Melanoma of the Scalp Harboring a Novel AGFG1 :: ERBB4 Fusion: A Case Report With Clinicopathologic and Genomic Characterization

ABSTRACT

ERBB4 (HER4) gain‐of‐function missense mutations have been implicated in cutaneous melanoma. In contrast, structural rearrangements involving ERBB4 are exceedingly rare. We describe a 59‐year‐old man with a rapidly enlarging 4.2‐cm melanoma of the scalp with high‐risk pathologic features, including epithelioid to plasmacytoid cytomorphology, ulceration, brisk mitotic activity, and microscopic satellitosis. RNA sequencing identified a novel in‐frame AGFG1::ERBB4 fusion joining AGFG1 exon 2 to ERBB4 exon 2. The ERBB4 portion retains the downstream coding sequence, including the tyrosine kinase domain. Concurrent molecular alterations included a pathogenic NF1 loss‐of‐function alteration, a TERT promoter mutation, CDK4 and STAT6 copy‐number gains, and TP53 copy‐number loss in a BRAF/NRAS/KIT‐wild‐type background.

1. Introduction

Cutaneous melanoma is highly genomically heterogeneous and has one of the highest somatic mutational burdens among human cancers [1]. Current molecular classification of melanoma recognizes several categories defined by their distinct genomic landscapes: high‐CSD (chronically sun‐damaged) melanoma, low‐CSD melanoma, acral melanoma, mucosal melanoma, uveal melanoma, and melanoma arising in (or resembling) blue nevus—blue‐nevus‐like melanoma [2], together with Spitz melanoma and the newer category of melanoma arising from Wnt‐activated deep penetrating/plexiform melanocytoma or nevus [3]. Both lentigo maligna melanoma (LMM) and desmoplastic melanoma are categorized within the high‐CSD melanoma group.

Although targeted therapies have transformed outcomes for patients with BRAF‐mutant melanoma, effective molecularly guided treatment options for BRAF‐wild‐type disease remain comparatively limited [4]. Many BRAF‐wild‐type melanomas harbor activating NRAS mutations or loss‐of‐function NF1 alterations that converge on RAS‐MAPK pathway activation; however, MEK inhibitor monotherapy and other strategies targeting downstream RAS‐MAPK signaling have shown limited clinical benefit in these settings [5]. These challenges highlight the need to identify additional actionable molecular alterations that may refine classification and therapeutic stratification of BRAF‐wild‐type melanoma. Gene fusions are rare in melanoma (approximately 2%) but may represent actionable therapeutic targets, supporting their evaluation in patients with advanced diseases who have progressed following immunotherapy [6].

ERBB4, also known as HER4, is a receptor tyrosine kinase within the ERBB/HER family, which also includes EGFR/ERBB1, HER2/ERBB2, and HER3/ERBB3 [7]. Activating ERBB4 point mutations have been described in melanoma, are enriched in BRAF‐wild‐type tumors, and have been shown to increase kinase activity, promote transformation, and confer sensitivity to ERBB inhibition [8, 9, 10]. These alterations may promote tumorigenesis by activating downstream RAS‐MAPK and PI3K‐AKT signaling pathways. However, compared with point mutations, ERBB4 structural rearrangements are rarely reported in melanoma, and their clinicopathologic significance remains poorly understood.

Here, we describe a novel AGFG1::ERBB4 fusion identified in a scalp melanoma within a BRAF/NRAS/KIT–wild‐type and NF1‐mutant genomic background.

2. Case Presentation

A 59‐year‐old man presented with a rapidly enlarging, non‐tender lesion on the right parietal scalp. He initially noticed the lesion after shaving and attributed it to a small cut. The mass rapidly enlarged over the months prior to evaluation. Physical examination revealed an approximately 4.2 cm × 4.2 cm exophytic, thick, friable plaque with peripheral hemorrhagic crust and scant purulent drainage. No cervical, axillary, or inguinal lymphadenopathy was identified.

Computed tomography of the head demonstrated an exophytic soft‐tissue mass arising from the right parietal scalp without invasion into subcutaneous fat or calvarium and without intracranial abnormality. Punch biopsy revealed an ulcerated malignant neoplasm composed of epithelioid to plasmacytoid cells arranged in sheets and nests with high nuclear‐to‐cytoplasmic ratios, marked pleomorphism, intranuclear pseudoinclusions, prominent nucleoli, and brisk mitotic activity. Immunohistochemistry demonstrated diffuse positivity for S100, SOX10, HMB45, and MART‐1 with negativity for CAM5.2, AE1/AE3, and p40, confirming the diagnosis of melanoma (Figure 1).

