ERBB2/ ERBB3-mutated S100/ SOX10-positive unclassified high-grade uterine sarcoma: first detailed description of a novel entity

During the last two decades, significant progress has been made in elucidating the genetic landscape of uterine mesenchymal neoplasms, with delineation of novel entities and characterization and expansion of the molecular drivers in the spectrum of established neoplasms. Notably, most of these recent advances concern identification of fusion genes [1]. For example, PLAG1 and PGR fusions have been recognized as genetic drivers in subsets of high-grade uterine sarcomas characterized by prominent myxoid and epithelioid/rhabdoid features, respectively, some of which were previously classified as leiomyosarcomas [2, 3]. On the other hands, ALK fusions have been confirmed in inflammatory myofibroblastic tumors [4]. In the spectrum of fibrosarcoma-like malignancies, a variety of recurrent tyrosine kinase fusions have allowed identification of distinct molecularly defined entities, including NTRK [5], COL1A1::PDGFB [5], and others. Uterine tumors resembling ovarian sex cord tumors (UTROSCT) represent another tumor with recently characterized recurrent fusion genes involving ESR1, GREB1, NCOA1-3, and others [6].

The molecular landscape of endometrial stromal neoplasms has emerged in both the low-grade (LG-) and high-grade (HG-) endometrial stromal sarcoma (ESS) categories [1]. Moreover, a new sarcoma category overlapping with ESS and harboring recurrent KAT6A/B::KANSL1 fusions has been recently identified [7]. Correctly identifying and precisely subtyping these emerging entities is mandatory for proper risk stratification, which is the basis for optimized treatment and follow-up strategies [8]. We herein describe the detailed clinicopathological and molecular findings of a novel uterine sarcoma characterized by a neurogenic-like (S100/SOX10-positive) immunophenotype and activating mutations involving members of the epidermal growth factor receptor (EGFR/ERBB) family of tyrosine kinases ERBB3 (HER3) and ERBB2 (HER2) to alert pathologists to this rare but possible underdiagnosed entity with therapeutic implications.

The cases were identified in our consultation files. The tissue specimens were fixed in formalin and processed routinely for histopathology. Due to the consultation nature of the cases, immunohistochemistry (IHC) was performed in different laboratories and the stains applied varied from case to case, based on tissue availability and initial differential diagnostic considerations (details of the staining protocols and antibody sources are available upon request).

With the wider application of next generation sequencing modalities in routine practice, the classification of uterine sarcomas has been dynamically evolving [1]. A variety of new entities have been delineated based on precise genotyping and most aggressive malignancies in the historical spectrum of undifferentiated uterine sarcomas could indeed be reclassified by molecular profiling, mostly into the spectrum of HGESS [13]. Most of these recent developments were represented by diverse recurrent oncogene fusions as driver events in distinct subsets of neoplasms of well established (endometrial stromal [1]) origin or of ambiguous (PLAG1, NTRK fusions and others [2, 3, 5]) histogenetic origin.

A series of 7 unclassified uterine neoplasms expressing S100 and SOX10, lacking gene fusions and harboring an activating ERBB2 (HER2) mutation at codon 777 were presented in an abstract form at the USCAP annual meeting (Los Angeles, 2023) [14]. Original diagnoses of the presented cases were mostly unclassified or undifferentiated sarcoma with discussion of metastases, MPNST or endometrial stroma sarcoma. Patient’s age range was 42 to 60 years (median, 53). Sites of primary involvement were cervix (n = 2), cervix/uterus/vagina (n = 2), cervix and uterus (n = 1), cervix and pelvis (n = 1) and uterus unspecified (n = 1). Hysterectomy was the treatment in 3 patients, three received biopsy only and one had removal of the mass with bilateral salpingo-oophorectomy. The tumor stage was stated as T1b (n = 1), T2 (n = 2), T3b (n = 1) and T4 (n = 1) [14]. All tumors were cellular with predominance of spindle cells and variable focal or multifocal epithelioid/ round cells with frequent necrosis (3 of 7), focal myxoid stromal changes (3 of 7) and brisk mitotic activity in the range of 1 – 30 mitoses/ 10 HPFs (median, 7). The genomic profiling of the 7 tumors revealed frequent truncating ATRX mutations (5 of 7) and homozygous CDKN2A (5 of 7) and CDKN2B (4 of 7) deletions. TP53 missense mutations (2 of 7) and HER2 co-amplification (3 of 7) were observed in subset of tumors [14].

The herein presented cases fit perfectly into this novel tumor category, both histologically, immunohistochemically and molecularly. We detected same ERBB2 point mutation (p.V777L) in Case 2 as described by Lin et al. [14] and a novel ERBB3 (HER3; EGFR3) point mutation (p.Glu928Gly) in Case 1. Remarkably, a ATRX mutation and deletion of CDKN2A (p16) together with HER2 co-amplification at a genomic level were detected in both tumors.

Being a member of the human EGFR family of tyrosine kinases, ERBB2/HER2-neu is an orphan receptor without a known activating ligand, however, it adopts a structure similar to the ligand-bound state and is active in a heterodimeric complex [15]. HER2 amplifications drive oncogenesis in a significant proportion of aggressive epithelial cancers (carcinomas) including in particular subsets of breast cancer [16] and salivary gland cancer [17] where it is detectable in 15 – 30% of cases.

Valine 777 of HER2 is located in the N-lobe (N-terminal lobe) of the tyrosine kinase domain and results in an elevated kinase activity, an aberrant and enhanced phosphorylation of downstream signaling molecules in cell culture experiments, as well as an increased and more rapid tumor growth compared to wild-type HER2 in xenograft models [18]. Mutation of Valine at position 777 represent one of the frequently detected HER2 alterations in cancer and is observed in approximately 4.0% of all HER2 mutated cases [19].

