Introduction

The importance of genetic analysis of pediatric brain tumors has dramatically increased in recent years. First, diagnosis based on the revised 2016 WHO classification of tumors of the central nervous system requires the analysis of certain missense mutations, fusion genes, and amplifications/deletions of certain chromosome regions [1, 2]. Second, for medulloblastomas, a molecular classification is used both as a prognostic marker and for stratification into risk groups [3]. Third, some genetic abnormalities, such as the BRAF V600E mutation, and the NTRK, MET, and ALK/ROS1 fusion genes, can be effective therapeutic targets [4,5,6,7]. In addition, methylation-based classification has become a powerful tool [8]. For example, posterior fossa ependymoma is molecularly classified into the PFA and PFB subgroups, which have distinct clinical features [9,10,11]. Strum et al. reported that CNS–PNET could be reclassified into several novel subgroups [12]. Further, in 2018, the German Cancer Research Center (DKFZ) demonstrated the utility and accuracy of methylation-based classification of all brain tumor entities, and this classification tool is now available on their website [13].

Here, we report a case of pediatric midline low-grade glioma (LGG) in the lateral ventricle harboring both the H3F3A K27M and BRAF V600E mutations. Based on the methylation analysis, the tumor was classified as a “diffuse midline glioma with an H3 K27M mutant.” However, the pathological and clinical features were distinct from those of this entity.

Clinical summary

A 2-year-old female was referred to our hospital because of an intraventricular mass. The patient was born at 33 weeks gestation and she stayed in the hospital for 1 month because of her low birth weight. At the age of 1 year, she presented with epilepsy, which was controlled with an antiepileptic drug. MRI showed a nodular mass in the left ventricle (Fig. 1a–d). The patient underwent gross total resection of the mass because it was slowly growing. Based on the intraoperative findings, the border between the tumor and the normal thalamic tissues was partially unclear, whereas that between the tumor and ventricular lumen was clear. The local pathological diagnosis based on the 2007 WHO classification was oligoastrocytoma. She received no chemotherapy or radiotherapy and has survived without the evidence of disease for more than 8.5 years. After the publication of the revised fourth edition of the WHO classification in 2016, a molecular analysis and a review of the central pathology were performed (Fig. 2, 3a–f).

Fig. 1
figure 1

Radiological imaging of the case. a–d T2-weighted (a, b) and contrast-enhanced T1-weighted MRI (c, d) demonstrating a circumscribed, partially enhanced mass in the left ventricle

Full size image
Fig. 2
figure 2

Pyrosequencing results showing the BRAF V600E and H3F3A K27M mutations

Full size image

Pathological findings

The tumor consisted of moderately cellular glial proliferation within a fibrillary background. The lesional cells were relatively uniform in appearance and were round, oval, and spindle shaped. Pleomorphic, giant, epithelioid, rhabdoid, and piloid cells were lacking. The round cell component was associated with a perinuclear halo and resembled oligodendroglial tumors, whereas the spindle cells resembled the astrocytic lineage (Fig. 3a, b). There were a small number of rosette-like glial structures (Fig. 3c). Dysmorphic ganglion cells were absent. Mitotic figures were not detected. Necrosis, microvascular proliferation, Rosenthal fibers, and eosinophilic granular bodies were not observed. Focal microcalcification was present. Immunohistochemically, the tumor cells were positive for glial fibrillary acidic protein and S100 protein but were negative for epithelial membrane antigen and synaptophysin. In agreement with the molecular findings, they were immunopositive for H3 K27M, with a concomitant loss of H3K27me3 expression (Fig. 3d, e). BRAF V600E immunohistochemistry was positive (Fig. 3f), and H3 K27M and BRAF V600E staining appeared to be present in the same cells. The tumor retained ATRX expression, and p53 staining showed wild-type labeling. The MIB1 labeling index was 6.8%. The tumor mostly lacked transgressing neurofilament–positive fibers, and there was no evidence of diffuse infiltration.

