Introduction

Pediatric low-grade gliomas (PLGGs) are clinically and genetically distinct from adult LGG [63]. PLGGs include pilocytic astrocytoma (PA), dysembryoplastic neuroepithelial tumor (DNET), ganglioglioma (GG), pleomorphic astrocytoma (PXA), subependymal giant cell astrocytoma (SEGA), diffuse astrocytoma (DA) and diffuse oligodendroglial tumors (d-OT), and other tumors [70]. Recent large-scale genomic analysis uncovered the genetic landscape of pediatric brain tumors, including PLGG. Importantly, there is a tight relationship between specific genetic alterations and anatomical tumor location in PLGG. Several clinical trials, based on molecular target therapy, are ongoing in PLGG patients. Herein, we summarize the genetic characterization of PLGG and the relevant molecular therapeutic targets potentially suitable for precision medicine.

Genetic characterization of PLGG

Somatic gene abnormalities

MAPK signaling pathway

Although, histologically, PLGGs comprise different tumors, the vast majority of PLGGs are caused by various genetic alterations affecting the MAPK pathway (Fig. 1) [53]. In particular, genetic alterations to promote MAPK pathway activation are found in almost the totality of PA cases [29, 56, 75].

Fig. 1
figure 1

Major oncogenic pathways (MAPK pathway and PI3K/AKT/mTOR pathway) in pediatric low-grade gliomas (PLGG). Blue fonts. Major oncogenic alterations in PLGG. Red fonts. Proposed molecular therapeutic inhibitors for suppressing major oncogenic pathways

Full size image

KIAA1549: BRAF fusion

A 2-Mb tandem duplication of 7q34, encompassing the BRAF gene, was commonly detected in PA cases [11]. Importantly, younger patients with infratentorial posterior fossa PA tend to display a high frequency of the BRAF fusion [23, 30]. In the tandem duplication, the N-terminal end of the KIAA1549 protein replaces the N-terminal regulatory region of BRAF, while the BRAF kinase domain is retained, resulting in constitutive activation of the MAPK pathway [30, 75]. Other BRAF fusions, involving FAM131B, RNF130, CLCN6, MKRN1, GNA11, QKI, FZR1, and MACF1, were also identified. These events also induce loss of the N-terminal regulatory region of the BRAF protein and constitutive activation of the kinase domain. The BRAF fusions arise from various genetic mechanisms, including deletions and translocations in PA [29, 75]. In addition to BRAF fusions, BRAF small insertions, activating BRAF kinase signaling, were also observed in a small proportion of PA [29]. BRAF fusions are almost exclusive to PA [70]. PAs carrying the KIAA1549:BRAF fusion occur in almost all anatomical locations, but most frequently in the cerebellum (Fig. 2A) [10].

Fig. 2
figure 2

a Axial (left) and sagittal (right) contrast enhancing MR imaging indicating contrast-enhanced lesion with cyst in cerebellum. Hematoxylin and eosin staining indicating pseudo-oligodendroglial pattern. b Axial (left) and sagittal (right) contrast enhancing MR imaging indicating weak contrast-enhanced lesion in suprasellar lesion. Hematoxylin and eosin staining showing typical biphasic pattern with Rosenthal fibers (arrows). These tumors are histologically diagnosed as pilocytic astrocytoma, but harboring distinct genetic alteration. Magnification, × 400

Full size image

BRAF V600E mutation

PLGGs bearing the BRAFV600E mutation comprise a heterogeneous group of tumors with divergent histology, location, cooperating genetic alterations, and outcomes [31]. The BRAFV600E status alone is not a sufficient diagnostic or prognostic biomarker of PLGG [31]. A strong association between the BRAFV600E mutation and tumor location was observed in PA (WHO grade I). More specifically, 20% of extra-cerebellar PA were found positive for the BRAFV600E mutation, whereas only 2% cerebellar PA harbored the mutation [65, 70].

The BRAFV600E mutation was observed in approximately 65% of PXA cases (WHO grade II and grade III) [14, 64], and is highly correlated with age. PXA patients carrying the BRAFV600E mutation were mainly observed in young subjects (mean age; 18 years), while PXA tumors related to wild-type BRAF display a later onset (mean age; 38 years) [64]. The BRAFV600E mutation was also found in 18–50% of GG (WHO grade I), 30–51% of DNET (WHO grade I), and 43% of SEGA (WHO grade I), respectively [8, 42, 55, 65].

