MEK and RAF inhibitors: time for a paradigm shift in the treatment of pediatric low-grade gliomas?

See article in this issue by Sun et al., pp. 774–785.

Pediatric low-grade gliomas (PLGGs), the most common group of brain tumors in children, are molecularly and clinically heterogeneous with a propensity for indolent progression, particularly if unresectable.¹ Although radiotherapy is effective, with a 5-year event-free survival of 87%, its sequelae (eg, neurocognitive, neuroendocrine dysfunction, vasculopathy, ototoxicity, secondary malignancies) can prove debilitating.² Thus, chemotherapy remains the mainstay of treatment for younger children. However, 5-year event-free survival in children with non-neurofibromatosis (NF)-associated LGG receiving standard front-line chemotherapeutic regimens is approximately 40%, reflecting the chronic nature of this disease and the multiple therapeutic interventions often required to achieve the 5-year overall survival of 85%.¹

PLGGs have a median of 1 coding and protein-changing mutation per tumor and generally demonstrate activation of the mitogen-activated protein kinase (MAPK) pathway.¹ Approximately 15% of NF type 1 (NF1) patients develop PLGG, in which loss of neurofibromin 1, a negative regulator of RAS, results in constitutive activation of the RAS/RAF/MAPK pathway.³ In non-NF1-associated PLGG, the most common genetic aberrations affect v-raf murine sarcoma viral oncogene homolog B1 (BRAF), a serine/threonine-specific protein kinase and key regulator of the MAPK pathway. Up to 90% of sporadic pilocytic astrocytomas (PAs) harbor a BRAF-KIAA1549 fusion gene (or fusions with alternative partners),³ which all lead to the deletion of the auto-inhibitory BRAF protein domain, resulting in constitutively activated BRAF dimers. The BRAFV600E mutation functions as a constitutively activated BRAF monomer and has been reported in approximately 10% of PA, but more commonly supratentorial PAs, diffuse astrocytomas (DAs), pleiomorphic xanthoastrocytomas, and gangliogliomas. Other alterations harbored by PLGGs include myeloblastosis/myeloblastosis oncogene-like 1 rearrangements, fibroblast growth factor receptor 1 mutations and rearrangements, cyclin-dependent kinase inhibitor 2A deletions, neurotrophic tyrosine receptor kinase 2 and 3 fusions, and, rarely, isocitrate dehydrogenase mutations.³

In this issue of Neuro-Oncology, Sun and colleagues report promising in vitro and in vivo activity of MLN2480, a brain-penetrant RAF dimer type II antagonist, in PLGG models.⁴ The BRAF catalytic domain can undergo a conformational change, with the active state assuming an N-(1-deoxy-D-fructos-1-yl)-glycine (DFG) conformation of “DFG in” and an inactive state assuming “DFG out.” Vemurafenib and dabrafenib, targeting monomeric forms of BRAF, are type I antagonists that bind the active or DFG-in conformation of the BRAF catalytic domain.⁴ MLN2480 is a type II inhibitor that preferentially binds to the DFG-out conformation and targets both monomeric and dimeric forms of BRAF. Having developed “pathway-relevant” BRAF mutant polymorphous low-grade adenocarcinoma (PLGA) models by transducing tumor protein (TP)53-null neural progenitors with KIAA1549:BRAF fusion or BRAFV600E expression vectors (although TP53 is typically not mutated in PLGG), Sun et al report that MLN2480 has equipotent activity against BRAFV600E, KIAA1549:BRAF, and other BRAF oncoproteins that function as dimers, and directly engages with BRAF in both BRAFV600E- and KIAA1549:BRAF-expressing neurospheres; in contrast, type I RAF inhibitors have much less activity on KIAA1549:BRAF cells. After oral drug administration in mice, MLN2480 demonstrates good blood–brain barrier (BBB) penetration (plasma:brain ratio 24%, AUC:brain 41.6h.µM). Confirmatory experiments including matrix-assisted laser desorption/ionization–mass spectrometry imaging demonstrate drug/metabolite localization in tumor and healthy mouse brain. MLN2480 treatment leads to phosphorylated extracellular signal-regulated kinase (ERK) suppression in tumor tissue and healthy brain and inhibits growth of primary neurosphere PA cultures with truncation/fusion BRAF mutations without triggering paradoxical activation or “rebound” ERK signaling in neural progenitors transformed with either BRAF oncoprotein. Indeed, MLN2480 circumvents paradoxical activation by binding with equal affinity to monomeric and dimeric BRAF subunits.⁴

The standard front-line chemotherapy regimen for PLGA consists of carboplatin ± vincristine, with a 5-year progression-free survival (PFS) of 39%.¹ Weekly vinblastine has a 5-year PFS of 53.2%.¹ Second-line chemotherapeutic regimens including vinblastine, temozolomide, and avastin + irinotecan have 2-year PFS ranging from 47% to 61%.¹ The prevalence of activated BRAF alterations in PLGG has made BRAF and other members of the MAPK pathway (eg, mitogen/extracellular signal-regulated kinase [MEK]1/2, ERK1/2) attractive drug targets, especially in the context of this single-pathway disease. The study by Sun et al⁴ addresses 3 concerns that have troubled investigators developing these agents in PLGG: (i) tailoring therapy by understanding distinct mechanism of action on the mutant kinase, (ii) delineating agents’ BBB penetration and its impact on efficacy, and (iii) potential for acquired resistance and secondary squamous-cell carcinomas with RAF inhibitors due to rebound signaling or paradoxical activation of wild-type RAS kinase dimers in cells with RAS activity.

