Fusing the Genetic Landscape of Infantile High-Grade Gliomas

Infantile high-grade gliomas (iHGG) are relatively uncommon but aggressive primary brain tumors that occur in very young children, and there is currently little efficacy of nontargeted chemotherapy and radiation (1, 2). To develop more targeted therapies, the molecular underpinnings of iHGG must be understood. Previously, it has been established that a subset of these iHGG harbor mutations in the histone H3–associated genes H3F3A and HIST1H3B/C. The genetic drivers of histone–wild-type iHGG have not been well characterized. In this issue of Cancer Discovery, Clarke and colleagues report multi-omic profiling data from the largest multi-institutional cohort of iHGGs (3). To focus their analysis on cerebral high-grade gliomas, an initial cohort of 241 infantile (<4 years of age) patients were screened and cases were excluded from further analysis based on tumor location and molecular resolution into other well-established diagnostic entities, such as pilocytic astrocytoma and ependymoma. This filtering led to molecular characterization of 130 iHGGs. Occasional mutations were found in genes involved in the MAPK pathway, including the BRAF^V600E mutation in rare cases. Half of these cerebral iHGGs were found to harbor recurrent gene fusions. Gene fusion partners included either a receptor tyrosine kinase (RTK; ALK, ROS1, MET) or a neurotrophic RTK (NTRK1-3; Fig. 1). The results presented by Clarke and colleagues dramatically increase our molecular understanding of iHGG, expanding upon a recent separate multi-institutional study involving low-grade and high-grade gliomas in the infant population (4).

Gene fusions are frequent events in pediatric cancers (5). In addition to the ALK, MET, ROS1, and NTRK receptor gene fusions described in infantile and pediatric gliomas, gene fusions have been identified in pediatric brain tumors (summarized in Fig. 1) including low-grade gliomas (KIAA1549–BRAF), supratentorial ependymoma (C11ORF95–RELA, YAP1–MAMLD1) and a subgroup of NF2–wild-type meningioma (YAP1–MAML2), as well as several peripheral pediatric cancers, such as Ewing sarcoma (EWS–ETS fusions, most commonly EWS–FLI1) and anaplastic large cell lymphoma (ALK gene fusions; refs. 6–8). Most of these tumor types harbor relatively few additional mutations, making these gene fusions the likely tumor-initiating events and attractive therapeutic targets.

Cancer-associated gene fusions can be categorized on the basis of whether they contain gain-of-function mutations leading to the formation of an oncogene, or if they lead to the loss-of-function of a tumor suppressor gene. Gain-of-function gene fusion products are further characterized based on the function of the involved proteins—receptor tyrosine kinases that are upstream of mitogenic pathways (i.e., ALK or NTRK fusions), signal transducer protein kinases (i.e., BRAF fusions), or transcription factors/transcriptional activators (TF/TA) that are downstream effectors of mitogenic pathways (i.e., RELA or YAP1 fusions).

The first category of gene fusion involves the fusion between the 3′ region of a receptor tyrosine kinase (retaining the kinase domain), such as ALK, ROS1, MET, NTRK1–3, or FGFR, and the 5′ region of any one of several other genes, such as those identified in infantile high- and low-grade glioma by Clarke and colleagues and others. The activity of RTKs is normally tightly regulated, and under normal physiologic conditions these receptors signal only upon binding to their natural ligands. The truncation of the regulatory domains located in the 5′ regions of RTKs leads to a dysregulation of RTK activity. Another type of gene fusion that lead to an activation of RTK pathways involves signal transduction protein kinases (i.e., BRAF) that are downstream of the actual RTKs and transduce the mitogenic signals upon receptor–ligand binding. These fusions lead to a dysregulation of BRAF activity by placing the 3′ BRAF kinase domain behind another 5′ gene, thereby removing the autoinhibitory domain of BRAF. These gene fusions somewhat mirror activating point mutations (i.e., BRAF^V600E) and both types of alterations are found in the same tumor types in a mutually exclusive manner. Both fusion types lead to constitutively active mitogenic signaling, independent of receptor–ligand binding, which in turn leads to constitutively activated mitogenic signaling pathways (i.e., MAPK and AKT pathways), cellular overgrowth, and ultimately cancer.

The second category of gain-of-function gene fusions involves putative TFs/TAs, such as C11ORF95–RELA, YAP1–MAMLD1/MAML2, SS18-SSX1/2, and EWS–ETS, that induce an oncogenic program by directly binding to a plethora of genes and activating their expression. Under normal physiologic conditions, the activity of these wild-type TFs/TAs is tightly controlled by upstream pathways (e.g., by phosphorylation, nuclear localization, or degradation), whereas the activity of TF/TA fusion proteins might be independent of these signals. The mechanisms of how exactly these gene fusion events lead to a deregulation and/or hyperactivation of these TFs/TAs (compared to the wild-type versions) are poorly understood to date. Likewise, little is known about the how functions of these fusion proteins can be therapeutically targeted.