FGFR1 mutations, fusions, and kinase domain duplications

Recent large-scale genetic analysis uncovered recurrent FGFR1 somatic mutations (p.N546K, p.K656E), FGFR1–TACC1 fusions, and a novel internal tandem FGFR1 kinase domain duplication in patients with PA and DNET [29, 56, 76]. Abnormalities in FGFR1 are frequent in PLGGs with oligodendroglial phenotype and composed of primary of oligodendrocyte-like cells, being present in 82% of DNETs and 40% of d-OT, respectively [56].

Fusions of the NTRK gene family

NTRK receptor transcript fusion is a very rare event in adult and pediatric tumors (0.3%), but relatively frequent in pediatric melanoma (11.1%) and PLGG (2.5%), as well as pediatric high-grade glioma (PHGG, 5.3%) [50]. Among NTRK family members, NTRK2 fusions (QKI-NTRK2 and NACC2-NTRK2 fusions) were identified in PA [29]. The NTRK gene fusions are characterized by various 5′partners with a dimerization domain that presumably leads to constitutive dimerization and kinase activation in PA [29, 75].

Homozygous CDKN2A deletion

CDKN2A is a tumor-suppressor gene and key regulator of cell cycle. A CDKN2A homozygous deletion inactivates cell cycle regulation and promotes malignant cancer progression. Consistently, loss of CDKN2A (p16) in immortalized human astrocytes abrogated the senescent features [26]. Although activated BRAF alone is not sufficient for tumorigenesis, the combination of BRAF activation and CDKN2A loss results in cell transformation [61], indicating a critical role for CDKN2A homozygous deletion in tumor progression. Indeed, BRAFV600E mutation and CDKN2A deletion were found to be early genetic events in PLGG with malignant transformation [46]. In BRAFV600E mutant PLGGs, CDKN2A deletion was rated as a poor independent prognostic factor [40]. CDKN2A homozygous deletion was found in 25% of the BRAFV600E mutant PLGGs [40], 60% of PXA, and 16.7% of BRAFV600E mutant GG. However, the number of available studies is relatively small [55, 72].

Amplification and/or rearrangement of MYB/MYBL1 and MYB–QKI fusion

A triple alteration consisting of MYB truncation, increased expression through enhancer hijacking, and loss of QKI tumor suppressor function was identified in a proportion of PLGGs with astrocytic phenotype (diffuse astrocytic and angiocentric glioma). An MYB–QKI fusion characterizes almost all angiocentric gliomas (87%), while MYB–ESR1 fusion and QKI rearrangement was also identified in a subset of angiocentric gliomas [56]. In contrast, MYB or MYBL1 rearrangement was detected in 41% of DA [56]. MYB/MYBL1 alterations were co-occured with BRAF alteration in a subset of DA [56, 57, 75].

ROS1 or ALK fusions

ALK or ROS1 rearrangements, which activate RAS/MAPK, PI3K, and JAK/STAT pathways, have been identified in a subset of PHGG and PLGG [9, 28, 48, 52]. These genetic events were found in infants and younger children.

Germline gene abnormalities

TSC1/TSC2

Tuberous sclerosis complex (TSC) is an autosomal dominant neurocutaneous disorder caused by mutations in either hamartin-encoding TSC1 or tuberin-encoding TSC2. In the central nervous system, TSC is characterized by the development of SEGA, subcortical heterotopic nodule, and cortical tuber [47]. SEGAs belong to the group of astrocytic neoplasms, even though they are also expressed in neurons. The majority of SEGAs exhibit loss of heterozygosity in the same gene in which the mutation of TSC1 or TSC2 was identified [3], while no evidence of BRAFV600E or other BRAF mutations was found in these tumors.

Neurofibromatosis type I (NF1)

The NF1 gene encodes an RAS GTPase-activating protein known as neurofibromin and is one of several genes that affect RAS–MAPK signaling [58]. The somatic NF1 loss occurred by frameshift mutation, loss of heterozygosity, and methylation and no additional gene alteration was identified in NF-1 associated PA, indicating NF1 loss is likely sufficient for PA formation [22]. About 20% of NF1 patients develop PAs [47], most of which are located in the supratentorial midline (Fig. 2B) [70].

