PDGFB:APOBEC3C fusion in congenital diffuse high-grade glioma of the brainstem

To the Editor,

Infant-type high-grade gliomas are rare neoplasms that have superior clinical outcomes and differing molecular characteristics from both childhood and adult high-grade gliomas (1–3). Recently, pediatric high-grade gliomas have been reclassified in the 2021 WHO Classification into 4 subcategories: diffuse midline glioma H3 K27-altered, diffuse hemispheric glioma H3 G34 mutant, diffuse pediatric-type high-grade glioma H3-wild-type/IDH-wild-type, and infant-type hemispheric glioma (4). Infant-type hemispheric gliomas have distinct molecular profiles with fusion genes involving ALK, NTRK1/2/3, ROS1, and MET (5, 6). However, the biology of infant-type high-grade gliomas located in the brainstem is largely unknown and molecular characterization is lacking. Here, we report a case of congenital high-grade brainstem glioma with a previously unreported PDGFB:APOBEC3C fusion.

Our female patient highlighted in this report initially presented at 2 months of age with failure to thrive and was found to have aspiration on a fiberoptic endoscopic evaluation of feeding. At that time, a nasogastric tube was placed, and she was discharged with a diagnosis of protein-calorie malnutrition. At 3 months of age, she presented with seizure-like activity and acute hypoxic respiratory failure in the setting of an upper respiratory viral infection. A CT head revealed a 2.8 cm × 2.9 cm × 2.8 cm posterior fossa mass with obstructive hydrocephalus. Subsequent magnetic resonance imaging revealed an expansile, T2 hyperintense, contrast-enhancing brainstem lesion centered at the medulla with extension into the upper cervical spine and pons (Fig. 1A). She underwent posterior fossa craniotomy for biopsy, which pathology revealed an H3 wild-type glioblastoma, WHO Grade 4. Histologically, the tumor was characterized as a high-grade glial neoplasm with primarily monomorphic cells with basophilic to clear cytoplasm, focally surrounded by a myxoid matrix. There was a small focus of necrosis visualized. Immunohistochemical (IHC) analysis showed that the tumor was strongly and diffusely positive for GFAP, SOX10, and Olig2, with moderate staining for synaptophysin. The tumor was negative for LIN28, EMA, and H3K27M. INI-1 was intact, p53 was wild-type, and H3K27me3 was retained (Fig. 1B). MIB-1 was 7%-10%. Her tumor was negative for a KIAA1549-BRAF fusion, IDH1, IDH2, and H3F3A by next-generation sequencing.

Figure 1.

Clinical characterization of congenital diffuse high-grade glioma. (A) T1-weighted contrast-enhanced magnetic resonance imaging showing a diffuse brainstem lesion. (B) H&E section at 20× magnification (top) shows sheet-like to trabecular growth of a monotonous, hypercellular tumor in a myxoid and highly vascular background. Tumor cells are characterized by scant, somewhat cleared cytoplasm, and slightly angled nuclear contours. Rare islands of necrosis are seen with pyknotic and karyorrhectic debris and early calcification. Tumor cells vaguely palisade around the necrosis. Immunohistochemistry for Olig2 (middle) shows diffusely positive staining. H3K27me3 staining by immunohistochemistry is retained (bottom). (C) T1-weighted contrast-enhanced magnetic resonance imaging 2.5 years after completion of chemotherapy and focal radiation.

DNA methylation profiling demonstrated a focal amplification at chromosome 22 (Fig. 2A). Transcriptome sequencing (RNA-seq) of her tumor identified a PDGFB:APOBEC3C fusion at this locus that has not previously been reported (Fig. 2B, C) (7). Given the co-occurrence of focal amplification with an identified fusion transcript, we performed fluorescent in situ hybridization (FISH) using a PDGFB break-apart probe (Zytovision, Bremerhaven, Germany). This revealed marked 3′ amplification in a dispersed pattern suggestive of double minutes (Fig. 2D). To interrogate the functional activity of this fusion, we performed single-sample gene set enrichment analysis (ssGSEA) (8, 9) of canonical PDGF-signaling gene sets using the patient RNA-seq dataset, as well as several infant-type diffuse gliomas (3) and normal brain samples from our institutional repository as controls. This showed an enrichment of PDGF-signaling pathway genes relative to neoplastic or normal controls (Fig. 2E). IHC for PDGFRA and pERK was also strongly positive (Fig. 2F), together supporting the evidence for upregulated PDGF signaling within this tumor sample.

