Pediatric low-grade glioma in the era of molecular diagnostics

Tumors of the central nervous system (CNS) are the most frequent solid tumors in children, with approximately 5.4-5.6 diagnoses per 100,000 [48, 154, 155]. Of those diagnosed, 0.7 per 100,000 will succumb to their disease, making CNS tumors the leading cause of cancer related death in children [154, 155, 168]. Within this group, pediatric-type low-grade gliomas (pLGG) are the most frequent, accounting for approximately 30% of all childhood brain tumors [154, 155]. pLGG are defined as World Health Organization (WHO) grade I or II malignancies and encompass a wide array of histologies that can arise throughout the neuro-axis (Fig. 1a-c) [131, 132].

Management of pLGG is intimately related to surgical resection, and complete resection remains the most favorable predictor of patient outcome [225]. Often, this is achievable for superficial lesions such as those arising in the cerebral hemispheres or posterior fossa, but is not always feasible for deep seated or highly infiltrative tumors [225]. In these cases, progressive residual disease has historically been treated with adjuvant chemotherapy or radiation [12, 55, 122, 124, 135, 139, 157, 183, 189]. Importantly, these treatments are associated with long-term sequelae and, particularly for radiation, increased mortality [55, 71, 118, 138, 145]. These concerns are poignant in a disease where 10-year overall survival (OS) exceeds 90%. However, with progression-free survival (PFS) at approximately 50%, up to half of patients will require adjuvant therapy. As such, a more robust risk stratification is required to help guide the type and intensity of therapy warranted. In the past, the degree of surgical resection, histological diagnosis and age were used to determine prognosis. However, more recently the molecular underpinnings of pLGG have emerged as a powerful tool to supplement the stratification of these tumors.

In the last decade, significant molecular data has emerged to suggest that pLGG near universally up-regulate the RAS-mitogen-activated protein kinase (RAS/MAPK) pathway [34, 96, 149, 229]. This data has led to increasing use of targeted therapeutics that supplement and/or replace older cytotoxic approaches. As the era of targeted therapeutics inevitably arrives, a concise classification scheme recognizing the molecular features of pLGG is needed. Here, we will review the histological spectrum of pLGG, the molecular alterations that have been identified in these entities and how to effectively test for them, and review the newest therapeutic agents and their utility in treating this disease. We conclude with proposing a multi-faceted approach for stratifying pLGG that considers clinical, histologic and molecular parameters and aims to aid clinicians in their future treatment decisions.

pLGG form a heterogeneous group of neoplasms that encompass tumors of primarily glial histology, including astrocytic and/or oligodendroglial, and tumors of mixed neuronal-glial morphology. These tumors are considered grades I and II according to the current WHO classification where they are distinguished from high grade glioma on the basis of specific morphologic features or, in the case of diffuse glioma, based on the absence of necrosis, mitoses and microvascular proliferation [131, 132, 166, 199]. Histologic diagnoses which fall under the umbrella of pLGG and their hallmark molecular alterations are listed in Table 1; their typical histologic features are depicted in Fig. 1d-j.

In many cases the different histologic entities are readily distinguished, however cases of overlapping morphology are well documented. These include, for example, reports of histological overlap between pleomorphic xanthoastrocytoma and ganglioglioma [3, 62, 208] and between dysembryoplastic neuroepithelial tumor and oligodendroglioma [70, 114]. In addition, tumors which are classically well circumscribed, such as pilocytic astrocytoma, may possess an infiltrative component [34], leading to confusion and difficulty in grading. A precise histologic diagnosis may be particularly challenging in deep seated midline tumors, for which a small biopsy is often all that is available. Rarely do these capture the true complexity of the tumor and the classic morphologic features by which diagnoses are made are often lacking.

In addition to these difficulties, pLGG overlap morphologically with entities more commonly found in adults. This creates confusion regarding appropriate grading and treatment and is exacerbated by use of similar terminology, namely diffuse astrocytoma and oligodendroglioma. In the most recent WHO iteration, both diffuse astrocytoma and oligodendroglioma have been split based on the presence or absence of IDH1 mutations, in addition to 1p/19q co-deletion for the latter. Tumors with the morphology of oligodendroglioma or diffuse astrocytoma in the pediatric age group often do not have IDH1 mutations and/or 1p/19q co-deletion and are therefore considered oligodendroglioma, NEC or, of even greater concern, diffuse astrocytoma, IDH-wildtype. The latter raising concern for molecular glioblastoma (GBM). Both of these diagnoses may lead to conventional adult diffuse glioma treatments involving cytotoxic chemotherapy and radiation, particularly in the adolescent and young adult age group. However, in IDH1 wild-type cases, pediatric oligodendrogliomas most frequently harbor alterations in FGFR1 including TKD-duplications or SNVs or BRAF p.V600E (Table 1). Recently, the entity polymorphous low-grade neuroepithelial tumor of the young (PLNTY) was described [88]. These tumors invariably possessed oligodendroglioma-like cellular components and highly infiltrative morphological features, yet boast a benign clinical course uncommonly seen in classic IDH–mutant oligodendroglioma [31, 88]. These tumors do not harbor IDH1 mutations, but rather FGFR2/3 fusions (discussed further below) or BRAF p.V600E. IDH1 wild-type diffuse astrocytoma most frequently harbor BRAF p.V600E mutations, accounting for ~40% of cases (Table 1). In addition, they sometimes contain KIAA1549-BRAF fusions, FGFR1 SNVs, or MYB or MYBL1 alterations. The latter alterations were originally described in series of pediatric diffuse astrocytomas [174, 212] and reports thus far suggest they have a benign clinical course without therapy [31]. Recently, these have been termed isomorphic diffuse glioma [223]. In these cases, misdiagnosis may result in over-treatment, leading to potentially harmful sequelae.

