Introduction
A diverse array of low-grade neuroepithelial tumors (LGNTs) has been associated with epilepsy in children and young adults. These neoplasms exhibit varied histopathologic features and glial or glioneuronal differentiation, recognized types including pilocytic astrocytoma, diffuse astrocytoma (DA), ganglioglioma, pleomorphic xanthoastrocytoma (PXA), dysembryoplastic neuroepithelial tumor (DNET), angiocentric glioma, and oligodendroglioma [
26]. Recent large-scale molecular profiling studies have established the genomic landscape characterizing the broad spectrum of LGNTs, while also pointing toward more molecularly uniform disease subgroupings [
31,
48]. In particular, genome-wide DNA methylation profiling and the identification of recurrent genomic aberrations (e.g., mutations, rearrangements, and copy number abnormalities) have been crucial in the delineation of biologically distinct disease entities and in refining tumor classification.
Here, we describe a unique epileptogenic LGNT variant affecting children and young adults, which we call polymorphous low-grade neuroepithelial tumor of the young (PLNTY). We show that PLNTYs, while morphologically variable, are uniformly characterized by the presence of oligodendroglioma-like cellular components, infiltrative growth patterns, and intense CD34 immunopositivity. Moreover, we demonstrate that PLNTYs exhibit a distinct DNA methylation signature, most closely related to that of ganglioglioma, and harbor molecular abnormalities involving mitogen-activated protein kinase (MAPK) pathway constituents. Our findings suggest that PLNTYs may represent a distinct biological entity within LGNTs of children and young adults.
Discussion
LGNTs of children and young adults are a heterogeneous group of neoplastic entities whose fundamental identities reflect both molecular and histogenic factors. In recent years, comprehensive genomic profiling in large patient cohorts has confirmed the biological distinctiveness of some well-established histopathologic variants, such as pilocytic astrocytoma, PXA, and angiocentric glioma [
3,
4,
22,
30,
31,
33,
38,
48]. However, other morphological subtypes, including oligodendroglial tumors, have exhibited more variable molecular features, implying the underlying existence of multiple distinct disease entities. In this study, we sought to characterize a specific oligodendroglioma-like LGNT with characteristic clinical behavior, localization, and histopathology. We found that these tumors form a cohesive biological subgroup characterized by intense CD34 expression, molecular abnormalities involving MAPK signaling, and a distinct DNA methylation signature.
The designation of PLNTYs as “polymorphous” is meant to acknowledge the fact that, while all harbored components prompting consideration of oligodendroglioma in the differential diagnosis, most also contained populations of patently astrocytic or ambiguous appearance and several manifested foci of vague or more fully developed perivascular pseudorosetting that raised question of ependymocytic differentiation. Consistent expression of GFAP and OLIG2 in the absence of dot- or ring-type EMA labeling and failure to express neuronal “markers” would argue for these lesions fundamentally representing non-ependymal gliomas. However, we think it prudent to reserve final judgment on their fundamental differentiating capacity and so refer to these generically as “neuroepithelial”. The issue of their molecular relationship to gangliogliomas is further addressed below. There can be little doubt that the neoplasms of the PLNTY group are part of a larger family of tumors reported as diffuse astrocytomas or oligodendrogliomas of pediatric type having MAPK pathway-activating genetic abnormalities [
31,
48]. Similarly, the lesions we describe as PLNTYs would appear to lie among CD34-expressing neoplasms constituting a subset of growths for which Blumcke and colleagues have proposed the designation “long-term epilepsy associated tumors” (LEATs) [
7]. Whatever terminology is eventually employed, the designation of PLNTYs should clearly segregate them from the ineluctably recurring and genetically distinct gliomas of infiltrating type encountered in generally older patients.
A constant and striking feature of our tumors was intense, frequently widespread expression of CD34 by tumor cells and by ramified neural elements in the regional cerebral cortex. In fact, it was this immunohistochemical observation, along with the consistent failure of the tumors in this series to manifest labeling for IDH1 R132H or chromosome 1p/19q codeletion, despite their oligodendrogliomatous appearances, that prompted us to consider these as distinct from the infiltrating gliomas of later life and to initiate additional molecular studies. While the immunohistochemical labeling of some glioblastomas for CD34 is recognized [
19,
25], this phenomenon is generally foreign to infiltrating astrocytomas of lower grade and to neoplasms of the oligodendroglioma group [
2,
5,
6,
10,
13‐
15,
28,
29,
32,
36,
43,
45‐
47]. Only one of the many studies that have examined CD34 expression in gliomas found tumoral (as opposed to endothelial) immunoreactivity in lesions classified as WHO grade II or III astrocytomas, oligodendrogliomas or oligoastrocytomas [
8]. Of note, and emphasized by the authors of this single outlier, such reactivity was confined to examples deriving from the temporal lobes of patients with chronic and refractory seizures whereas “tumor samples derived from non-epileptic patients with lesions outside the temporal lobe never showed CD34 -immunoreactive neural cells” [
8]. It is likely that at least some of these ostensibly conventional gliomas represented neoplasms of the type we describe.
