An Overview of Optic Pathway Glioma With Neurofibromatosis Type 1: Pathogenesis, Risk Factors, and Therapeutic Strategies

Abstract

Optic pathway gliomas (OPGs) are most predominant pilocytic astrocytomas, which are typically diagnosed within the first decade of life. The majority of affected children with OPGs also present with neurofibromatosis type 1 (NF1), the most common tumor predisposition syndrome. OPGs in individuals with NF1 primarily affect the optic pathway and lead to visual disturbance. However, it is challenging to assess risk in asymptomatic patients without valid biomarkers. On the other hand, for symptomatic patients, there is still no effective treatment to prevent or recover vision loss. Therefore, this review summarizes current knowledge regarding the pathogenesis of NF1-associated OPGs (NF1-OPGs) from preclinical studies to seek potential prognostic markers and therapeutic targets. First, the loss of the NF1 gene activates 3 distinct Ras effector pathways, including the PI3K/AKT/mTOR pathway, the MEK/ERK pathway, and the cAMP pathway, which mediate glioma tumorigenesis. Meanwhile, non-neoplastic cells from the tumor microenvironment (microglia, T cells, neurons, etc.) also contribute to gliomagenesis via various soluble factors. Subsequently, we investigated potential genetic risk factors, molecularly targeted therapies, and neuroprotective strategies for tumor prevention and vision recovery. Last, potential directions and promising preclinical models of NF1-OPGs are presented for further research. On the whole, NF1-OPGs develop as a result of the interaction between glioma cells and the tumor microenvironment. Developing effective treatments require a better understanding of tumor molecular characteristics, as well as multistage interventions targeting both neoplastic cells and non-neoplastic cells.

Optic pathway gliomas (OPGs) are the most predominant pilocytic astrocytomas (PAs), classified as World Health Organization (WHO) grade I, constituting up to 7.0% of all gliomas.^1,2 OPGs are most commonly found in patients with neurofibromatosis type 1 (NF1), the most common tumor predisposition syndrome. According to a retrospective study, 59% of the 133 patients with OPGs are diagnosed as NF1.³ Meanwhile, among children with NF1, OPGs are also the most predominant central nervous system (CNS) tumors, with an estimated prevalence of 15% to 20%.^4–7 Generally, NF1-associated OPGs (NF1-OPGs) are typically diagnosed in children younger than 7 years old (mean of 4.5 years old) and affect the optic pathway from the optic nerves (ONs), optic chiasm, and optic tracts to optic radiations.^8–11 Specifically, 75% to 85% of OPGs in individuals with NF1 occur in the ONs and chiasm (anterior optic pathway).¹² A retrospective study showed that approximately 34.8% of patients with NF1-OPGs have bilateral gliomas.¹⁰ There are various distinctions between NF1-OPGs and sporadic OPGs despite their common pathologic classification. Sporadic OPGs are more likely to affect chiasmal and postchiasmal lesions, which exhibit more severe clinical symptoms and require more clinical intervention.^13–17 Without NF1 loss, most sporadic OPGs are characterized by activation of BRAF via fusion with KIAA1549, which activates downstream pathways.^18–21 

Although many NF1-OPGs tend to be indolent low-grade gliomas (LGGs), 30% to 50% of patients have visual disturbance owing to compression of the optic nerve^5,22,23 or precocious puberty caused by hypothalamic involvement.^24,25 The deterioration of vision usually manifests as a progressive visual decline or normal visual symptoms with anomalous ophthalmological tests.²⁶ Ophthalmological signs of OPGs range from optic nerve edema, optic nerve pallor, optic nerve atrophy, and/or proptosis.⁴ Notably, although toddlers may suffer from deteriorative vision loss, they seldom complain about it.²⁶ Hence, early screening and risk assessments are essential. A series of clinical studies have found that in patients who are younger than 2 years old^27–29 and girls^27,30 as well, as having tumors located in the postchiasmal optic pathway^{9,10,22,27,29} or hypothalamus,²⁵ NF1-OPGs may produce more aggressive symptoms and have worse prognoses. However, more accurate prognostic risk factors are needed for precision medicine strategies for related patients. 

As mentioned above, most of NF1-OPGs are indolent and asymptomatic. Hence, the main objective of therapy is to reduce the risk of a permanent, clinically significant visual impairment.³¹ Therapy intervention is frequently provided for symptomatic children who have worsening vision loss (a 2-line reduction in visual acuity).^4,32 The role of surgical resection is limited due to the danger of irreversible nerve damage. Secondary neoplasms³³ and radiotherapy-related vasculopathy³⁴ also prevent radiotherapy from being used. Therefore, chemotherapies, such as carboplatin and vincristine combined,^35–37 are the most common effective treatments in symptomatic children. Although chemotherapy can successfully attenuate tumor progression, it is estimated that nearly 30% of affected children remain visually impaired after treatment.^27,38 Therefore, there is an urgent need to design new therapeutic strategies to prevent vision deterioration and improve visual outcomes in patients with NF1-OPGs before irreversible vision deficits occur. In this review, we summarized the latest advances in NF1-OPGs in terms of activated Ras effector pathways in optic glioma stem cells (o-GSCs), astrocytes and retinal ganglion cells (RGCs) as well as complex interactions within the tumor microenvironment, aiming to identify potential risk assessment algorithms and therapeutic modalities for affected children. 

