In this study we report two cases that fully meet the diagnostic criteria of DLGNTs as proposed by the sixth cIMPACT-NOW update [9], although neither of them presented diffuse growth nor leptomeningeal involvement in spite of a long follow-up. Indeed, both were glioneuronal neoplasm composed of oligodendrocyte-like cells; chromosome arm 1p deletion and a mitogen-activated protein kinase (MAP-Kinase) pathway gene alteration.
In the first case, diagnosis was made on the late relapse, 20 years later, of a tumor initially diagnosed as “diffuse astrocytoma grade II”. In the second case, initial pathological diagnosis was “ganglioglioma grade I” and this tumor was reclassified as DLGNT 8 years later after evidence for both 1p/19q codeletion and KIAA1549:BRAF fusion. A t-SNE analysis (t-distributed stochastic neighbor embedding) was performed (Fig. 3) including the following glioma reference classes: ganglioglioma (GG, 21 cases), anaplastic astrocytoma with piloid features (AAP, 20 cases), DLGNT (26 cases), LGm6-GBM (Ceccarelli et al., 2016, 13 cases) [14], PA-like low grade gliomas (PA-like, Ceccarelli et al., 2016, 26 cases), pilocytic astrocytoma (PA_INF, 28 cases), polymorphous low-grade neuroepithelial tumor of the young (PLNTY, 10 cases) and normal hemisphere (Norm_Hemi, 12 cases). Our first case clustered with the DLGNT tumor group, highly supporting its diagnosis. As expected, our second case, which did not demonstrated a significant confidence score on the classification algorithm, possibly due to a moderate proportion of normal cells / immune infiltration in the tissue sample, did not segregated in a clear methylation group. It showed similarity to AAP but was detached from the main cluster of this group. However, in contrast with AAP, this case did not demonstrated features of anaplasia, piloid cytology, ATRX loss nor CDKN2A/B homozygous deletion. In addition, clinical course was favorable in contrast with the reported evolution of AAP. It was a case of difficult diagnosis, for which pathological and molecular features fully met the diagnostic criteria of DLGNT with no argument towards a differential diagnosis. These two cases point out that the DLGNTs can occur in the absence of leptomeningeal dissemination and in the adult setting, in contrast to initial reports [5]. In such circumstances it is particularly challenging to think of this diagnosis since the pathological features remain non-specific and could mimic pilocytic astrocytoma, ganglioglioma, oligodendroglioma IDH-mutant and 1p/19q codeleted, extraventricular neurocytoma, as well as recently reported polymorphous low-grade neuroepithelial tumors of the young (PLNTYs) [15] and tumors possibly related to the last reported as BRAF V600E mutant oligodendroglioma-like tumors with chromosomal instability (BRAF-OLT) [16]. All these tumors have in common a focal oligodendrocyte-like proliferation (i.e pilocytic astrocytoma, ganglioglioma) or a more diffuse pattern (oligodendroglioma IDH-mutant and 1p/19q codeleted,extraventricular neurocytoma and PLNTYs/BRAF-OLT). Some pathological criteria, when present, might be more characteristic of one entity than others: Rosenthal fibers in pilocytic astrocytoma, lymphocytic cuffing, and eosinophilic granular bodies in ganglioglioma. Ganglion cells, a common feature of ganglioglioma might also be observed in DLGNT and in extraventricular neurocytoma [5, 17]. Extraventricular neurocytoma and oligodendroglioma IDH-mutant and 1p/19q codeleted often display a dense branched vasculature. Moreover pilocytic astrocytoma, ganglioglioma, oligodendroglioma IDH-mutant and 1p/19q codeleted or extraventricular neurocytoma share with DLGNT the OLIG2 and synaptophysin protein expression whereas PLNTYs are OLIG2-positive but most commonly lack synaptophysin expression [15]. In addition, ganglioglioma and PLNTYs/BRAF-OLTs frequently demonstrate CD34-immunopositivity. Furthermore, as observed in our cases, GFAP expression is commonly lacking in DLGNT whereas it is positive in the differential diagnosis mentioned above [5, 6]. The two cases reported here demonstrated oligodendrocyte-like proliferations of cells co-expressing OLIG2 and synaptophysin associated with a dense, often branched vasculature. Although rare, pathological features of anaplasia such as those observed in our first case (microvascular proliferation, high mitotic count and proliferation index) might occur in DLGNTs. Genetic findings help to distinguish DLGNTs from their main differential diagnoses. DLGNTs share MAP-Kinase alteration with pilocytic astrocytoma, ganglioglioma or PLNTYs/BRAF-OLTs. However, 1p deletion is not a common feature of pilocytic astrocytoma, when present it is usually subclonal [10], and 1p/19q codeletion does not occur. Furthermore, the 1p/19q codeletion is not encountered in ganglioglioma nor in PLNTYs/BRAF-OLTs, but by definition, is the hallmark of oligodendroglioma IDH-mutant and 1p/19q codeleted. The diagnosis of our second case was particularly challenging because it was supratentorial, occurring in an adult, and the pathological examination was suggestive of anaplastic oligodendroglioma. However, in spite of a 1p/19q codeletion, IDH mutation was lacking and KIAA1549:BRAF fusion was present. It is to note that, Badiali et al. [18] previously reported rare cases of gliomas IDH-mutant and 1p/19q-codeleted with co-occurrence of KIAA1549:BRAF fusion. Therefore, in adults, an oligodendrocyte-like tumor demonstrating 1p/19q codeletion, in the absence of IDH mutation should point towards DLGNT and the search for an associated MAP-Kinase alteration should be done by appropriate testing even if the tumor is supratentorial without leptomeningeal dissemination. The first case was also very unusual by the association of 1p/19q codeletion, PIK3CA mutation and BRAF V600E mutation. As far as we know PIK3CA mutation has never been reported. Although MAP-Kinase alteration is always recorded in DLGNT, the most frequent MAP-Kinase alteration reported in DLGNT is the KIAA1549:BRAF fusion although NTRK1/2/3 and TRIM33:RAF1 fusions might also occur [4]. The BRAF V600E mutation seems very infrequent in DLGNT. We could identify only one previously reported case of low-grade glioneuronal tumor with oligodendroglial features and BRAF V600E mutation that presented as an isolated temporal lesion that later disseminated. However, 1p data was not available for this case and thus whether it corresponded to a DLGNT remains questionable [19]. Two subtypes of DLGNT have been described by DNA-methylation profiling: DLGNT-MC-1 and DLGNT-MC-2 by Deng et al. [4]. Compared to DLGNT-MC-2, the age at diagnosis was lower in the DLGNT-MC-1 group, (median 5 vs 14 years, p < 0.01) and the clinical course less aggressive (5-year OS 100% vs 43% in DLGNT-MC-2). DLGNT-MC-2 is enriched for superimposed 1q gain whereas codeletion of 1p/19q was more frequently observed in DLGNT-MC-1. Our two cases demonstrating 1p/19q codeletion and excellent outcome likely belong to the DLGNT-MC-1 subtype (confirmed in one case, and assumed based on absence of 1q gain in the second).
To conclude, the terminology of “diffuse leptomeningeal glioneuronal tumor” appears unsuitable in our cases which had neither diffuse growth nor leptomeningeal dissemination suggesting that DLGNTs comprises a spectrum of tumors that has yet to be fully clarified. These tumors would be further classified in the large group of “gliomas and glioneuronal tumors driven by MAP-Kinase pathway alterations” in the fifth edition of the WHO classification of CNS tumors.