Figure 2.

Identification of a PDGFB:APOBEC3C fusion. (A) DNA methylation reveals focal amplification at chromosome 22 (insert) without other copy number variations. (B) Circos plot visualizing RNA-seq-based detection of fusion event on chromosome 22. (C) Fusion transcript mapped to 22q13.1 containing PDGFB PDGF/VEGF domain and APOBEC3C N-terminal domain. (D) Interphase FISH using PDGFB break-apart probe. 3′ end is labeled in red and 5′ with green; a normal arrangement is indicated by gold or fused red/green signal. (E) ssGSEA enrichment scores for patient sample (n = 1, red), infant-type diffuse gliomas (n = 4, black), or normal brain samples (n = 8, grey) using canonical PDGF signaling gene sets. (F) Immunohistochemistry for PDGFRA (left) and pERK (right).

The patient had a tracheostomy placed for long-term airway support. She was treated with a chemotherapy regimen consisting of 2 cycles of carboplatin and etoposide (10) followed by 2 cycles of carboplatin, etoposide, and imatinib before experiencing a progression on surveillance imaging. She then received palliative radiation to a total dose of 35 Gy in 10 fractions, using volumetric modulated arc therapy (VMAT) targeting the tumor with a 1 cm expansion. Despite noncurative intent, her tumor displayed marked radiographic regression (Fig. 1C). Her course was complicated by radiation necrosis, which resolved following treatment with dexamethasone and bevacizumab. Remarkably, almost 3 years have now elapsed from her last therapy without evidence of recurrent or progressive disease.

Without more robust preclinical modeling, we cannot conclusively assert that this fusion is sufficient for pathogenesis. However, the PDGF/VEGF domain is in-frame with evidence of functional activity, and comprehensive molecular characterization failed to identify another putative oncogenic event. APOBEC3C upregulation was identified as a high-risk feature in adult-type glioblastoma multiforme, but it has not been previously implicated in the biology of infant-type high-grade glioma (11). It has also been linked as a marker or poor prognosis in adult breast cancer (12). The platelet-derived growth factor (PDGF) family has been implicated in the carcinogenesis of high-grade glioma, including PDGFRA alterations in diffuse midline glioma, H3 K27-altered tumors of the brainstem (formerly diffuse intrinsic pontine glioma, [DIPG]) (13). There is evidence that PDGFB signaling induces cancer stroma in HGG, and that PDGFB induces a pro-inflammatory state in murine models of HGG (14). Together, these findings support the plausibility of the novel fusion described here as the inciting oncogenic event for our patient.

The 2021 WHO revised classification of pediatric diffuse gliomas adds needed pathologic clarity in the era of molecular diagnosis, but it may still fall short in categorizing rare patient populations. Congenital high-grade gliomas are rare entities that differ clinically and molecularly from their adult and childhood counterparts. Here, we discuss a previously unreported fusion of PDGFB:APOBEC3C in a patient with congenital brainstem diffuse high-grade glioma with a favorable clinical course. This highlights the importance of routine molecular characterization, both to better understand the complex biology of this rare disease and to guide prognosis and clinical decision-making for individual patients and families.

Funding

This work was directly supported by the Morgan Adams Foundation; the National Institute of Neurological Disorders and Stroke (1K08NS121592) to N.A.D. The University of Colorado Genomics Shared Resource and Pathology Shared Resource are supported by a Cancer Center Support Grant (P30 CA046934).

Conflict of interest

The authors have no duality or conflicts of interest to declare.