Activated Ras Effector Pathways Mediate Glioma Pathogenesis

Located on chromosome 17q11.2, NF1 is a major tumor suppressor gene with 57 exons encoding a protein product of 220 to 250 kDa called neurofibromin.^39–42 Neurofibromin acts as a negative operator of the p21-Ras proto-oncogene by converting the active GTP-bound Ras proto-oncogene into an inactive GDP-bound state.^26,42–45 Based on the two-hit hypothesis of tumorigenesis,⁴⁶ young patients with NF1 are born with one nonfunctional NF1 allele (germline NF1 gene mutation) in cells throughout the body that are heterozygous for NF1 (NF1^+/−) but form gliomas only upon somatic mutation of another functional allele in stem cells.^26,47,48 Whole-genome sequencing (WGS) analysis of NF1-PA samples revealed that loss of heterozygosity (LOH), frameshift mutations, and methylation were all possible mechanisms of somatic NF1 gene silencing.⁴⁹ Deprivation of neurofibromin in NF1-related gliomas leads to increased Ras activation,^50–52 which plays a critical role in promoting cell development and survival in various tissues, including the developing brain of mammals.^53,54 

There are 3 distinct Ras effector pathways: the phosphatidylinositol-3 kinase (PI3K)/protein kinase-B (AKT)/mechanistic target of rapamycin (mTOR) pathway,^51,52,55 the mitogen activated protein kinase (MEK)/mitogen activated protein kinase (ERK) pathway,^55–57 and the cyclic adenosine monophosphate (cAMP) pathway.^58,59 Upregulated Ras downstream pathways have distinct effects on diverse cells, such as o-GSCs, astrocytes, and RGCs, to determine tumorigenesis (Fig. 1). 

Figure 1.

 

Ras signaling pathway. Deprivation of neurofibromin in NF1-related OPGs led to activate Ras and its 3 effector pathways. Nf1^−/− astrocytes displayed both upregulated MEK-ERK and PI3K-AKT manners, which individually contributed to activating mTOR activity and promoting tumor proliferation. Specifically, PI3K/AKT manner promoted the proliferation of o-GSCs and MEK/ERK pathway determined the progenitors multilineage differentiation. Besides, neurofibromin decreased the levels of cAMP in Nf1^−/− astrocytes to positively accelerate tumor development while downregulated cAMP levels attenuated Nf1^+/− RGC growth through PKCζ, which was related with visual dysfunction. This figure was created using the BioRender software.

Optic Glioma Stem Cells

A widespread mechanism behind pediatric brain cancers is stalled developmental programs, resulting in the development of stem or progenitor populations to abnormally maintain their early stem-like properties.⁶⁰ The rarity of NF1-OPGs in adults suggests that there might be transient neural progenitor cells (NPCs) susceptible to tumorigenesis through loss of neurofibromin during specific developmental periods.⁴⁸ Researchers have discovered that o-GSCs are likely to reside in the third ventricular zone (TVZ) rather than the lateral ventricular subventricular zone (lv-SVZ).^61,62 There may be a potential correlation between the specific anatomic location of putative o-GSCs and their impact on normal ONs growth. During embryonic development, astrocytes are produced locally from astrocyte precursor cells (APCs).^63,64 Oligodendrocytes are derived from brain-derived oligodendrocyte precursor cells (OPCs) that arrive at the distal ONs at birth.⁶⁴ Migrating OPCs are also referred to as migrating glial progenitors (GPs), owing to their role in differentiating into type 2 astrocytes.⁶³ As early as the late embryonic stages, migratory GPs were thought to originate in radial glia (RG) in the TVZ.^65–68 

Following Nf1 loss, overactive Ras effector pathways interrupt the equilibrium of original cell maintenance in the TVZ and promote abnormal persistence of NPCs.^65,69 Specifically, the PI3K/AKT pathway promoted the proliferation of progenitors, and the MEK/ERK pathway regulated progenitor multilineage differentiation.⁷⁰ When ERK was hyperactivated, NPCs from the hypothalamic ventricular zone (hVZ) migrated into the hypothalamic mantle zone (hMZ) and failed to differentiate normally.⁶² Under MEK/SMAD family member 3 (Smad3)-dependent signaling, hairy and enhancer of split 1 (Hes1) mediate neuron and astrocyte differentiation, whereas oligodendrocyte lineage transcription factor 2 (Olig2) induces oligodendrocyte differentiation.⁷⁰ In contrast, APCs in the ONs were independent of Nf1 loss, so they might not be the source of NF1-OPGs.⁶² In addition, migrating GPs induced the formation of OPGs in the neonatal ONs, leading to RGC apoptosis in a BCL-2-associated X protein (Bax)-dependent manner.⁶² Transient neonatal pharmacological inhibition with MEK or genetic loss of Mek1/Mek2 attenuated RGC loss and axon degeneration, which inhibited OPGs formation before irreversible neurological damage occurred.⁶² 