Molecular target therapy for pediatric low-grade tumors

Target for BRAF V600E mutation in pediatric gliomas

A large clinical study indicated that patients with BRAFV600E mutant PLGG exhibited poorer prognosis compared to patients with wild-type BRAF. Moreover, conventional therapies were scarcely effective in BRAFV600E mutant PLGG patients [40]. Thus, novel therapeutic strategies are needed for BRAFV600E mutant gliomas.

Since BRAFV600E triggers the constitutive activation of MAPK signaling pathway (Fig. 1), the latter has been proposed as a therapeutic target. An early preclinical study demonstrated that BRAF knockdown and pharmacological BRAF inhibition induce p-ERK inactivation and antiproliferative activity in BRAFV600E-expressing, but not in wild-type BRAF-expressing malignant gliomas. The first-generation BRAF inhibitor, PLX4720, extended the overall survival in a BRAFV600E mutant orthotopic xenograft model, indicating that BRAFV600E is a promising therapeutic target in BRAFV600E-expressing brain tumors [49].

Burger et al. demonstrated a durable response to dabrafenib (a first-generation BRAF inhibitor) in three patients with disseminated leptomeningeal BRAFV600E-positive glioma. Of note, one patient was stable for up to 27 months. In addition, they established a patient-derived cell line and found that dabrafenib reduced the cell density and inhibited p-ERK [5], indicating that drug sensitivity is through on-target effect. A phase-1/2 clinical trial demonstrated that the objective response rate of dabrafenib is 38% in PLGG [13]. Another clinical trial for pediatric BRAFV600E mutant relapsed/refractory PLGG demonstrated that dabrafenib was also effective (44% overall response rate) and tolerable (clinicaltrials.gov, NCT01677741). Another first-generation BRAF inhibitor, vemurafenib (a PLX4720 analog), controlled tumor progression in patients with recurrent PXA [7, 43]. In addition, a vemurafenib-based regimen decreased tumor size in patients with brain stem GG expressing the BRAFV600E mutation [62]. In another study, all six BRAFV600E-mutant PLGG patients favorably responded to BRAF inhibitors [40]. The VE-BASKET study, focusing on BRAFV600E-mutant non-melanoma cancers, demonstrated the characterization of 24 patients with glioma received vemurafenib until disease progression. The objective response rate was 42.9% (3/7) in PXA, 9.1% (1/11) in malignant diffuse glioma, and 33.3% (2/6) in other tumors, including one pilocytic astrocytoma [32]. Another phase-2 basket study also indicated a 75% (3/4) response after vemurafenib monotherapy in PXA patients with the BRAFV600E mutation [25].

Unfortunately, resistance to the BRAF inhibitor is frequently observed due to MAPK pathway reactivation, which is mediated by RAF-independent activation of MEK and ERK [69]. Additionally, acquired gene alterations conferring drug resistance, including NRAS, BRAF, MEK1, and MEK2, were identified after treatment with BRAF inhibitors [71]. In contrast, combined BRAF/MEK inhibition was reported to block p-ERK signaling and prevent the acquisition of resistance in melanoma cell lines [54]. In addition, a phase-1/2 clinical trial demonstrated that a combination of dabrafenib and trametinib (MEK inhibitor) prolonged progression-free survival (PFS) in patients with metastatic melanoma [44]. Various phase-3 studies demonstrated improved overall survival after treatment of naïve unresectable or metastatic melanoma patients with BRAFV600E mutation [45, 60]. Similar combination effects were also observed in other solid cancers, including colorectal cancer [12].

