Neurocytomas in the pediatric population have limited representation in the literature due to their low incidence [18, 19]. We provide a diverse five patient pediatric case series that represents histologically diagnosed central and extraventricular neurocytomas, typical and atypical neurocytomas, and recurrent neurocytomas. As neurocytomas generally exhibit benign characteristics and respond well to surgical intervention alone, all patients underwent initial surgical resection without adjuvant chemoradiation. Initial disease recurrence was managed by surgical intervention alone or with subsequent radiation therapy. Despite initial surgical interventions, four of the five patients demonstrated recurrent disease which differed markedly from a single institution long-term follow up of adult neurocytomas in which two thirds of patients had durable progression free survival after surgical resection alone [22]. This higher-than-expected disease recurrence may have been due to discordance between the histologic diagnosis and integrated diagnosis of these CNS tumors. Notably, despite a higher-than-expected recurrence rate, the patients in this series have had a 100% survival rate at time of last follow up (range 44–112 months).

This series demonstrates the wide range of radiological characteristics, tumor locations, and sizes that can be seen in the initial presentation of histologically diagnosed neurocytoma. One patient with nausea, vomiting, and visual disturbances was found to have central neurocytoma with secondary hydrocephalus on their MRI. Both patients with histologically diagnosed extraventricular neurocytoma had presenting symptoms of seizures and were initially managed with consideration for epilepsy surgery prior to their eventual tumor identification.

Histologic classification of tumors prior to molecular testing depends largely on proliferation architecture, cytologic form, nuclear details, and background features, which must be taken in the context of the broader clinical and radiologic differential. Diagnosing neurocytomas is often a challenging endeavor and is particularly difficult based on histology alone. Histologically, our tumor samples generally demonstrated a uniformly distributed, moderately to highly cellular proliferation, with round monotonous to moderately pleomorphic nuclei with stippled chromatin and eosinophilic cytoplasm (Fig. 2). The heterogeneity of these tumors became more apparent upon immunohistochemistry staining (Table 1), with two tumors (cases 1 and 2) demonstrating high rates of MIB-1 staining for Ki-67 antigen consistent with a diagnosis of atypical neurocytomas. There was recurrence of disease in both patients identified as having atypical features, with one patient having multiple recurrences, which is consistent with the higher rates of disease recurrence reported for atypical neurocytomas in the literature. Several histologic and molecular features of the tumor identified in case 2 raised question of other tumor diagnoses, particularly an intra-axial mesenchymal tumor of the CNS owing to the EWSR-ATF1 gene fusion. Ultimately, this diagnosis was lowered on the differential when external epithelial membrane antigen testing (EMA) was found to be negative and internal testing was found to only have scattered light positivity.

Next generation sequencing has proven as a useful tool in identifying the molecular drivers of individual tumors, and chromosomal microarray has been used to identify copy number variants that a tumor has accumulated in its course—but even with these molecular tools the diagnosis may remain elusive. In our cohort, next generation sequencing identified an EWSR-ATF1 gene fusion and MUTYH mutation in one tumor (case 2), variants of uncertain significance in two tumors (cases 1 and 5), and no clinically significant finding in two tumors (cases 3 and 4). Clinically, case 2 had multiple recurrences and was more refractory to treatment than typical neurocytomas which may be due to its unique molecular drivers. Notably, there were no cancer predisposition germline mutations identified on whole genome sequencing of this patient, no family history of cancer, and no report of polyps identified in siblings to suggest a somatic type MUTYH variant. Case 3 highlights the utility of negative next generation sequencing, as histologically a septal dysembryoplastic neuroepithelial tumor (DNET) and myxoid glioneural tumor (GNT) were considered. As NGS was negative for PDGFRA mutations including p.k385, myxoid GNT was lowered on the differential [23]. Methylation analysis would have proved helpful in further narrowing the differential on this tumor, however this analysis could not be performed as the sample did not meet the minimum tumor content cutoff. Clinically, case 4 had multiple recurrences requiring repeated surgical interventions which is not typical for the expected clinical course for either neurocytoma or ganglioglioma.

Chromosomal microarray analysis generally does not demonstrate significant copy number abnormalities in neurocytomas, which is relatively consistent with the findings in our first three cases. The tumors ultimately classified as gangliogliomas had chromosomal microarrays with more broad copy number gains, thought to be less consistent with their histologic diagnosis of neurocytoma. Interestingly, gangliogliomas are typically also not associated with the significant broad copy number changes seen in cases 4 and 5, however they may rarely demonstrate gains in chromosome 7 and X, as seen in both cases. Diffuse glioneuronal tumor with oligodendroglioma-like features and nuclear clusters (DGONC) was additionally considered on the differential for case 5 due to its relatively low calibrated score for ganglioglioma on v11b6 methylation analysis, however the monosomy 14 more typically seen in DGONC was not seen on chromosomal microarray [24] and the tumor did not cluster with DGONC on v12 of the methylation classifier (publicly available, unpublished).

Methylation profiling has emerged as a powerful tool in the field of neuro-oncology, particularly in identifying new medulloblastoma subgroups that can aid in prognosis and management [25,26,27,28,29]. Though this diagnostic tool is not readily available for widespread clinical use, its strength in guiding integrated diagnoses and identifying new molecular drivers of disease suggests that it may have an emerging role in the clinical practice of neuro-oncology [29]. Methylation profiling affirmed our diagnosis of a central neurocytoma (calibrated score 0.999) in one case (case 1) and identified two cases of histologically diagnosed neurocytomas that clustered more closely with gangliogliomas, providing an alternative integrated diagnosis (calibrated scores 0.766—0.927). It should be noted that although methylation profiling was able to confidently classify the histologically diagnosed central neurocytoma in case 1 with a calibrated score of 0.999, this technique is not yet able to distinguish between typical and atypical neurocytomas. This series is the first to date to utilize methylation profiling results to provide an integrated diagnosis on histologically identified neurocytomas and highlights the diagnostic complexity of this rare brain tumor in the pediatric population. As both cases initially histologically diagnosed as extraventricular neurocytomas clustered with gangliogliomas on methylation profiling, this study highlights extraventricular location as a feature of histologically diagnosed neurocytomas that may benefit from increased diagnostic accuracy through this emerging molecular diagnostic technique.

This series is limited by its small sample size and lack of exclusion criteria, owing to the rarity of histologically diagnosed neurocytomas in the pediatric population. While this study demonstrates the potential future clinical utility of methylation profiling in providing an integrated diagnosis for difficult to diagnose CNS tumors, it also highlights the current limitations of available histologic and radiographic tools in distinguishing biologically similar rare CNS tumors. Two of four analyzed tumors in this study had methylation profiles that differed from their initial histologic diagnosis (calibrated scores 0.766–0.927), though it is challenging to interpret the significance of this finding as there are is no historical comparison for use of this novel technique in the integrated diagnosis of histologically diagnosed neurocytoma. Overall, this study highlights that neurocytomas warrant further investigation with particular attention to the role for next generation sequencing and methylation profiling in providing early and precise integrated diagnoses.