FIGURE 1.

Histologic and immunohistochemical features of the scalp melanoma. (A–D) The tumor is composed of epithelioid to plasmacytoid cells exhibiting marked cytologic atypia, prominent nucleoli, and brisk mitotic activity. No peripheral extension of melanoma in situ is identified beyond the invasive component (B). (E–H) Immunohistochemical stains demonstrate diffuse positivity for SOX10 (E), S100 (F), MART‐1 (G), and HMB45 (H). (A, E–H, 40×; B and C, 100×; D, 200×).

The patient underwent wide local excision with sentinel lymph node excision and sampling of right external jugular lymph nodes. Pathologic examination of the excision specimen demonstrated invasive melanoma with a Breslow thickness of 15 mm and Clark Level V invasion into subcutis. Ulceration was present with an extent of 20 mm, and the mitotic rate was 13 mitoses/mm². Lymphovascular and perineural invasion were not identified. Microscopic satellitosis was identified. Resection margins were negative, and 12 regional lymph nodes examined, including 10 sentinel nodes and 2 right external jugular lymph nodes, were negative for metastatic melanoma. The tumor was staged as pT4bN1c (Stage IIIC) according to the American Joint Committee on Cancer (AJCC) 8th edition melanoma staging criteria.

Given the aggressive clinical presentation and high‐risk pathologic features, comprehensive genomic profiling was performed using a clinically validated DNA‐ and RNA‐based next‐generation sequencing panel from Tempus (Chicago, IL, USA). RNA sequencing (xR) identified an in‐frame AGFG1::ERBB4 fusion joining the end of AGFG1 exon 2 to the start of ERBB4 exon 2, retaining the ERBB4 tyrosine kinase domain (Figure 2). On the GRCh37/hg19 reference, the breakpoints were chr2:228356356 in AGFG1 (NM_001135187.2, exon 2) and chr2:212989628 in ERBB4 (NM_005235.3, exon 2). The fusion junction sequence is 5′‐[AGFG1] ATGGAAAATGAA | TGTGTGCAGGAA [ERBB4]‐3′, which translates to MENECVQE in the AGFG1 reading frame, consistent with preservation of the open reading frame and retention of the downstream ERBB4 coding sequence. The Tempus xR assay is a whole‐transcriptome, hybrid‐capture next‐generation sequencing (NGS) assay that detects chromosomal rearrangements and altered splicing events using custom‐designed IDT probes. The assay uses the IDT xGen Exome Research Panel v2 as its backbone, which comprises more than 415 000 individually synthesized probes targeting approximately 34 Mb across 19 433 genes. The post‐microdissection tumor content in this case was 70%. The coordinates chr2:228356356 (AGFG1) and chr2:212989628 (ERBB4) were derived from the RNA fusion pipeline as the inferred genomic positions of the transcript junction and were not independently observed as a DNA structural variant. Additionally, orthogonal validation was not performed, and the functional significance of this fusion remains to be established.

FIGURE 2.

Schematic representation of the AGFG1::ERBB4 fusion, showing the fusion of AGFG1 exon 2 to ERBB4 exon 2. The breakpoint lies within the N‐terminal ArfGAP domain of AGFG1; therefore, only a partial ArfGAP domain is predicted to be retained, while the remainder of the ArfGAP domain and the downstream FG‐rich region are not retained. The ERBB4 contribution begins in exon 2, downstream of the native signal peptide, near the N‐terminal portion of extracellular domain I. The predicted fusion therefore retains near‐complete extracellular domain I, complete extracellular domains II–IV, the transmembrane region, the intracellular tyrosine kinase domain, and the C‐terminal signaling region.

DNA sequencing identified a tumor mutational burden of 15.3 mutations/Mb. Concurrent molecular alterations included an NF1 splice‐site variant, c.1641 + 2T>G, predicted to result in loss of function; a TERT promoter mutation, c.‐146C>T; CDK4 and STAT6 copy‐number gains; and TP53 copy‐number loss. The tumor was microsatellite stable and showed no pathogenic alterations in BRAF, NRAS, or KIT.

Although identification of the ERBB4 fusion suggested a potential role for ERBB‐directed targeted therapy, tyrosine kinase inhibitor treatment was not pursued because this approach remains investigational in melanoma. The patient received standard adjuvant anti‐PD‐1 immunotherapy with pembrolizumab and had no evidence of recurrence at 15 months of follow‐up.