Fig. 3
figure 3

Histological and immunohistochemical findings. The tumor consisted of glial proliferation with oligodendroglial (a) and astrocytic (b) morphology. There were a small number of rosette-like glial structures (c). Immunohistochemically, the tumor cells were positive for H3 K27M (d), with a concomitant loss of H3K27me3 expression (e). BRAF (V600E) immunohistochemistry was also diffusely positive (f)

Full size image

Molecular analysis

DNA and RNA were extracted from frozen tumor tissue using the DNeasy Blood and Tissue Kit (Qiagen, Tokyo, Japan) and the miRNeasy Mini Kit (Qiagen, Tokyo, Japan), respectively. Hot spot mutations, including IDH1 R132, IDH2 R172, BRAF T599, BRAF V600, H3F3A K27, H3F3A G34, TERT promoter C250, TERT promoter C228, FGFR1 N546, and FGFR1 K656, were analyzed by pyrosequencing using the AQ assay with a PyroMark Q96 (Qiagen, Tokyo, Japan). The polymerase chain reaction (PCR) for pyrosequencing and the pyrosequencing assay were performed as previously described [14]. Primer sequences are listed in supplementary Table 1. The BRAF V600E and H3F3A K27M mutations were detected at frequencies of 40% and 46%, respectively (Fig. 2). The hot spot mutation in other analyzed genes was not detected. Multiplex ligation-dependent probe amplification (MLPA) using the SALSA MLPA probemix P088-C2 (MRC-Holland, Amsterdam, the Netherlands) revealed no CDKN2A deletion. The existence of KIAA1549-BRAF fusion was assessed by reverse transcriptase PCR using previously reported primers and it was not detected [15].

Table 1 Reported cases of pediatric intracranial tumors harboring the H3F3A K27M and BRAF V600E double mutations
Full size table

A methylation classifier assay was then performed. DNA methylation was analyzed using an Infinium HumanMethylation450 BeadChip array (Illumina, San Diego, CA, USA), and the IDAT-files from the sample were uploaded to the online classifier developed by the DKFZ (https://www.molecularneuropathology.org/mnp). The tumor was classified as a “diffuse midline glioma, H3 K27M mutant” with a calibrated score of 0.98.

Discussion

Here, we report a case of midline glioma harboring concurrent H3F3A K27M and BRAF V600E mutations. The tumor was classified as a “diffuse midline glioma, H3 K27M mutant, WHO grade IV” by methylation analysis. However, in several aspects, the case was not entirely compatible with typical diffuse midline gliomas with H3F3A K27M mutations. Importantly, the present tumor was likely localized rather than diffusely infiltrating. Radiologically, it seemed well circumscribed; that is most of the tumor surface being surrounded by cerebral fluid. However, the border between tumor and normal thalamic tissue was partly unclear. Histologically, there was a paucity of transgressing neurofilament, suggesting a non-diffuse process, although this finding needs to be interpreted with caution because it may simply reflect the intra-ventricular growth. “Diffuse midline glioma with H3 K27M mutant WHO grade IV” was introduced as a new entity of WHO classification 2016, based on the knowledge at that time that recurrent mutation at K27 in H3F3A, HIST1H3B, and HIST1H3C is detected in high-grade glioma (HGG) from the pons, thalamus and spinal cord and that these mutations occur exclusively in diffuse midline gliomas, as stated in WHO blue book [2, 16]. However, some cases of non-diffuse glioma with H3F3A K27M were subsequently reported [17, 18]. Accordingly, Consortium to Inform Molecular and Practical Approaches to CNS Tumor Taxonomy—Not Official WHO (cIMPACT-NOW) up date 2 emphasizes that “diffuse,” “midline,” “glioma” and “H3 K27M-mutant” are requirements for the diagnosis of diffuse midline glioma with H3 K27M [16]. Therefore, based on the cIMPACT-NOW-modified view of the WHO Classification, the present case seems incompatible with the “diffuse mid-line glioma with H3 K27M” designation. Whereas some of the reported localized gliomas with H3 K27M demonstrated the histology of well-established entities, the present case was difficult to classify histologically. The original diagnosis of oligoastrocytoma is likely invalid due to its mostly localized nature. Although histological findings, along with the presence of the BRAF mutation, may suggest a possibility of pilocytic astrocytoma, characteristic features of that entity were mostly lacking, including piloid or microcystic loose tissues, Rosenthal fibers, eosinophilic granular bodies, and microvascular proliferation. The morphology did not fit well with other BRAF-mutant localized gliomas either, such as gangliogioma and pleomorphic xanthoastrocytoma. Taken together, the present case would be best labeled descriptively as low-grade glioma, likely localized, not elsewhere classified as per the Consortium to Inform Molecular and Practical Approaches to CNS Tumor Taxonomy—Not Official WHO cIMPACT-NOW update 1 [16, 19].

