Rare IDH hotspot mutations in dysembryoplastic neuroepithelial tumors expand the spectrum of IDH-altered CNS tumors

Since the 2016 World Health Organization (WHO) classification, IDH mutations are considered an essential criterion for adult-type diffuse gliomas [6] and may be thought to be restricted to them. Adult-type IDH mutated gliomas are rare in the pediatric population, generally associated with more favorable outcomes than adults [5]. Rare dysembryoplastic neuroepithelial tumors (DNETs) with IDH1/2 mutations have been reported [1–3]. As these preceded the era of DNA methylation profiling, their initial diagnosis remains uncertain. Herein, we report three glioneuronal tumors harboring IDH1/2 hotspot mutations. All three patients were adolescents (aged 12, 13, and 10 years at surgery), with a history of epilepsy. Tumors were located in the posterior temporal lobe (case #1), insula (case #2), and frontal lobe (case #3). Neuroradiological review revealed well-delineated cortico-subcortical masses with low density on computerized tomodensitometry, low signal on T1-weighted images, and high signal on T2-weighted images with slightly lower signal on FLAIR images, without FLAIR/T2 mismatch or edema. Contrast enhancement was absent or minimal (slight partial enhancement in case #2). These features closely resembled typical DNETs and lacked an infiltrative pattern (Fig. 1, Supplementary material). Patient #3 required two complete resections, 7 years apart. Histopathologically, tumors #2 and the first resection of #3 were prototypical DNETs with the specific glioneuronal element. Tumor #1 displayed few neurons in a myxoid background whose residual nature or part of a specific glio-neuronal element was difficult to determine (Fig. 1). In all tumors there was no atypia, mitosis (with low proliferation index accordingly), necrosis, nor microvascular proliferation. In the recurrence of patient #3, histopathological features remained consistent with a DNET, although without obvious specific glioneuronal element and with the appearance of microvascular proliferation at the edge of a cystic formation. All tumors harbored IDH1/2 hotspot mutations. Tumors #1 and #2 had IDH1 p.R132H mutation with variant allele frequency (vaf) of 30% and 40%, respectively, confirmed by immunohistochemistry. Tumor #3 harbored an IDH2 p.R140W mutation with a vaf of 5%, which, although canonical, has not previously been reported in gliomas [6] (Supplementary Table 1). None of the tumors carried FGFR1 kinase domain duplications, point mutations, or FGFR1 fusions. For patient #2, TSO500 panel sequencing was performed from tumor and germline (blood). This confirmed the somatic nature of the IDH1 p.R132H mutation and in addition the presence of an unusual somatic PDGFRA indel (c.1139_1141delinsCGG; p.K380_L381delinsTV; vaf 73%) and LOH of chromosome 4 where PDGFRA is located. No PDGFRA mutation was found in the other tumors, although the gene was included in the other panels. Tumor #3 additionally had low vaf alterations of ALK and KIT detailed in Supplementary Table 1. DNA methylation profiling (Heidelberg classifier V12.8) classified all tumors as glioneuronal tumors with calibrated scores ranging from 0.73 to 0.90 for DNET methylation subclass. Also, they clustered with DNETs by t-distributed stochastic neighbor embedding (tSNE) analysis (Fig. 1, arrowheads). Copy number variation (CNV) generated from methylation data was flat, without 1p/19q co-deletion (Fig. 1). Of note, for patient #3, the detailed molecular analysis described above was performed on the recurrence due to lack of material from initial surgery. All patients remained alive and seizure free at 12, 192 and 240 months of follow-up, without adjuvant treatment.

Fig. 1.

