the WHO [4] repertoire widening considerably the range of differential diagnoses. The result is also the confirmation of the importance that molecular genetics is gaining with increasing momentum within the diagnostic armamentarium.

This chapter covers two newcomers, GLI1-altered mesenchymal tumors and pseudoendocrine sarcomas, both presenting cytoarchitecturally as small round and epithelioid cell proliferation.

2. GLI1-altered mesenchymal neoplasms

2.1 General considerations

GLI1 (or Glioma-associated oncogene homolog 1) is a zinc finger-type transcription factor of the krüpell family of proteins playing a distinctive role as a downstream effector in the terminal region of the hedgehog pathway. Among the three known isoforms of GLI (GLI1, 2, and 3), GLI1 appears to single out as the only one entirely engaged in transcriptional activation behavior [5, 6]. Alterations of the gene lead to the recognition of two molecular subtypes of the generally small round epithelioid cell tumor with uniform morphology currently encountered and forwarded as emerging lesions, those with amplification and those with fusions. Among the cases thus far reported, our assessment shows that roughly two-thirds belong to the fusion subtype and one-third to the amplified subtype.

The first description of a GLI1-altered mesenchymal tumor dates from 2004 harboring the translocation t(7; 12) with fusion of GLI1 with the ubiquitous B-actin gene (GLI1::ACTB) [7]. This was a lesion of spindle cell morphology labeled as pericytoma or of pericytic phenotype described in soft tissue, bone, and stomach. Although appearing “myopericytomatous” with perivascular features, it is nevertheless distinct from classic myopericytoma at least at the molecular level where the latter exhibits a quite different gene profile characteristic of its pathogenetic make-up: recurrent mutations of PDGFRB that define myopericytomatosis and myofibroma as well [4]. Subsequently, two non-myopericytomatous lesions with GLI1 fusions were classified, located in the gastric wall: one benign with spindle cell morphology, plexiform fibromyxoma [8, 9] and the other malignant and biphasic, gastroblastoma [10].

2.2 Clinical aspects (including prognostication) and gross pathology

Characterization of emerging GLI1-altered tumors is ongoing and combines multifactorial features [11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22]. Their occurrence spans the broadest possible age group between 1 and 80 years, the mean age being 35–41 years, with roughly equal frequency in males and females.

Primary tumors can involve practically any site including soft tissue (trunk including chest wall and abdomino-pelvic regions, limbs, and extremities), head, and neck representing globally 40% of cases (of which around 75% target specifically the tongue), the viscera (lung, endometrium, ovary, kidney, genito-urinary region, and gastro-instestinal tract) and bone such as the tibia, vertebrae, or the scapula.

Based on the published data, we find that 82% of cases are reportedly malignant, with diverse metastatic sites, the most frequently involved being the brain, lung, and lymph nodes. Whereas the remaining 18%, conveniently and provisionally coined “Distinctive Glomoid Nested Tumors,” have been observed to pursue an indolent course with no metastasis and no tumor-induced death. If local recurrences have been documented, they are presumably due to too narrowly obtained safe margins at excision. No difference in prognosis related to a particular molecular alteration has been noted thus far.

Tumors may be deep-seated or superficial arising in the subcutis. They may measure anywhere from 0.9 to 12 cm, the median size being 3–5 cm. They are generally lobulated, compact or partially cystic, white-colored, or sometimes hemorrhagic. Overtly malignant cases have infiltrative borders.

2.3 Light microscopy features

Tumors are generally composed of multinodular sheets of nested or trabecular small round monotonous cells (Figure 1) with more or less vesicular nuclei or lightly spread chromatin, and separated by fibrous septa. There is possible focal spindling, and only rarely a biphasic architecture. On most occasions, these cells are embedded in a delicately vascularized stroma, which can be myxoid, conveying a perivascular configuration. Necrosis is an eventuality. A very frequent feature is lymphovascular invasion [11, 12, 16, 20, 21].

Figure 1.

GLI1-altered mesenchymal tumor- distinctive nested Glomoid neoplasm (courtesy of Dr. D Papke, Brigham and Women’s hospital, Harvard Medical School, Boston, USA).