Various Cre driver lines have been applied to target somatic Nf1 loss in different populations of NPCs in genetically engineered mice (GEM) to identify the specific o-GSCs of NF1-OPGs. Previous studies incorporated Cre-loxp strategies and revealed that Nf1^+/− mice with conditional Nf1 mutations in Gfap⁺,⁷¹ CD133⁺ (Prom1),⁷² or Blbp⁺⁵⁸ progenitors all formed OPGs by 3 months of age. In contrast, OPGs were not generated by somatic Nf1 loss in Ng2⁺ (nerve/glial antigen 2) cells during the embryogenic period⁷³ or in GFAP⁺ astrocytes after birth.⁶⁵ Consistent with the above results, further investigations identified 3 NPCs within the TVZ and revealed that Nestin⁺ Blbp⁺ CD133⁺ cells are the most likely source of murine Nf1 optic gliomas. They are the most clonogenic, proliferative, and abundant population during embryogenesis and then dramatically decline during early postnatal life.⁶¹ When multiple original cells have unique differentiation pathways to generate gliomas, transforming events often produce different tumor latencies and partially determine glioma biology. For example, Nf1 loss in Olig2⁺ OPCs postponed optic glioma latency to 6 months.⁷² Whole-tumor RNA-sequencing analyses also confirmed that FMC (Nf1^flox/mut; GFAP-Cre) and FMOC (Nf1^flox/mut; Olig2-cre) optic gliomas were completely distinct entities.⁷⁴ 

Somatic Nf1 mutation during a restricted developmental window was also critical to OPG gliomagenesis in children. Using the Nf1^flox/null, Prom1-cre mouse model,⁷⁵ studies have shown that murine Nf1 optic glioma formation requires somatic Nf1 loss during late embryonic development because postnatal loss within the same o-GSCs does not cause tumorigenesis.⁷² It was also mentioned above that progenitors within the TVZ were most plentiful during embryogenesis and sharply disappeared by postnatal day 1. In human TVZ, proliferation of progenitors also decreased with age,⁶⁹ which might explain the NF1-OPG development in young children. There was a combination of factors that contributed to the unique patterning and penetrance of Nf1 optic gliomas, including the regulation of overactivated Ras signaling, composition of o-GSCs in TVZ, changes in progenitor proliferation over time, and the requirement for somatic Nf1 gene inactivation during embryonic development. 

Astrocytes

Nf1 ^−/− astrocytes displayed both upregulated MEK/ERK and PI3K/AKT signaling, which individually contributed to the activation of mTOR and promotion of tumor proliferation.⁵⁵ In this regard, MEK/ERK activates mTOR by p90RSK, independent of the tuberous sclerosis complex (TSC)/Ras homolog enriched in the brain (Rheb).^55,76 The inhibition of PI3K, MEK, or mTOR prevented Nf1 optic glioma development and decreased tumor volumes in vivo.^55,77 Unfortunately, inhibition of these pathways was only effective during treatment, and tumors proliferated to pretreatment levels after the treatment had ended.⁷⁷ Therefore, it is possible that combining treatments targeting mTOR with different effector arms may yield longer lasting improvements in OPGs outcomes.⁵⁵ 

Compared to the above two signaling pathways, neurofibromin decreased the level of cAMP in Nf1^−/− astrocytes to positively accelerate tumor development.^78–80 A reduction in tumor proliferation could be observed by increasing intracellular cAMP levels with the phosphodiesterase-4A1 (PDE4A1) inhibitor rolipram.⁵⁹ Furthermore, when PDE4A1 was ectopically expressed in the cortex, it produced glioma formation in the brain regions where OPGs did not grow naturally.⁵⁹ Glioma progression can also be dramatically blocked by rolipram in Nf1 GEM in vivo. Therefore, it is possible to explain the pattern of NF1 gliomagenesis by differences in cAMP levels between brain regions. 

Retinal Ganglion Cells

In contrast to its positive role in Nf1-deficient astrocytes, impaired neurofibromin function downregulates cAMP levels to attenuate Nf1^+/− RGC growth through the atypical protein kinase C-zeta (PKCζ), which is related to visual dysfunction.^81,82 In Nf1­mutant mice, it was shown that a combination of progressively damaged RGC axons,⁸³ accumulated RGC death,⁸² and reduced visual acuity³⁰ led to vision loss. There was evidence that inhibition of Ras activity (lovastatin) or upregulation of cAMP levels (rolipram) can inhibit RGC apoptosis in mice with optic gliomas.^82,84 

Tumor Microenvironment Permissiveness Contributes to Gliomagenesis

The tumor microenvironment plays a significant role in the formation and development of tumors, holding both positive and negative impacts on tumor cell growth.^85–87 The relevant non-neoplastic cells include microglia, T cells, neurons, endothelial cells, Müller cells, and so on, except fibroblasts and mast cells in both human NF1-related OPGs and Nf1 GEMs.^47,88 Previous studies have shown that homozygous inactivation of Nf1 in o-GSCs alone was insufficient for tumor formation. In contrast, in the context of the Nf1 heterozygous brain environment, Nf1 inactivation in o-GSCs led to the growth of OPGs by 3 months of age.^71,89,90 These observations suggested that nonneoplastic Nf1^+/− cells may have an essential impact on Nf1-null astrocyte growth, which provides the supportive environment that is needed for tumor growth. 