Consistently, whereas monotherapy with a BRAF inhibitor results in transient MEK-ERK inhibition in BRAFV600E mutant glioma cell lines, combined BRAF/MEK inhibition prevented MAPK reactivation, resulting in enhanced antitumor effects, both in vitro and in vivo [77]. Moreover, in a BRAFV600E mutant-expressing glioma cell line, combined BRAF/MEK inhibition was found to be more effective compared to the combination between BRAF and mTOR inhibitors (everolimus) or that between an MEK1/2 inhibitor (AZD6244) and everolimus [51]. A higher efficacy of the combined BRAF/MEK inhibition was also observed in a BRAFV600E mutant-expressing and CDKN2A-deficient syngenic mice glioma model [21], which is considered a high-risk group in BRAFV600E mutant PLGG [40]. Brown et al. reported a clinical–radiological response in two cases of anaplastic PXA treated with a combination of BRAF and MEK inhibitors [4]. Another group described the results of BRAF/MEK inhibitor treatment of two patients with BRAFV600E-positive tumors (one anaplastic PXA and one glioblastoma (GBM)) after conventional treatment. Accordingly, the anaplastic PXA case exhibited a partial response for 14 months, although progression was observed thereafter. On the other hand, the GBM patient was stable after 16 months of treatment [66]. In a phase-2 open-label trial (clinicaltrials.gov, NCT02034110), patients with the BRAFV600E mutation received dabrafenib plus trametinib until unacceptable toxicity, disease progression, or death. In that study, 37 high-grade glioma (HGG) patients were enrolled, 31 of whom were evaluable for response. The overall response rate was 26% (8/31), including one complete response. Of note, five of eight (62.5%) responding patients exhibited a response duration of more than 12 months. A global, open-label phase-2 study (clinicaltrials.gov, NCT02684058), evaluating dabrafenib in combination with trametinib in pediatric patients with BRAFV600E-mutant HGG, is currently ongoing, as well as a phase-1/2 study employing the trametinib/dabrafenib combination in PLGG (clinicaltrials.gov, NCT02124772). In addition to MAPK pathway reactivation, it has been demonstrated that resistance to BRAF inhibitors may occur through activation of the EGFR signaling pathway in glioma [74]. Therefore, a combination of BRAF inhibition and EGFR target therapy might be an alternative therapeutic option for BRAFV600E-mutant glioma.

Target for KIAA1549-BRAF fusion gliomas

Importantly, a paradoxical activation of MAPK signaling after treatment with PLX4720 (first-generation BRAF inhibitor) was reported in KIAA1549-BRAF fusion cells [68]. Similarly, sorafenib, a multi-kinase inhibitor targeting BRAF, VEGFR, PDGFR, and c-kit induced MAPK-dependent stimulation of tumor growth, mostly in recurrent/progressive PLGGs carrying the BRAF1549-KIAA fusion [33]. To overcome this issue, a second-generation BRAF inhibitor was developed and found to prevent the paradoxical MAPK activation [68]. However, another study reported that after treatment with the second-generation BRAF inhibitor, PLX8394, inhibitor-resistant cells emerged because of PI3K/AKT/mTOR pathway activation in cells stably expressing the KIAA1549-BRAF fusion [27], further emphasizing the difficulty in controlling KIAA1549-BRAF fusion tumors using a single BRAF inhibitor.

In addition to the BRAF inhibitor, the MEK1/2 inhibitor demonstrated antitumor effect in a BRAFV600E mutant xenograft model [38]. A phase-1 Pediatric Brain Tumor Consortium study indicated that selumetinib (AZD6244, MEK1/2 inhibitor) displays promising antitumor activity in PLGG [1]. Currently, selumetinib effects are under examination in young patients with recurrent or refractory LGG (clinicaltrials.gov, NCT01089101). Preliminary results indicated partial response in 29% (2/7) of BRAFV600E mutant tumors and 33% (6/18) of tumors carrying the BRAF-KIAA1549 fusion.

A preclinical study demonstrated that the KIAA1549-BRAF fusion regulates neuroglial cell growth in an mTOR-dependent manner [34], implying that targeting of mTOR and downstream pathways may be another suitable therapeutic strategy for KIAA1549-BRAF fusion tumors. Notably, KIAA1549-BRAF-containing cells were sensitive to the MEK1/2 inhibitor, AZD6244, and the combination of AZD6244 and everolimus-induced enhanced effect, when compared with BRAF wild-type cells [51]. AZD6244 and everolimus reduced p-ERK and p-S6, respectively, and combination treatment enhanced suppression in both signaling [51]. Additionally, an acquired PI3K/AKT/mTOR pathway-mediated resistance to the MEK inhibitor emerged in KIAA1549-BRAF fusion cells, whereas the combination treatment of MEK and mTOR inhibitors enhanced cell toxicity in vitro and in vivo [27].