2.4 Immunohistochemical aspects

No consistent immunophenotypic profile has been described, except the almost constant GLI1 expression (Figure 2). Parrack et al. found GLI1 nuclear or nucleo-cytoplasmic tumor positivity with 98% specificity and 91.3% sensitivity using the C1 clone antibody of Santa Cruz in test cohort and control cohort cases [21]. S100 protein and smooth muscle actin immunostaining are observed in a little less than half of cases and keratin AE1-AE3 in 15% of cases, at least focally [11, 13, 20]. Neuroendocrine markers such as CD56 [11, 12] and synaptophysin [14] are reported positive albeit with no clear rationale. More importantly, MdM2 and CDK4 [12, 20] as well as STAT6 [20] have been proven to be not surprisingly overexpressed in GLI1-amplified lesions, the genes coding for these proteins being part of the amplicon 12q13–15 with possible co-amplifications with GLI1.

Figure 2.

Diffuse immunohistochemical expression of GLI1 antibody in GLI1-altered mesenchymal tumor (courtesy of Dr. D Papke, Brigham and Women’s hospital, Harvard Medical School).

In tumors arising, especially in the oropharynx and tonsils, P16 is usually positive, probably due to the negative feedback loop whereby Rb silencing by hyperphosphorylation-inducing GLI1-activated cyclin/cyclin dependant kinase complexes leads to P16 overexpression [17]. Thus purportedly GLI1-activated Cyclin D1 can be found to be positive. A number of tumors are immunostain with CD10 and BCOR as well [18].

2.5 Molecular genetics

In GLI1-rearranged tumors, various fusion partners of the gene have been identified by sequencing methods (RNA sequencing), including, for example, ACTB1, PTCH1, FOXO4, PAMR1, HNRNPA1, NEAT1, and QTXNIP. Promoter swapping seems to be the generally accepted mechanism resulting in GLI1 oncogenic activation. GLI1 rearrangements as well as amplifications are readily highlighted by fluorescent in situ hybridization (FISH) techniques. Some cases presumably show rearrangement and concomitant amplification of the gene with discordant FISH and RNAseq results and attributed by some authors to genetic instability phenomenon [14].

Moreover, since GLI1 shares with DDIT3 the locus 12q13.3, some seven authors propose re-arrangements of DDIT3 as observed with some commercial break-apart probes (VYSIS or CytoCELL) as a surrogate marker for GLI1 rearrangement [17].

GLI1 highly amplified lesions can harbor co-amplifications not only of some neighboring genes but of the whole oncogene-rich amplicon 12q13–15 as demonstrated by molecular karyotyping, the most frequently involved genes being MdM2, CDK4, HMGA2, STAT6, and DDIT3 and much less frequently the genes LRP1, TSPAN31, FRS2, and ARHGEF25 [14].

2.6 Major differential diagnoses

Undifferentiated carcinoma and small round cell sarcomas of unknown differentiation (SRCSUD) represent the bulk of differential diagnoses, including the currently emerging pseudoendocrine sarcoma (cf. infra). In this context, variants of dedifferentiated liposarcoma morphologically similar to SRCSUD must be recognized for they can immunostain for GLI1 [21].

Neuroendocrine tumors (carcinoma, paraganglioma) are to be considered because, like these, GLI1-altered tumors can express CD56 and/or synaptophysin (cf supra par.2.5). Other pitfalls are ossifying fibro-myxoid tumor (nonossifiying variant) neuroblastoma, neuroectodermal tumors, (including melanocytic tumors), PEComa, and glomus tumors. GLI1-altered tumors share with myoepithelioma possible S100 protein positivity. A fairly recent case report describes a myoepithelioma with GLI1 fusion (TUBA1::GLI1) without any other alterations known in classic myoepithelioma [23]. According to our view, if this is confirmed by other reports, it raises at least the possibility of the existence of a lesional continuum between conventional myoepithelioma at one end and GLI1-fusion tumor at the other. High-grade endometrial stromal sarcoma can be a mimic of gynecologically located GLI1-altered tumors inasmuch as both lesions can express Cyclin D1, CD10, and BCOR (cf supra par.2.5). SDH-deficient gastrointestinal stromal tumor should also be mentioned as a differential diagnosis if consistent with primary tumor location. In head and neck localizations, GLI1- tumor remains an alternative possibility to HPV-induced oropharyngeal neuroendocrine carcinoma sharing with it P16 positivity, and to sialoblastoma notably in the pediatric population.