Microglia

Microglia have been commonly identified as major stromal constituents in gliomas, including PAs.^91–95 In a mouse model (Nf1^flox/mut; GFAP-Cre), increased Nf1^+/− microglia were identified as early as 3 weeks of age (WOA), before the formation of obvious tumors, which was consistent with a period of fast astrocyte proliferation in vivo.⁸⁹ The critical role of microglia in glioma formation has been identified by several independent approaches in Nf1 GEMs. First, Nf1^+/− microglia could promote the proliferation of Nf1^−/− astrocytes in vitro, and treatment with minocycline (a microglia inhibitor) dramatically inhibited mouse OPGs growth in vivo.⁹⁶ Second, neurofibromin mediated microglial properties by the c-Jun NH2-kinase (JNK)-relevant pathway involving mixed lineage kinase (MLK) and Ras-related C3 botulinum toxin substrate 1 (Rac1) manner, so that SP600125 (a JNK inhibitor) treatment could reduce optic glioma proliferation in vivo.⁹⁷ Further observation used CD11b-TK mice⁹⁸ to ablate resident brain microglia genetically by ganciclovir administration, which resulted in attenuated OPG development in vivo.⁹⁹ 

Further evidence revealed that o-GSCs expressed chemokine C-X3-C-motif ligand 1 (CX3CL1) to recruit microglia,¹⁰⁰ and microglial chemokine C-X3-C-motif receptor 1 (CX3CR1) was a chemokine receptor that regulated microglia migration.^101,102 Genetic CX3CR1 reduction can impair microglia dispersal, resulting in an accumulation of microglia in ONs⁹⁹ and postponed OPG formation¹⁰³ in vivo. In turn, to affect astrocyte proliferation and tumor progression, microglia incorporate additional molecules, including stromal cell-derived factor 1 (CXCL12), chemokine ligand 5 (Ccl5), meningioma-expressed antigen-5 (MGEA5), and interleukin-1β (IL-1β).^80,96 

CXCL12 could boost the proliferation of Nf1^−/− astrocytes and the apoptosis of wild-type astrocytes by sustaining the suppression of cAMP levels.⁸⁰ Whereas the injection of lentiviral-encoded CXCL12 into the cerebral cortex of mice with Nf1 optic gliomas resulted in a decreased tumor growth rate, AMD3100 (a CXCR4 inhibitor) treatment did not inhibit OPG growth.¹⁰⁴ This may be due to differences in the bioavailability of the CXCL12 used in that study and the counterregulatory mechanisms of CXCR4 between the optic pathway and the cortex. The existence of other cAMP-regulatory mechanisms might also neutralize the effects of CXCL12.¹⁰⁴ Ccl5 was uniquely expressed by microglia in 3 to 6 WOA for OPG maintenance.^105–107 Culminated Ccl5 acted in glioma cells through the AKT/glycogen synthase kinase-3 beta (GSK3β)/cyclic-AMP response binding protein (CREB) pathway to restrain stem cell apoptosis and boost tumor proliferation.^105,107,108 Administration of Ccl5 neutralizing antibodies prevented murine optic glioma proliferation and ameliorated retinal defects in vivo.¹⁰⁵ Importantly, accumulated Ccl5 expression had a negative correlation with survival in patients with gliomas.¹⁰⁸ In addition, the expression of MGEA5 by Nf1^+/− microglia in vivo was also enhanced, but future studies are needed to clarify its specific impact on the body. 

In Nf1 GEMs, female subjects displayed greater RGC reduction, a thinner retinal nerve fiber layer (RNFL; composed of RGC axons) and more decreased visual acuity than male subjects, despite identical tumor proliferation in both sexes.^30,109 It was demonstrated that estrogen induced microglia by estrogen receptor-β (ERβ), which was exclusively expressed in microglia. ERβ-activated microglia produce neurotoxins, such as IL-1β and other inflammatory cytokines, resulting in vision deterioration.^84,109 Pharmacologic inhibition of microglial ERβ reversed visual defects in female Nf1 GEMs.¹⁰⁹ There may be a potential model in which microglia play a positive role in glioma growth in both male and female Nf1 GEMs by producing nonsexually dimorphic paracrine factors, such as Ccl5,¹⁰⁷ whereas female gonadal sex hormones activate microglia, creating a neurotoxic microenvironment resulting in increased axonal damage and retinal dysfunction.¹⁰⁹ 