Classification based on methylation profiles is a highly robust method for facilitating diagnosis and subgrouping brain tumors. However, the interpretation of methylation-based classification may sometimes require caution. H3 K27M would cause a considerable change in the tumor’s epigenetic landscape, which may mask fine differences among K27M-harboring tumors, some of which are not compatible with the definition of diffuse midline glioma with H3 K27M mutation, as in our case. Other mutations, such as BRAF, may on the other hand less affect epigenetic landscape, which could make them more difficult to distinguish by methylation profiling alone. The principle of defining a tumor entity, whether using should histology, genetics, methylation, or their combination, warrants further discussion.

In addition to the location and histological findings presented above, the patient’s age at diagnosis and her relatively long survival of the present case are also unusual for H3F3A K27M–positive glioma. The H3F3A K27M mutation is characteristically detected in pediatric gliomas. The mean age at diagnosis is 6–10 years [6, 20,21,22]. Most H3F3A K27M -positive gliomas arise in the pons or thalamus, and only approximately 5% arise in the ventricle [6, 20]. In most cases, the histological grade is high. The median overall survival of H3F3A K27M-positive midline glioma is less than 1 year, which is significantly worse than that of H3F3A K27 wild-type glioma. Based on an analysis of 77 pediatric patients with diffuse midline glioma over the age of 3 years, Karremman et al. reported that H3 K27M status has a negative impact on survival, regardless of tumor location or pathological grade [20]. There is a case report of pediatric LGG in the thalamus harboring H3 K27M transformed to HGG [23]. Pratt et al. also reported that H3 K27M has a negative impact on survival in circumscribed/non-diffuse glioma [24].

In contrast to H3F3A K27M, BRAF V600E is detected in a variety of tumors arising in patients of all ages, including thyroid and colorectal cancers as well as Langerhans cell histiocytosis [3, 25,26,27,28,29,30,31]. It is also detected in various glioma subtypes arising in any location, and the frequency is high in some subtypes, such as pleomorphic xanthoastrocytoma. The prognostic implications of BRAF V600E are dependent on various factors, including patient age, patient gender, tumor type, and the presence of other concurrent molecular abnormalities [3, 25,26,27,28,29,30,31,32]. For pediatric LGG, several reports suggest that BRAF V600E is associated with an aggressive clinical course, although this is somewhat controversial [3, 28, 32, 33]. Lassaletta et al. reported that the progression-free survival of patients with BRAF V600E–positive LGG is significantly worse than that of patients with wild-type BRAF LGG [3]. Mistry et al. reported that the frequency of BRAF V600E in pediatric LGG that underwent malignant transformation was higher than the frequency in non-transformed LGG (44% vs. 6%) [31].

Although the coexistence of H3F3A K27M and BRAF V600E mutations is rare, there have been several reported cases of pediatric glioma harboring these concurrent mutations (Table 1) [20, 22, 32, 34,35,36,37,38]. Interestingly, these include both diffuse midline glioma and non-diffuse glioma. In view of the diagnostic and prognostic information of this alternation, the recently published cIMPACT-NOW update 4 proposed the new classification name “Diffuse glioma, BRAF V600E mutant.” The diagnosis of some cases shown in Table 1 may warrant further discussion [39]. It is also noteworthy that the prognosis of these patients was not necessarily dismal but instead appeared to be somewhat better than that of patients with midline glioma harboring only the H3F3A K27M mutation, and that no patients died within the first year after the initial diagnosis. Notably, four patients, including all three patients under the age of 3 years, survived for more than 5 years. However, the number of reported cases with concurrent mutations is small and histological diagnosis, grading and age of diagnosis are variable. The follow-up periods are too short and recent widespread application of BRAF-targeted therapy may change their survival [40]. Therefore, further analysis of more cases is required to clarify these points.

In summary, we report a case of a relatively long-term survivor of pediatric midline glioma harboring concurrent BRAF V600E and H3F3A K27M. Although the tumor was classified as “diffuse midline glioma with H3K27M mt, WHO grade IV” based on methylation analysis, the pathological and clinical features of this case did not fit those of “diffuse midline glioma with H3 K27M mt” as defined in the current WHO classification (2016). Our case points to certain limitation of the current methylation classifier and it calls for further discussion of the integrated diagnosis of brain tumor. Further investigations of non-diffuse glioma with H3F3A K27M and glioma with both BRAF V600E and H3F3A K27 M are warranted.