In addition to distancing lesions of the PLNTY group from conventional “adult-type” gliomas of the infiltrating variety (a matter of great practical import given that the latter are rarely curable by surgical or adjuvant means, are prone to progression and eventually fatal in the great majority of cases), shared CD34 expression may well reflect a link between PLNTYs and other tumors long recognized for their association with chronic epilepsy in young subjects. Specifically, expression of CD34 is a common attribute of pediatric gangliogliomas [
8‐
10,
17] and PXAs [
17,
35], having also been recorded by some observers in the setting of DNET [
42,
44]. There is, furthermore, a strong association of these epileptogenic tumors with the presence in regional cortical tissues of the ramified, CD34- expressing neural elements observed in complex with each of our cases. Foreign to the normal human brain, these have also been found in association with glioneuronal microhamartomas (“hamartia”) and some forms of cortical dysplasia [
8,
17]. The transient expression of CD34 in the murine neural tube [
12,
24] has elicited speculation that these ramified cells represent developmentally arrested/dysregulated neural progenitors that testify to regional dysontogenesis and that may give rise to the CD34-expressing tumors with which they are identified [
8‐
10]. The superficial positioning of the neoplasms reported here, their predilection for the temporal lobes, presentation in the young, association with chronic epilepsy, general indolence, CD34 expression, MAPK pathway-activating genetic lesions and methylation profiles would all support a link to the pediatric ganglioglioma and other “developmental” tumors. In fact, it could be argued that the tumors of our PLNTY group represent “lopsided” variants of ganglioglioma in which neuronal components and even immunophenotypic evidence of neuronal differentiation may not be demonstrable, though whether these ostensibly infiltrative lesions are as biologically stable and surgically curable as classical gangliogliomas remain to be seen.
Using focused molecular profiling, we found that nearly all of the PLNTYs harbored either BRAF V600E mutations or fusion events involving FGFR2/FGFR3. These abnormalities occurred in a mutually exclusive fashion, consistent with the notion that shared downstream molecular sequelae fundamentally drive the pathogenesis of PLNTYs. Indeed, both V600E-mutant BRAF monomers and FGFR fusion chimeras have been shown to activate MAPK signaling, the latter by way of constitutive receptor dimerization [
16,
39], a mechanism likely operative in all such cases in our sample set (see above). That being said, abnormal MAPK pathway activation, by either BRAF V600E or FGFR fusion, is hardly specific to PLNTYs as a transforming molecular event. BRAF V600E has been widely implicated in the pathogenesis of LGNTs, with particularly high rates of incidence in gangliogliomas and PXAs [
37]. Similarly, recent work has identified frequent FGFR fusion events in DNETs, and diffusely infiltrating pediatric gliomas exhibiting either astrocytic or oligodendroglial histopathology [
31]. As stated above, we believe that PLNTYs likely represent a subset of this latter group. BRAF V600E mutations and FGFR fusions are even found in a small minority of glioblastomas [
11]. Notably, none of the tumors in our sample cohort harbored KIAA1549-BRAF fusions, which are commonly associated with pilocytic astrocytomas and were recently reported in a small set of two neoplasms classified as “pediatric oligodendrogliomas” [
23]. The broad spectrum of brain tumors exhibiting MAPK pathway activation attests to its importance as an oncogenic mechanism, while also indicating that its precise transformative consequences are highly dependent on biological context.
Over the past several years, genome-wide DNA methylation profiling has become an influential tool informing the classification of primary brain tumors. For instance, methylation signatures have been repeatedly utilized to establish endogenous subclasses within morphologically based disease entities, most notably for glioblastoma and medulloblastoma [
20,
41]. Methylation profiling has also facilitated more appropriate tumor classification within poorly defined histopathologic groupings. In a recent study of supratentorial primitive neuroectodermal tumors (PNETs), methylation arrays delineated the vast majority of sampled neoplasms into other more well-established diagnostic categories, including glioblastoma and ependymoma [
40]. In these respects, specific patterns of CpG island methylation capture the biological distinctiveness of a given neoplastic process. In our analysis, PLNTYs exhibited a unique DNA methylation signature whose distinctiveness even allowed the identification of additional cases from a much larger sample cohort that had already been subjected to profiling. Interestingly, the PLNTY methylation signature was most closely related to those of other, morphologically diverse LGNTs—gangliogliomas, pilocytic astrocytomas, and DNETs—characterized by MAPK pathway activation. This finding implies that the unique aspects of the PLNTY methylation signature relate to issues of histogenesis and differentiation state rather than the primary oncogenic driver mechanism.
Applying our findings to the diagnostic setting, several considerations should be kept in mind. While the demonstration of aberrant CD34 expression by neoplastic and ramified neural elements certainly suggests PLNTY in the appropriate context, calling into serious question the diagnosis of any conventional glioma of infiltrating type, it would be premature to conclude that this finding obviates the need for assessments of
IDH1/2 mutation and chromosome 1p/19q codeletion status in the evaluation of neurosurgical specimens. In a similar vein, CD34 expression cannot be proposed as a surrogate marker of MAPK pathway activation at this time. Candidate cases should be studied for
BRAF and
FGFR2/3 abnormalities and an integrated diagnosis rendered on the basis of the histologic, immunophenotypic, and molecular genetic features. In fact, we believe that ostensibly infiltrating glial neoplasms deriving from young individuals (especially those posing as chromosome 1p/19q non-codeleted oligodendrogliomas) should be assessed in similar fashion irrespective of CD34 expression status, as the CD34-labeling tumors we describe may constitute but a subset of fundamentally unified neuroepithelial neoplasms sharing a spectrum of MAPK pathway-activating genetic lesions. In this connection, we would point out that the “pediatric oligodendrogliomas” (and other low-grade gliomas of the young) reported to date as displaying BRAF and FGFR abnormalities seem not to have been systematically studied for CD34 expression [
23,
31,
48].
In summary, we report the characterization of PLNTY, a molecularly distinct LGNT that likely represents a significant subset of oligodendroglioma-like tumors arising in pediatric populations. PLNTYs exhibit an almost invariably benign clinical course and appear to be well controlled by gross total resection. Moreover, their driving molecular alterations, which uniformly activate MAPK signaling, are potentially targetable with existing small molecular inhibitors, providing additional therapeutic options for affected patients.