T Cells

It was found in 59 NF1 glioma samples that, in comparison with high-grade gliomas, NF1-LGGs exhibited more strongly enriched immune signatures, such as cytolytic T lymphocyte infiltration and mutation-derived neoantigens with enhanced HLA binding.⁴⁷ In addition, murine o-GSCs could not generate tumors in T-cell-deprived mice, indicating that the activation of T cells was necessary for Nf1 optic glioma proliferation in vivo.^106,110 Herein, CD8⁺ lymphocytes constituted the majority of T cells, and deprivation of CD8⁺ T cells inhibited OPGs growth in vivo.¹⁰⁸ 

Previous investigations have found that o-GSCs secrete Ccl2 to activate T cells by Ras-dependent signaling. T cells infiltrate glioma from the meningeal space in an integrin-dependent mechanism.^105,108 Then, activated T cells produce Ccl4 by the low-density lipoprotein receptor-related protein 1 (Lrp1)/calcineurin pathway,¹⁰⁸ promoting microglia to excrete Ccl5 by nuclear factor kappa-light-chain-enhancer of activated B cells (NFκB), which provides supportive microenvironment for o-GSC integration.^106,107 Notably, treatments with anti-Ccl2 antibody and anti-Ccl5 antibody abrogate OPG growth but make no difference in microglial contents and ON volumes.¹⁰⁵ It is likely that the interruption of the stromal axis hinders Nf1 optic glioma development but does not eliminate the initiation of glioma. 

Furthermore, there is an inverse epidemiological risk relationship between asthma and OPGs, in which T-cell-mediated disorders reduce glioma morbidity.¹¹¹ After asthma induction in Nf1 GEMs at 4 to 6 WOA, no tumor had grown at 12 to 24 WOA, and ON volumes decreased.¹¹² The potential mechanism was elucidated: upregulated expression of decorin (a member of the small leucine-rich proteoglycans) in T cells attenuated Ccl5 expression in microglia by inhibiting the NFκB pathway.¹¹² Hence, treatments with decorin or NFκB inhibitor in GEM at 4 to 6 WOA could hinder tumor formation at 12 WOA.¹¹² 

Neurons

In recent studies, it has been demonstrated that as constituents of the tumor microenvironment, hyperexcitable neurons contribute greatly to tumor growth by the paracrine pathway and electrochemical communication.^113,114 There is evidence that neurofibromin modulates hyperpolarization-activated cyclic nucleotide-gated (HCN) channels,¹¹⁵ which can affect the activity of neurons.^116,117 Further study has shown that tumor-related NF1-mutant RGCs exhibit increased excitability, controlled by HCN channel activity.¹¹⁸ Then, RGCs trigger upregulated expression of midkine to activate T-cell and microglial signaling cascades by the nuclear factor of activated T-cell-1 (NFAT-1) pathway in both human and murine OPGs.¹⁰⁸ Therefore, targeting the HCN channel with the antiseizure medication lamotrigine prevented the progression of Nf1-OPG in vivo.¹¹⁸ In addition, optic gliomas grow around the axons of RGCs, that is, ONs, which are responsible for transmitting light-induced signals from the retina to the cerebral cortex. The upregulation of ON activity induced by visual experience (light) was shown to initiate and promote optic glioma progression during a developmental period, whereas light deprivation inhibited tumor formation and rescued RGCs from death via the neuronal activity-dependent neuroligin 3 (NLGN3) signaling axis.^119,120 Similarly, the secretion of NLGN3 has also been shown to promote the proliferation of high-grade gliomas by active neurons.^121,122 

Other Non-Neoplastic Cells

Additionally, other cells within the tumor microenvironment may also contribute to the development of OPGs. There were elevated numbers of endothelial cells and overexpression of vascular endothelial growth factor (VEGF) in both human and murine OPGs, which were negatively correlated with the progression-free survival rate.^89,123–125 In the inner retinal layer, there are retina-specific glia named Müller cells, which secrete various trophic factors and cytokines into the surrounding environment via glia-glia (e.g. microglia) and glia-neuron interactions in the absence of NF1.⁹⁰ There is still much to discover about how the microenvironment and paracrine cells interact in tumor formation in future preclinical studies. 

Risk Factors for Glioma Formation and Progression

There has been an increasing focus on the early detection of NF1-OPGs by developing a scientific risk assessment algorithm to stratify patients based on their risk levels. Potential risk factors may include germline gene mutations, co-existing gene mutations, genomic alterations, and sexually dimorphic differences (Fig. 2). 

Figure 2.

 

The risk factors of NF1-OPGs. *A series of clinical and preclinical trials have found some potential risk factors of NF1-OPGs, including age, sex, tumor location, germline gene mutations, co-existent gene mutations, and genomic alterations. Further investigation is needed to develop a scientific risk assessment algorithm to stratify patients based on their risk levels. This figure was created using the BioRender software.