Understanding the Mechanism of Action

Therapies targeting BRAF must be specific to the distinct mechanisms of action on the mutant kinase; the fusion kinase in cells expressing KIAA1549-BRAF functions as a homodimer, rendering cells resistant to BRAF inhibition while exhibiting CRAF-independent paradoxical activation of MAPK signaling. Recent studies have confirmed that all adenosine triphosphate (ATP)–competitive RAF inhibitors such as vemurafenib, dabrafenib, and sorafenib lead to paradoxical activation of the MAP kinase pathway in BRAF wild-type cells, including cells harboring BRAF fusions.⁵ A phase II study of sorafenib, a multi-kinase inhibitor of BRAF, vascular endothelial growth factor receptor, platelet derived growth factor receptor, and c-Kit in children with progressive PLGG was prematurely halted because of unexpected progression among 9 of 11 patients, including 3 each with KIAA1549-BRAF and NF1.⁶ In vitro studies confirmed that this effect was related to paradoxical ERK activation, also reported in NF1 loss. Thus, PA, the most common PLGG, which commonly harbors the KIAA1549-BRAF fusion, cannot be treated with currently approved RAF inhibitors. In contrast, MLN2480’s equipotent activity on both BRAF oncoproteins gives this agent a clear advantage over type 1 RAF inhibitors. An alternative and promising class of agents are MEK inhibitors, which inhibit downstream of RAF and should be effective in PLGAs harboring either BRAF aberration.

Impact of BBB Penetration on Efficacy

The combination of dabrafenib and trametinib and single agent vemurafenib are FDA approved for BRAF-mutated metastatic melanoma. Despite preclinical evidence of limited BBB penetration, patients with BRAFV600E-mutant metastatic melanoma with CNS metastases experienced impressive (39%) intracranial response rates with dabrafenib.⁷ Similarly, early results from PLGG studies using type I RAF inhibitors (dabrafenib, vemurafenib) and MEK inhibitors (selumetinib) have shown promising activity.¹ Dabrafenib, an inhibitor of RAF kinase mutants BRAFV600E and BRAFV600K, demonstrated a 32% objective response rate in PLGG with BRAFV600E mutations; no squamous-cell carcinoma or keratoacanthoma was reported. Selumetinib, a non-ATP-competitive small-molecule inhibitor of MEK1/2, is undergoing phase II trials in PLGG in the Pediatric Brain Tumor Consortium. In the phase I study in children with progressive PLGG, without RAF status stratification, 25 of 38 patients were treated at the recommended phase II dose (RP2D). Among 5 (20%) patients with sustained, centrally reviewed partial responses (PRs) at the RP2D, 4 had BRAF alterations.⁸ The open phase II study has strata for BRAF-altered PA, optic-pathway gliomas, NF-associated LGG, and BRAF-altered non-PA LGG. Recently, Dombi et al reported that 71% of patients with NF1 related plexiform neurofibromas treated with selumetinib experienced PRs without disease progression in any patients.⁹ Finally, phase I trials of trametinib with or without dabrafenib are also under way in children with BRAFV600E-mutated LGG. Thus, the promising early activity reported with RAF inhibitors in PLGG may be explained by the typically enhancing nature of PLGG and the fact that the BBB is unlikely to be intact after prior surgeries and therapies. Regardless, introducing a BBB-penetrant type II RAF inhibitor would be a welcome and, potentially, more effective treatment option.

Potential Development of Acquired Resistance or Secondary Squamous-Cell Carcinomas

In melanoma, resistance to RAF or MEK inhibitor therapy occurs through upregulation of bypass pathways (eg, through receptor tyrosine kinases), acquisition of de novo neuroblastoma-RAS or MEK mutations, dimerization, or variant splicing of mutant BRAFV600.¹⁰ Acquired resistance to RAF and MEK therapies and secondary squamous-cell carcinomas (also caused by paradoxical activation of the MAPK pathway in BRAF wild-type cells) have not been reported in the small series of children with prolonged treatment to date. In contrast, approximately 20% of melanoma patients treated with RAF inhibitors develop squamous-cell carcinomas,¹⁰ raising particular concerns in PLGG, which requires chronic or intermittent treatment for maximal benefit. In contrast, MLN2480 does not trigger paradoxical activation or “rebound” ERK signaling and, thus, will be less likely to lead to squamous-cell carcinomas or acquired resistance (if relevant for PLGG).

Type II RAF inhibitors, such as MLN2480, are promising agents for PLGG patients, regardless of the type of BRAF alteration, based on the potentially lower likelihood of resistance induction and the increased BBB permeability. However, it remains to be seen how activity in patients compares with MEK inhibitors, how relevant resistance induction is in PLGG, and to what extent increased BBB penetration may potentiate neurological toxicities or improve outcome.