ERBB3 (Erb-B2 Receptor Tyrosine Kinase 3, synonym: HER3) belongs to the EGFR family of receptor tyrosine kinases. Despite the identification of neuregulin-1 (NRG1) as specific ligand, ERBB3 is lacking a significant intrinsic kinase activity and has been considered as kinase dead receptor. However, this paradigm has been continuously challenged and, in fact, the role of ERBB3 in pathogenesis of different malignancies has been demonstrated (as summarized in Black et al. [20]).

The c.2783A > G ERBB3 mutation replaces glutamic acid, a polar, negatively charged amino acid, by glycine with nonpolar and neutral properties at position 928 in the kinase domain. Glutamic acid at position 928 is implicated in the dimer formation interface of ERBB3 and, in fact, ERBB3 displays oncogenic potential when co-expressed with ERBB2 [21]. In this HER3/HER2 heterodimer, E928G increases the catalytic activity of HER2, an effect which was less pronounced when glutamine acid at position 928 was replaced with Alanine or Lysine, implicating varying functional effects of particular residues at this position [22]. Of note, in a heterodimeric complex, instead of reactivating catalytic activity of ERBB3, the E928G mutation rather increases the dimerization affinity of ERBB3 thereby enhancing its allosteric activation potential [23]. This highlights the necessity of including a functional, catalytical active dimerization partner in the heterodimer, for ERBB3 E928G to execute its full oncogenic potential. Our Case 1 falls into this pathogenetic category and represents a novel observation in this type of uterine sarcomas, indicating that mutant ERBB3 concurrent with ERBB2 amplification represents a novel mechanism driving oncogenesis in tumors lacking the V777L ERBB2 mutation.

The exact nosology of this neoplasm remains enigmatic. The lack of any detectable gene fusion and the strong and homogeneous expression of SOX10 and S100 exclude all known genetic subtypes of high-grade endometrial stromal sarcoma (HEESS) [1]. In this regard, the diffuse Cyclin D1 expression noted in both of our cases represents a diagnostic pitfall, given that this marker has been proposed as a surrogate for HGESS harboring YWHAE and BCOR gene fusions [24]. On the other hand, the expression of SOX10 and S100 makes the possibility of an MPNST and a metastatic melanoma important considerations. MPNST-like tumors have been reported in the female genital tract [25]. Their neurogenic nature was favored on the basis of patchy expression of S100 in addition to CD34 reactivity [25]. However, most of these tumors have been recently reclassified as genetically defined tyrosine kinase fusion associated sarcomas including in particular NTRK and COL1A1::PDGFB fusions [5]. Variable reactivity for S100 is a common feature in NTRK fusion sarcomas/neoplasms. However, these fusion tumors lack SOX10 expression, in line with a non-neurogenic and non-melanocytic origin. Indeed, coexpression of S100 (usually variable and patchy) and CD34 represents a valuable clue to tyrosine kinase fusion associated mesenchymal neoplasms originating at different anatomic sites [10]. In the context of non-epithelial and non-myoepithelial neoplasia, coexpression of S100/SOX10 is considered specific for a schwannian or melanocytic line of differentiation. In this regard, the detection of a NRAS mutation at position 61 in one of our cases represents a pitfall as this genotype is noted in 20—40% of melanomas, the frequency varying with the clinicopathological tumor type [26]. In our cases, the tumor did not cluster with MPNST or melanoma arguing against both possibilities. Moreover, the distinctive genotype with activating HER2 mutation, CDKN2A deletion and truncating ATRX mutation, which are identical to those reported in the cited abstract [14], all argue for a distinctive entity unrelated to melanoma, MPNST, or any of the established sarcoma types of the female genital tract or the soft tissues.

The literature on HER2 mutations in mesenchymal neoplasms is limited. Ronellenfitsch et al. reported activating ERBB2 mutations (p.Leu755Ser, p.Asp769Tyr, p.Val777Leu) in 3 of 7 (43%) patients with schwannomatosis, but in none of 8 NF2-asscoiated or sporadic hybrid schwannoma-neurofibroma cases [27]. One schwannomatosis patient had three tumors all harboring the same V777L (p.Val777Leu) ERBB2 mutation as reported in our cases. We have recently observed a similar mutation (p.Val777Leu) in a case of multiple hybrid schwannoma-neurofibromas unassociated with clinically recognizable NF1 or NF2 syndrome (Agaimy, unpublished data). These reported ERBB2 kinase domain mutations are known to occur in breast cancer and rarely in carcinomas of other organs and are treatable by pan-ERBB2 inhibitors [28‐31].

Finally, Lim et al. have recently reported a high-grade uterine sarcoma carrying a novel ERBB4 fusion (fused to CIQTNF1) in a 49-year-old woman [32]. The tumor morphology was suggestive of HGESS with variable expression of desmin, ER, PR, AE1/3 and cyclin D1 [32]. This report and our current cases point to an emerging role of members of the human EGFR family of tyrosine kinases (fusions and mutations) in the oncogenesis of rare uterine sarcomas.

In summary, we herein describe the first detailed study of a novel ERBB2/ ERBB3-mutated S100/SOX10-positive unclassified highly aggressive uterine sarcoma type. The histogenesis of this tumor, its appropriate classification and the potential benefit of targeting the underlying ERBB2/ ERBB3 tyrosine kinase mutation remain to be verified in the future. Inclusion of SOX10 in high-grade unclassified gynecological sarcomas would be a valuable and cheap screening tool to enhance recognition of this entity, particularly in putative cases of undifferentiated uterine sarcomas.