Germline Gene Mutations

It is generally known that different germline Nf1 gene mutations give rise to various expression levels of neurofibromin and distinct biological properties.^126,127 There are GEMs with two typical NF1-patient-originating Nf1 gene mutations (Gly848Arg¹²⁸ and Arg681X¹²⁹). However, it was only the mice with the Arg681X mutation that generated optic gliomas with upregulated proliferative rates, glial fibrillary acidic protein (GFAP) immunoreactivity, and RGC loss.^100,130 In addition, genetic mutations also affect the recruitment of non-neoplastic cells and the paracrine effects of the tumor microenvironment.¹⁰⁰ There was more Ccl5 secretion and JNK^{Thr183/Tyr185} activation in accumulated microglia infiltration, as well as increased Ccl5-regulated glial AKT^Thr308 in neoplastic astrocytes in Arg681X^CKO mice.¹³⁰ It appeared that germline mutations of the Nf1 gene had both effects intrinsic to the cell and effects from the surrounding tissue (stromal effects) on optic glioma phenotypes.¹³⁰ 

There were also potential genotype-phenotype relations in NF1-OPGs in clinical trials, which suggested that the germline mutations of NF1 may be useful as a biomarker for OPGs risk prediction. Studies have considered that patients with NF1-OPGs were more likely to have mutations in the 5′-tertile of the NF1 gene,^131–133 whereas other studies did not confirm this.¹³⁴ Given the small samples in the above studies, a combined cohort study was implemented to investigate the predictive role of NF1 mutations in 215 patients with NF1 (100 of whom had OPGs),¹³⁵ and the results showed that those who had mutations in the cysteine/serine-rich domain (CSRD; located in the fifth tertile) were more likely to develop OPGs, whereas those with mutations in the HEAT-like repeat region (HLR; amino acids 1825-2428) had a lower risk of forming OPGs. However, because there are many known variants of the large NF1 gene without definite genotype-phenotype correlations, genetic screening has proven to be challenging.^136,137 

Co-Existent Gene Mutations

It is generally accepted that NF1-related gliomas have complete NF1 gene inactivation, but recent studies have revealed that NF1-OPGs may also contain additional heterozygous deletions of PTEN, duplications of KIAA1549: BRAF^19,138,139 or FGFR1 alteration (needs further evidence).⁸⁸ Based on an integrated analysis of 1000 pediatric LGGs (pLGGs), genetic rearrangements were associated with a younger age of occurrence and less aggressive clinical symptoms; therefore, they were classified as low risk.⁸⁸ Single-nucleotide variant (SNV)-driven pLGGs, on the other hand, were more likely to be diagnosed later in life and necessitated multiple therapies and longer follow-ups.⁸⁸ 

GEM experiments indicated that OPGs expressing KIAA1549: BRAF exhibited no extra biological characteristics, whereas PTEN deletion and the loss of Nf1 may cooperate genetically to increase tumor aggressiveness.¹⁴⁰ PTEN counteracts the action of PI3K/AKT and inhibits the development of OPGs.¹⁴⁰ In Nf1^flox/mut, Pten^flox/wt, and GFAP-Cre (FMPC) mice, there was a greater proliferative index and microglial infiltration in larger optic gliomas, as well as a greater thinning of the RNFL by AKT activation, independent of mTOR.¹⁴⁰ According to whole-tumor RNA sequencing and deconvolution, FMPC mice displayed differentially expressed genes caused by downregulation of PTEN, which was normalized by PI3K inhibition (BKM120).⁷⁴ PTEN-enriched tumors showed significantly reduced progression-free survival compared to those depleted in BKM120 treatment signatures in low- and high-grade gliomas.⁷⁴ 

Genomic Alterations

There is a hypothesis that the progression of pLGGs is also affected by epigenetic alteration and specific miRNA expression differences, which might enhance stroma-induced MEK activation.^141,142 In this regard, epigenetic characteristics contribute greatly to the risk identification of glioma subgroups.¹⁴³ For example, a subset of PAs with H3K27 alteration are more likely to have worse clinical progression. In a sex-specific manner, a modification gene (Arlm1) on mouse chromosome 12 conferred astrocytoma resistance.¹⁴⁴ Although the modifier genes are not totally clear, potential candidates include those that regulate neurofibromin progression, o-GSC ability, and chemokine or microglial activation.¹⁴⁵ In addition, genetic backgrounds also had dramatic influences on gliomagenesis in Nf1 GEMs. For example, heterozygous inactivation of Nf1 and Trp53 in mice resulted in astrocytoma progression at a greater incidence on the B6/C57BL/6J (B6) genetic background than on the 129S4/SvJae (129S4) background.^146,147 To determine how epigenetic events play a crucial role in NF1-OPG pathogenesis, future research is needed. 