3. Pseudoendocrine sarcoma

3.1 Clinical aspects (including prognostication) and gross pathology

The mean age of occurrence is 52, with extremes being 29 and 74, thus more in the elderly age group than GLI1-altered tumors [24, 25]. There is a slight predilection for men.

The prevalent localization is the paravertebral region, other areas of the trunk being less frequent (chest wall, buttock, and groin) as well as head and neck sites.

This sarcoma of unknown differentiation is currently considered to be a tumor of intermediate malignancy with 43% risk of local recurrence and 21% risk of metastatic spread, knowing that these may be eventually underestimated. Frequent metastatic sites are lung and lymph nodes.

Tumors are deep-seated and their size varies from 2 to 19 cm, the median size being around 6 cm. They are grossly described as compact and hemorrhagic growths.

3.2 Light microscopy and ultrastructural features

The term “pseudoendocrine” refers to the histological architecture reminiscent of well-differentiated neuroendocrine carcinoma [24, 25, 26]. It is generally lobulated, nested, or trabecular like GLI1-altered tumor, packed with small round cells displaying stippled chromatin embedded in a delicately woven stroma (Figure 3). Mitoses can be found in great numbers (more than 5 per high power field). No necrosis has been thus far recorded. Electron microscopy undertaken in two tumors shows no characteristic neurosecretory granules in favor of neuroendocrine differentiation, but intercellular junctions with cytoplasmic cell processes [27]. Other occasional features include pseudoglandular aspects, calcifications, and stromal hyaline globules. Tumors have generally poorly circumscribed margins.

Figure 3.

Pseudoendocrine sarcoma: Endocrine-like morphology in light microscopy (courtesy of Dr. D Papke, Brighamn and Women’s hospital, Harvard Medical School, Boston USA).

3.3 Immunohistochemical aspects

Tumors show diffuse nuclear B-catenin positivity in 95% of cases (Figure 4) and S100 protein reactivity (Figure 5), at least focal, in 40% of cases. Desmin is reported positive in slightly less than 30% of cases and CD34 in 25% of tumors. Expression of keratin and/or neuroendocrine markers is not characteristic features [24, 25, 26].

Figure 4.

Diffuse B-catenin immunohistochemical expression in pseudoendocrine sarcoma (courtesy of Dr. D Papke, Brighman and Women’s hospital, Harvard Medical School, Boston, USA).

Figure 5.

Diffuse S-100 protein immunohistochemical expression in pseudoendocrine sarcoma (courtesy of Dr. D Papke, Brighman and Women’s hospital, Harvard Medical School, Boston, USA).

3.4 Molecular genetics

Recurrent B-catenin gene mutations identified by DNA sequencing methods are the hallmark of this tumor such as D32H, S33C, S331f, S37C, and S37F [24]. Their range does not generally overlap with that of desmoid fibromatosis but exceptionally the latter may harbor the D32H mutation. No other recurrent abnormality is found with RNA seq and no wild-type variant has been thus far discussed in the literature.

3.5 Differential diagnoses

The differential diagnoses are virtually the same as for GLI1-altered tumors (cf. supra par.2.6). It should nonetheless be emphasized that neither speckled chromatin nor B-catenin mutation is a known feature of any other known round cell sarcoma.

We add in this category of differential diagnoses pancreatic and ovarian solid pseudo-papillary tumors, which are small round cell non-sarcomatous tumors with characteristic b-catenin nuclear expression as well [28].

4. Conclusion

GLI1-altered mesenchymal round epithelioid cell tumors are ubiquitous and can be malignant generally speaking or more rarely of limited malignant potential irrespective of the oncogenic molecular abnormality involved, fusion, or amplification. Fusion tumors are more frequent than amplified ones. Immunohistochemistry with anti-GLI1 antibody is the simplest method of choice to distinguish these tumors from their potential mimics.

Pseudoendocrine sarcoma is a descriptive term with no common grounds with endocrine differentiation, considered to be a rare sarcoma of intermediate grade most frequently located in the paravertebral region. Its hallmark is a recurrent B-catenin mutation and immunohistochemical nuclear expression. Literature is still scarce with only around 24 cases reported.