3. Discussion

The present case harbored an NF1 loss‐of‐function alteration and a TERT promoter mutation, together with a high tumor mutational burden in a BRAF/NRAS/KIT–wild‐type background, which is consistent with a high‐CSD molecular profile [1]. This profile also corresponds to the NF1‐mutant molecular subtype in the TCGA four‐group framework, which denotes BRAF/NRAS–wild‐type tumors with high mutation burden arising in sun‐damaged head‐and‐neck skin of older patients [11].

Structural rearrangements involving ERBB4 may activate receptor tyrosine kinase signaling across multiple tumor types, with the location of the fusion breakpoint determining the predicted structure of the chimeric protein and its mechanism of activation. One pattern comprises kinase‐only fusions, in which a 5′ partner is joined to the C‐terminal portion of ERBB4 containing the tyrosine kinase domain; examples include EZR::ERBB4 in lung adenocarcinoma [12] and PARD3B::ERBB4 reported in HPV‐associated adenosquamous carcinoma of the oropharynx [13]. A second pattern comprises full‐receptor (extracellular‐domain–retaining) fusions, in which breakpoints near the 5′ end of ERBB4 preserve the extracellular, transmembrane, and intracellular tyrosine kinase domains; IKZF2::ERBB4, a recurrent event in peripheral T‐cell lymphoma, has been proposed as an example, although its precise breakpoint and domain architecture warrant confirmation [14].

The novel AGFG1::ERBB4 fusion described here follows the receptor‐preserving pattern. This junction places the N‐terminal AGFG1 segment upstream of a largely intact ERBB4 receptor. AGFG1 and ERBB4 are both located on the long arm of chromosome 2 but in opposite transcriptional orientations, with AGFG1 on the plus strand and ERBB4 on the minus strand. Because the two genes are antiparallel, a simple interstitial deletion would not co‐orient the partners into a sense transcript; the observed in‐frame AGFG1::ERBB4 fusion therefore most plausibly arises from an inversion or a more complex intrachromosomal rearrangement, and the exact genomic mechanism cannot be determined from RNA fusion data alone [15]. AGFG1 encodes a phenylalanine‐glycine (FG) repeat‐containing ArfGAP family protein with reported roles in nucleocytoplasmic transport and vesicle/RNA trafficking [16, 17]. In the present case, the breakpoint fell within the N‐terminal ArfGAP domain. Therefore, only part of the ArfGAP domain was predicted to be retained, whereas the remainder of that domain and the entire downstream FG‐rich region were lost. Because the ERBB4 contribution begins at exon 2, downstream of the native signal peptide, translation was predicted to initiate at the AGFG1 start codon. We compared the present case with the only ERBB4 fusion identified in the TCGA cutaneous melanoma cohort, CPSF4L::ERBB4 in TCGA‐ER‐A194 [11]. Despite different 5′ partners, both share an ERBB4 exon 2 breakpoint. We hypothesize that the retained AGFG1 moiety may contribute to aberrant, ligand‐independent ERBB4 activation through oligomerization or disruption of normal receptor autoinhibition. However, AGFG1 is not an established dimerization partner; therefore, this remains speculative and requires functional validation.

Elenius and colleagues have demonstrated that the oncogenicity of ERBB4 mutant alleles is isoform‐specific [10]. In this tumor, the ERBB4 breakpoint in exon 2 lies upstream of both major alternatively spliced regions that generate the JM‐a/JM‐b and CYT‐1/CYT‐2 isoforms [7]. Therefore, the breakpoint itself does not define the downstream ERBB4 splice isoform. Although the fusion was detected by whole transcriptome RNA sequencing, the available clinical report does not provide sufficient information to resolve the complete JM/CYT splice configuration. Because the biologic effects of AGFG1::ERBB4 may be isoform‐dependent, this limitation warrants further study.