Sexually Dimorphic Differences

Currently, a sexually dimorphic interpretation of NF1 clinical abnormalities is intriguing. According to clinical investigation, girls with NF1-OPGs were 5 to 10 times more likely to experience visual dysfunction and require therapies than boys.^30,148 Female GEMs, as mentioned earlier, also displayed decreased visual capacity due to neurotoxins produced by ERβ-activated microglia.¹⁰⁹ In addition, two clinical studies demonstrated that boys were more likely to suffer from specific learning problems,^149,150 and only male Nf1 GEMs exhibited spatial learning/memory difficulties related to higher Ras activity and lower dopamine levels in the hippocampus.³⁰ The above findings suggested that sex, as a genomic determinant, was a key prognostic factor responsible for neuronal malfunction in NF1 through epigenetic mechanisms and that future preclinical or clinical studies should take this into account. Currently, a prospective observational multicenter study is underway to determine prognostic factors for visual outcomes. 

Targeted Therapies and Vision Recovery Strategies

Considering that tumor growth is slow in young patients with still developing brains, it is imperative that treatments can halt further tumor progression and neurologic degeneration without causing long-term neurocognitive consequences. Currently, promising targeted therapies for neoplastic cells or stromal cells have attracted much attention, and vision conservation treatments have flourished as well. Aside from functional outcomes, it is equally important to understand the best initial timing and length of therapy, the duration of efficacy, and the late outcomes.^151,152 

Targeting Neoplastic Cells

With chemotherapies having both short- and long-term complications,^153,154 therapeutic strategies targeting Nf1-null neoplastic cells by Ras pathways have emerged. As an inhibitor of B-RAF, sorafenib was used in a phase II clinical trial to treat children with low-grade astrocytomas.¹⁵⁵ The trial was halted due to an unprecedented promotion of tumor development by paradoxical ERK activation in patients.^156,157 To bypass this unexpected effect, selumetinib, an MEK inhibitor, is currently being tested in children with NF1-related LGGs.^158,159 These results indicated that selumetinib had an encouraging antitumor impact on patients with recurrent or refractory LGGs and could become an alternative for chemotherapy. At the time of this writing, there are several ongoing clinical studies to assess the treatment outcome or maximum tolerated dose of selumetinib (NCT03871257, NCT03326388, and NCT01089101), the effectiveness and correct dose of an orally selective MEK1/2 inhibitor called binimetinib (NCT02285439) and the efficacy of an oral pan-RAF inhibitor tovorafenib (NCT05566795; see the Table). In addition, treatment with the mTOR inhibitor everolimus also demonstrated considerable disease stability/shrinkage with stabilized VA in individuals with recurrent/progressive NF1-related LGG.^160,161 Rapamycin was combined with erlotinib, an EGFR inhibitor, to treat children with recurrent LGGs, which revealed prolonged disease stabilization in some patients.¹⁶² 

Table.

 

The Ongoing Clinical Trials for LGG or OPG

 

However, partial NF1-OPGs continued to grow after common therapies and needed alternative second-line treatments on account of the resistance mechanisms. For example, rapamycin treatment did not have a durable curative effect in OPGs because of its decreased sensitivity observed in o-GSCs through high-level expression of ATP binding cassette subfamily G member 1 (ABCG1).¹¹⁰ Given that mTOR activity in o-GSCs could be increased by p90RSK hyperactivation, combined p90RSK and mTOR inhibitors are needed to ameliorate rapamycin resistance and improve clinical outcomes.¹¹⁰ 

Targeting Non-Neoplastic Cells

There has been a growing recognition of the importance of the tumor microenvironment in disease pathogenesis, leading to new therapies targeting non-neoplastic cells and related signals from the tumor stroma.¹⁶³ Preclinical studies have demonstrated that inhibiting microglial function could mitigate tumor growth, but clinical trials have not been conducted.^96,97,99 In contrast, bevacizumab, an anti-VEGF receptor agent, combined with irinotecan or traditional chemotherapy regimens has been successfully used in recurrent LGGs.^164–167 Despite the small sample sizes, in clinical trials, treatment with bevacizumab resulted in disease control and even improved vision function with no significant toxicity.^168–170 Recently, in a large-scale clinical trial in the United Kingdom, 88 children with LGG received bevacizumab-based treatment.¹⁷¹ In addition to inducing a better overall visual response, bevacizumab treatment significantly controlled LGGs in the short term and prevented further deterioration. An ongoing trial (NCT02840409) is assessing the efficacy of bevacizumab and vinblastine combination in patients with NF1-LGG. In addition, a phase I trial of the anti-angiogenic agent lenalidomide has exhibited good tolerability and promising treatment results in pediatric LGGs.¹⁷² In addition, using IL-12, interferon beta/gamma gene delivery and immunomodulatory gene therapies against high-grade tumors enhanced the T-cell-mediated immune response,¹⁷³ providing new directions and reference significance for the treatment of OPGs. There have also been some vaccine- and interferon-based immunotherapeutic clinical trials conducted on patients with NF1-LGG (NCT02358187 and NCT02343224). 