In recognition of the increased understanding of the molecular underpinnings of diffuse gliomas in adults, IDH1 mutation and 1p/19q deletion status were incorporated into the most recent WHO revision in order to improve diagnostic reproducibility and provide important prognostic information for patients [132]. A similar incorporation of molecular features into the classification of pLGG will help to more accurately identify these entities and, importantly, distinguish them from adult-type gliomas, which carry a worse prognosis and require more aggressive therapy.

Below, we discuss some of the common testing strategies used to molecularly profile pLGG and include their tissue requirements, cost, turn-around time, and target-specific applicability (Table 2).

Given the array of molecular alterations and their overlap amongst different tumor histologies, devising a simple testing recommendation for pLGG can be difficult. Ultimately, as proposed by Miklja et. al. [141], there are two primary approaches to the problem (i) sequential testing of specific alterations in a tier-based approach or (ii) upfront NGS panels optimized for pLGG. The latter may be used in centers with access to this technology and for whom cost is not an issue. For other centers the use of a sequential testing strategy is largely supported by the fact that the vast majority of pLGG harbor a single molecular driver within a subset of recurrently altered genes. These events primarily occur as either gene fusions or mutations, but almost never both. Rarely, exceptions arise where multiple mutations, either within the same gene or in other traditional pLGG targets, arise. These are almost exclusively observed with mutations and not gene fusions. The crux of this strategy lies in its ability to accurately identify a molecular driver prior to the number of tests conducted exceeding the cost and turn-around-time of an NGS based approach. A possible testing strategy highlighting the most probable molecular alterations present based on the tumor's clinical features is included in Fig. 4.

The importance of molecular testing in tumor diagnostics is increasingly recognised and became formalised for brain tumors in the most recent WHO classification [132]. As we gain a better understanding of the molecular underpinnings of pLGG, it is becoming evident that, while certain histologies may be enriched for particular molecular events (and vice versa), they are not exclusively associated with a particular event. Furthermore, while classic morphologies exist for the entities encompassed within the umbrella of pLGG, there remain cases with overlapping features between histologic categories, as was discussed above. Importantly, whether a particular molecular event carries the same prognostic significance across different pLGG entities is currently unclear. Given this, a layered diagnostic approach is recommended where both the histologic classification and molecular findings are reported in an integrated diagnosis [133]. Most importantly, pLGG need to be distinguished from their adult-type counterparts as both clinical management and long term outcome are drastically different.

A comprehensive risk based classification of pLGG lies in an integrated model, utilizing clinical, imaging and molecular information to concisely categorize tumors based on their potential clinical risk (Fig. 5) [95, 97, 123, 171, 229, unpublished data]. The scheme we propose here attempts to incorporate these factors into one tool. For example, a pLGG with typically benign histology, a KIAA1549-BRAF fusion, and arising in a child between 3-12 years would typically be viewed as low risk and a "watch and wait" strategy may be employed, followed by less aggressive therapies if the tumor were to progress. In contrast, a tumor in an unfavorable location or highly disseminated with high risk molecular features will require close clinical follow-up and a more aggressive therapeutic approach (Fig. 5). This schematic approach would also allow for amendments incorporating adjunct strategies such as the methylation classifier [20] or other novel molecular targets upon their discovery.

The era of precision medicine for pLGG has arrived. Molecular stratification of pLGG resulting in significant clinical implications is currently available and has been seen in trials for specific inhibitors such as BRAF p.V600E- and MEK-inhibitors. The expected availability of FGFR-targeted agents, as well as other tyrosine kinase inhibitors for rare fusions, makes precision diagnostics key to the management of these patients. Indeed, the current National Cancer Institute–Children’s Oncology Group Pediatric MATCH trial (NCT03155620) aims to match actionable mutations to 9 investigational therapies, providing a glimpse into the future of pLGG treatment. In this context it is important to be aware of which methods are available to be used that are not reliant on expensive NGS-based technologies, and here we provide a testing pipeline to aid in testing decisions. Importantly, molecular stratification is only one factor influencing the behavior of pLGG. Other factors such as age, tumor location, and histopathology are required to inform a comprehensive approach to prognostication and treatment of pLGG. We therefore propose a pLGG risk classification schema that utilizes the breadth of clinical and molecular information available to best equip clinicians as we transition to this new era of pLGG classification and treatment.