In contrast to the two categories of gain-of-function fusion types that lead to the formation of a potent oncogene, a third category of gene fusions involves the loss-of-function of tumor suppressor genes, i.e., by truncation and or frame shift of the tumor suppressor gene (TRMT11–GRIK2 and mTOR–TP53BP1, for example). Whereas gain-of-function gene fusions are likely strong oncogenic drivers (occurring in tumors with an otherwise relatively silent genome) and tumor-initiating events (exemplified by their ability to induce tumor formation when expressed in mice), it is unknown if loss-of-function gene fusions are driver mutations (i.e., sufficient for tumorigenesis) or simply contribute to aggressive behavior.

Although Clarke and colleagues detected multiple variants of each RTK fusion type (with different 5′ fusion partners) in iHGG, some 5′ fusion partners were more common and identified in several samples or as part of different RTK fusions (such as ETV6 and EML4). In addition, the same RTK gene fusions have been identified in other pediatric and adult cancers, such as secretory breast carcinoma, congenital fibrosarcoma, congenital mesoblastic nephroma, colon carcinoma, and non–small cell lung carcinoma. It is unclear why certain variants of RTK gene fusions are more common than others and how the choice of 5′ fusion partner affects disease behavior. ETV6 itself is a tumor suppressor and loss thereof has been observed in several cancers. Therefore, in addition to dysregulation of NTRK3 kinase activity, loss of ETV6 function might contribute to the oncogenic functions of ETV6–NTRK3. Another ETV6 fusion gene, ETV6–RUNX1, occurs in around 25% of pediatric acute lymphoblastic leukemia (ALL) cases and is characterized by a similar ETV6 breakpoint compared with the iHGG ETV6–NTRK3 fusion identified by Clarke and colleagues. Importantly, up to 90% of analyzed ETV6–RUNX1 samples also experienced loss-of-heterogeneity of the nontranslocated ETV6 allele (9).

Using two types of genetically engineered mouse models, Clarke and colleagues subsequently demonstrated that the expression of PPP1CB–ALK is sufficient to cause the formation of tumors that histologically resemble human glioblastoma, demonstrating that these RTK gene fusions are the likely tumor-initiating events in these cancers. They then used their preclinical model to test the efficacy of different ALK inhibitors in vitro (using cells from disseminated PPP1CB–ALK-driven mouse tumors) and identified lorlatinib as the inhibitor with the greatest inhibitory effect. Subsequent in vivo studies, using an allograft system of short-term cultured PPP1CB–ALK tumor cells transplanted into CD1 mice, showed that lorlatinib treatment was superior to both vehicle and temozolomide treatment and led to tumor regression during treatment duration. Beyond preclinical models, ALK inhibitor therapy stabilized disease in a one-month-old boy following failed initial temozolomide chemotherapy for iHGG. These results provide hope for future clinical ALK inhibitor efficacy in fusion-positive infant high-grade gliomas, which has been shown to be useful in treating adult ALK fusion–positive non–small cell lung carcinoma (10).

Furthermore, Clarke and colleagues established patient-derived in vitro cultures from two NTRK fusion–positive iHGG samples (TPM3–NTRK1 and ETV6–NTRK3) and found that treatment of these cells with small-molecule inhibitors of tropomycin receptor kinase (TRK; entrectinib, crizotinib, or milciclib) led to a significant growth inhibition that was accompanied by a reduction in AKT and ERK signaling. Finally, treatment with TRK inhibitors stabilized disease in two patients diagnosed with ETV6–NTRK3-positive infantile gliomas. The first was a girl diagnosed with congenital glioblastoma who initially received multiple chemotherapeutic agents before undergoing subtotal resection. Four months post-surgery she was started on crizotinib treatment, which led to a 56% reduction of solid tumor mass. Addition of larotrectinib therapy increased tumor reduction to a total of 73%. The second patient had a pontine low-grade neuroepithelial neoplasm discovered at 11 months of age and underwent gross total resection. The patient developed two recurrences and underwent chemotherapy and a second gross total resection. Following the second recurrence, the patient received larotrectinib therapy and has remained without recurrence for 12 months.

Overall, the mutational and experimental data from this comprehensive study has implications for diagnosis, prognosis, and therapeutic management of iHGG. It is likely that the fusions and other alterations present in histone–wild-type iHGG will lead to a widespread reclassification of iHGG based on molecular subtyping. Incorporating molecular profiling into any clinical trials or routine clinical practice will allow for the development of molecular subtype–specific therapeutic strategies, which are desperately needed for these types of tumors.

No potential conflicts of interest were disclosed.