Although canonical Spitz‐associated kinase fusions involve ROS1, ALK, NTRK1–3, MET, RET, BRAF, and MAP3K8, ERBB4 has been documented as a rare, noncanonical fusion partner within the Spitz spectrum [18, 19]. Quan et al. identified a previously unreported MYO5A::ERBB4 fusion in a Spitz tumor, expanding the recognized repertoire of kinase fusions in these lesions [20]. However, the presence of an ERBB4 fusion is not sufficient to assign Spitz lineage, as fusion partner, morphology, and the broader molecular background must be interpreted together. Kinase gene fusions were historically regarded as markers of Spitz lineage, but they are now well documented in bona fide conventional melanomas of non‐Spitz lineage, including aggressive fusion‐driven tumors lacking spitzoid cytomorphology [6, 21, 22, 23, 24]. The present tumor harbored a pathogenic NF1 loss‐of‐function alteration, characteristic of conventional, non‐Spitz melanoma rather than of Spitz neoplasia. Taken together, the morphological and molecular findings support a diagnosis of conventional non‐Spitz melanoma and argue against a Spitz lineage.

In the present case, the co‐occurrence of ERBB4 fusion with NF1 loss is particularly notable because recent studies have linked NF1‐mutant melanoma to ERBB‐family receptor activation [25]. NF1 loss increases RAS‐GTP levels and promotes constitutive RAS–MAPK pathway activation [5]. Beyond this canonical MAPK effect, NF1 loss has been shown to increase EGFR/ERBB1 expression and activation in melanoma, resulting in activation of both the RAS‐MAPK and PI3K‐AKT signaling pathways and sensitivity to EGFR inhibition in NF1‐mutant melanoma models [25]. Because ERBB4 can activate both pathways, AGFG1::ERBB4 may provide an additional upstream ERBB‐family signaling input, whereas NF1 loss drives downstream RAS‐MAPK pathway dependence and may promote broader ERBB‐family signaling vulnerability. This molecular context supports the possibility that AGFG1::ERBB4 functions as a cooperating event in NF1‐mutant melanoma.

Gene fusions are rare in melanoma but may represent clinically actionable genomic alterations [6]. A distinctive biologic feature of ERBB4 is that its intracellular domain can undergo proteolytic cleavage and translocate to the nucleus, where it may influence gene transcription [26]. The retention of the ERBB4 kinase domain raises the possibility of both broader ERBB‐family signaling and altered transcriptional regulation, providing a structural rationale for potential sensitivity to pan‐ERBB inhibition, although these possibilities require validation. Preclinical data suggest that NF1‐mutant melanomas do not respond consistently to single‐agent MEK, ERK, or PI3K‐AKT signaling inhibition [27]. MEK inhibitor monotherapy may also be limited by adaptive ERBB‐family signaling, including upregulation of ERBB3/ERBB2 signaling in BRAF/NRAS‐wild‐type melanoma [28]. Consistent with this, ERBB‐activated driver‐negative melanoma models show suppression of both RAS‐MAPK and PI3K–AKT signaling with afatinib, and combined afatinib plus trametinib produces stronger antiproliferative effects than either agent alone [29]. Thus, a pan‐ERBB inhibitor could theoretically suppress both fusion‐associated ERBB4 signaling and NF1‐loss‐associated ERBB‐family activation, whereas a MEK inhibitor could target downstream RAS‐MAPK pathway dependence. This molecular configuration provides a conceptual rationale for combined pan‐ERBB and MEK inhibition, supporting further investigation of this combination strategy.

4. Conclusion

In summary, to our knowledge, this is the first reported AGFG1::ERBB4 fusion in cutaneous melanoma. This case illustrates the value of integrating histopathologic analysis with DNA‐ and RNA‐based molecular profiling to identify rare but potentially actionable gene fusions in melanoma. Additional cases and functional studies will be needed to determine whether receptor‐preserving ERBB4 fusions confer sensitivity to pan‐ERBB inhibition.

Author Contributions

X.C. contributed to study conception and design, clinicopathologic and genomic data review, literature review, figure/table preparation, and manuscript drafting. H.Y, K.L.M., L.L., Y.P., and Y.T.H. contributed to pathologic review, clinicopathologic interpretation and molecular/genomic interpretation. R.M.J. contributed to literature review and data organization. C.C.Y. analyzed the genomic data and critically reviewed the manuscript. S.Z. supervised the study and contributed to study conception and design. All authors critically revised the manuscript and approved the final manuscript.

Funding

The authors have nothing to report.

Ethics Statement

The study was performed in accordance with the Declaration of Helsinki. The case was reviewed under Baylor College of Medicine IRB protocol H‐58097, “The Impact of Genomic Profiling and Biomarker Testing on the Diagnosis and Management of Malignancies and/or Rare Diseases.”

Consent

Informed consent was waived due to the retrospective nature of the study and the use of de‐identified archival material.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request.