MRI, histopathology, CNV, and t-SNE visualization of DNA methylation profiles from the three cases (black arrows). t-SNE parameters, reference dataset, and legend are available in supplementary material. MC: methylation class. Scale bars: 100 μm

We identified four additional reported DNETs with IDH1 p.R132H mutation in patients aged 3–24 years with a history of epilepsy (Supplementary Table 1). All four tumors lacked DNA methylation confirmation. Three of the tumors were reported in a 2011 study [3]; they showed IDH1 p.R132H mutation by Sanger sequencing, without loss of heterozygosity of 1p/19q. Two patients remained seizure free during long-term follow-up (27–108 months); the third was lost to follow-up. The fourth reported tumor [1], histopathologically represented a complex DNET with LHFPL3::NTRK2 fusion and IDH1 p.R132H mutation. Follow-up information was not available. While the interpretation of previously reported tumors is limited by absent methylation data, our study demonstrates that IDH1/2 hotspot mutations—either alone or with other alterations—may occur in pediatric cortical tumors with a benign glioneuronal phenotype and a DNA methylation profile consistent with DNET rather than IDH-mutant glioma. These findings raise the question of whether the spectrum of IDH-altered tumors needs to be expanded, either by introducing a new type of low-grade IDH-mutated tumor or by defining a subtype of DNET with IDH1/2 hotspot mutation. Radiological, histological, and DNA methylation analysis, as well as benign clinical course would favor considering our three tumors as DNET. None of the tumors carried FGFR1 alterations typically seen in DNETs, suggesting a different molecular pathogenesis from classical DNET. Conversely, they did not carry ATRX or TP53 mutations, common in IDH-mutant astrocytomas, indicating that they are also genetically distinct from astrocytomas, IDH mutant.

In the current WHO Classification, IDH mutations are an entity defining alteration for IDH mutant gliomas. Our highly molecularly characterized tumors represent a warning that IDH1 and IDH2 hotspot mutations may also be observed in other brain tumors. The situation parallels the K27 histone mutation, initially regarded as specific for diffuse midline glioma, but later identified in other tumor types [4]. Our limited data suggest that the clinical course of IDH mutant DNET, including a progressive case with a follow-up of 20 years, may more closely resemble classical DNET, in which malignant transformation is exceptional [2], than gliomas, IDH-mutant, which almost invariably progress to malignant gliomas; however, caution is warranted given the limited follow-up for one patient.

In conclusion, we present molecular evidence for a previously unanticipated occurrence of IDH hotspot mutations in molecularly defined DNETs. This observation emphasizes the importance of additional diagnostic layers in the integrated diagnosis beyond mutational analysis, as these tumors can be identified as DNETs in particular through histology and DNA methylation profiling, and, when methylation is not available, characteristic MRI features may provide additional support for the diagnosis.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

A.M., D.C., P.V.: study conception and design. P.V., G.S., M.R., A.T.E., A.M., D.C.: histopathological review and interpretation. P.V., M.R., A.M., D.C.: figure conception. R.S., A.K., S.S.: molecular biology experiments (DNA extraction, sequencing); analysis of IDH1/2 mutations and additional variants. E.S.: DNA methylation profiling, t-SNE clustering, copy-number analysis; bioinformatic data interpretation. V.D.R.: neuroradiological review and interpretation of MRI data. J.O., S.B., T.B., M.B.: clinical data collection, patient follow-up, and integration of clinical outcomes. M.R.: literature review and comparative analysis with previously published cases. M.R., A.M., D.C.: Drafting and redaction of the manuscript; preparation of supplementary data and references. All authors: critical revision of the manuscript and approval of the final version.

Funding

D.C.’s work was supported by the Everest Centre for Research into Paediatric Low-Grade Brain Tumours (GN-000707, The Brain Tumour Charity, UK). The other authors declare that they have not received funding for this study.

Declarations

Conflict of interest

David Capper receives license fees from the sales of the IDH1 R132H mutation-specific antibody from DIANOVA GmbH/BIOZOL Diagnostics GmbH. David Capper further is a cofounder and a shareholder of Heidelberg Epignostix GmbH. The other authors declare that they have no conflicts of interest related to the topic of this article.

Ethical approval and consent to participate

Ethical approval for reporting the case from Charité Berlin was granted by the Charité Ethics Committee (EA2/053/25). Written informed consent or non-objection to participate in the study was provided by the patient’s legal guardian for patients from Hôpital Necker Enfants Malades.