NF1 mutation targeting in pediatric gliomas

The NF-1 gene encodes neurofibromin, a negative regulator of Ras activity, and loss of NF-1 leads to Ras dysregulation and activation of downstream pathways, including MAPK and PI3K pathway, as well as tumorigenesis [59]. Initial preclinical experiment demonstrated that an MEK inhibitor sensitized an NF-1-deficient acute myeloid leukemia model [41]. Of note, the MEK1/2 inhibitor, selumetinib, decreased tumor size in 71% (17/24) of NF-1-related pediatric plexiform neurofibromas [16]. Therefore, Ras–MAPK signaling may be a novel therapeutic target in gliomas carrying the NF-1 mutation. See et al. demonstrated that NF-1 deficiency is associated with MEK inhibitor sensitivity in GBM cell lines and an MEK inhibitor (PD0325901) suppressed NF-1-deficient GBM orthotopic xenograft formation. In addition, they found that inhibition of PI3K/AKT/mTOR signaling enhanced the sensitivity to MEK inhibition [67]. In the NCT01089101 trial, 10 of 25 (40%) NF-1-associated LGGs under treatment with selumetinib achieved partial response, with a 2-year PFS of 96%. Further large-scale studies are required to establish the efficacy of MEK inhibition in NF-1-deficient PLGG.

In addition, it has been demonstrated that the NF-1-related mTOR pathway regulates glial cell growth and gliomagenesis [2]. Kaul and colleagues reported that the growth of NF-1-deficient optic glioma is suppressed by both AKT- or MEK-mediated mTOR inhibition in vitro and in vivo [35]. Moreover, rapamycin suppressed the proliferation of NF-1+/-GFAPCKO mice optic glioma [24]. Interestingly, when an mTOR inhibitor and the tyrosine kinase inhibitor, erlotinib, were tested on recurrent PLGG, two NF-1 patients had disease control for more than 1 year [73]. A phase-2 study employing everolimus for recurrent or progressive low-grade gliomas, with or without NF-1 deficiency, is currently recruiting patients (clinicaltrials.gov, NCT01734512).

Other molecular targets in pediatric gliomas

A breakthrough in molecular target therapy has been the employment of everolimus for tuberous sclerosis-associated SEGA. SEGAs are the most common brain tumors in TSC and occur in 5–20% of these patients. Mutations in TSC1 (hamartin) or TSC2 (tuberin) upregulate the mTOR complex 1, leading to cell proliferation in SEGA, raising the possibility that mTOR inhibitor may be a specific therapeutic agent. Several case reports demonstrated that rapamycin caused the regression of TSC-related astrocytomas [19, 37]. Importantly, large clinical trials (ClinicalTrials. gov number, NCT00789828, NCT00411619, and NCT01713946) indicated that everolimus reduced tumor volume and seizure frequency in patients with TSC-associated SEGA [18, 20, 39]. Additionally, treatment with everolimus provided long-term clinical benefit to patients with SEGA [17].

Furthermore, recurrent genetic alterations, such as those involving NTRK, FGFR1, and MYB, were identified in PLGG and are promising therapeutic targets [29, 56, 76]. Receptor tyrosine kinase tropomyosin-related A, B, and C (TRKA, TRKB, and TRKC) are encoded by NTR1, NTRK2, and NTRK3, respectively. These NTRK oncogenic fusions lead to downstream activation of signaling pathway, including the PI3K pathway and MAPK pathway [36].

Various inhibitors targeting TRK family members have been developed and tested in clinical trials. In 2018, the U.S Food and Drug Administration (FDA) approved the use of larotrectinib for solid tumors containing NTRK gene fusions. Importantly, larotrectinib led to dramatic tumor regression in a pediatric HGG patient with ETV6-NTRK3 fusion [78]. Clinical trials testing larotrectinib for NTRK fusion-positive solid tumors, including brain tumors, are currently ongoing (NCT02576431). Further studies are required to assess the therapeutic potential of TRK inhibition in PLGG.

Recurrent FGFR1 mutations (hotspot residues at N546 and K656), as well as PTPN11 mutations, were identified in a subset of PAs [29]. Moreover, duplication of the FGFR1 tyrosine kinase domain and FGFR1-TACC fusion were identified in PLGG [76]. These mutations were found to be mutually exclusive with RAF/RAS alterations and also result in MAPK pathway activation [29]. Anti-FGF/FGFR targeting drugs have been introduced and are being tested in clinical trials [6, 15].

In summary, large-scale genomic analyses uncovered the genetic landscape of pediatric low-grade gliomas, which are distinct from adult gliomas. Since most of the PLGGs harbor gene alterations that activate the MAPK pathway, the efforts to develop a therapeutic strategy targeting this pathway are intensifying. The results of several ongoing clinical trials are expected to improve PLGG patients’ outcome by novel molecular target therapies.