Therapeutic Strategies for Visual Recovery

Previous studies considered tumor volume and vision dysfunction to be the main outcome measures of optic glioma evolution in children with NF1. However, it should be noted that the size of NF1-OPGs may not be related to visual function.²⁷ Reductions in visual function are often misinterpreted as an indication of tumor progression, which can lead to unnecessary interventions. There are some OPGs that can develop dramatically but cause no discernible deterioration of vision acuity or function.¹⁴ Meanwhile, smaller gliomas with irregular growth may result in severe visual problems. Additionally, spontaneous regression or shrinkage of gliomas may accompany vision loss partially owing to pressure gradients and kinking of axons.¹⁷⁴ A conference report also outlined that visual acuity should be considered as a primary outcome measure in future clinical studies.¹⁵² Given that current successful chemotherapies and molecularly targeted treatments of NF1-OPGs could lead to decreased tumor volume without satisfactory improvement in visual function,¹⁷⁵ it is necessary to cautiously assess the effectiveness of future therapies in clinical trials for NF1-OPG and develop feasible strategies to enhance vision function. 

There was a vulnerable window between the onset of RGC death and the loss of more than 30% of RGCs (predicted by RNFL thickness of 7.5 microns), according to promising preclinical studies.⁸⁴ During this window of susceptibility, administration of rolipram (increasing cAMP levels),⁸² lovastatin (inhibiting Ras activation),^81,84 or PHTPP (antagonizing ERβ)¹⁰⁹ can reduce RGC loss and reverse vision impairment. Future clinical trials should focus on these therapeutic directions and identify optimal dosing intervals for the greatest benefits in visual outcomes. 

Furthermore, neuroprotective strategies may be promising treatments for visual impairment. In a randomized phase II clinical trial investigating treatment with nerve growth factor (NGF) eye drops, 13 patients with NF1-OPGs exhibited improved RGC function and enlarged visual fields with few systemic side effects.¹⁷⁶ Currently, a randomized trial is being conducted to determine the safety and effectiveness of painless NGF CHF6467 eye drops on patients with OPGs (NCT05733572). There are various approaches to regenerate RGCs and improve retinal function in mice with other RGC-degenerated or -injured diseases, such as altering RGC gene expression,^177,178 reprogramming Müller cells into RGCs,^179,180 and transplanting induced pluripotent stem cell (iPSC)-derived RGCs.¹⁸¹ Treatments for NF1-OPGs involving vision restoration can benefit from the above approaches in the future.¹² 

Conclusions and Perspectives

These observations illustrate that the gliomagenesis of OPGs demands complex interplay between neoplastic and non-neoplastic cells (Fig. 3).¹⁴⁵ Susceptible o-GSCs differentiate and grow into astrocytes following tumor-initiating genetic mutations during a specific developmental window. Then, o-GSCs and astrocytes both secrete molecules to recruit nonneoplastic cells, which form a permissive tumor microenvironment. In turn, the non-neoplastic cells in the tumor stroma support neoplastic cell proliferation and tumor progression by a paracrine mechanism. 

Figure 3.

 

The complex communication between neoplastic and non-neoplastic cells. *O-GSCs expressed CX3CL1 to recruit microglia which elaborated molecules to support neoplastic cells proliferation, including Ccl5, CXCL12, and MGEA5, and, in turn, neoplastic cells also secreted molecules (“stromagens”) to recruit microglia. Activated by Ccl2 secreted from o-GSCs, T cells then produced Ccl4 to promotes microglia to excrete Ccl5. As immune system regulators, RGCs held upregulated expression of midkine to activate T cells and initiated optic gliomas progression by NLGN3 signaling axis. By contrast, in female GEMs, estrogen induced microglia to produce neurotoxins, like IL-1β and other inflammatory cytokines, resulting in vision deterioration. Besides, other cells within the tumor microenvironment, like endothelial cells and Müller cells, might also contribute to the development of OPGs. This figure was created using the BioRender software.

In addition, there is temporal and spatial heterogeneity among NF1-OPGs. Studies on o-GSCs have shown that the original cells determine the timing of murine OPG formation and somatic Nf1 loss given the temporal window.^61,62,72 The tumor microenvironment also contributes to this effect, such as a greater increase in Nf1^+/− microglia than WT microglia during 5 to 6 WOA⁹⁹ and a decrease in the expression of CXCL12 with increasing age.⁸⁰ Astrocytes from various regions of the CNS have unique gene expression patterns and biological characteristics. Astrocytes from the ON and brainstem expressed higher levels of the Nf1 gene and had more increased numbers following Nf1 gene inactivation than neocortical astrocytes, in agreement with the rarity of astrocytomas in the cortex.¹⁸² Cues from the surrounding stroma have also suggested that low cAMP levels accompanied by high CXCL12 expression along the optic pathway could account for the preferential localization of NF1-related gliomas.^59,80 

Although our understanding of NF1-OPGs is constantly evolving from a molecular and immunological perspective, there are many unsolved mysteries hidden in the tumor microenvironment. First, further research is needed to determine how stromal cell paracrine factors influence the activated Ras signaling axis in neoplastic cells. In addition, scientists observed that T-cell-mediated microglial Ccl5 expression could be hindered by decorin expression from T cells in asthma.¹¹² Understanding the molecular mechanisms that underlie OPGs with T-cell-mediated systemic diseases, such as eczema, rheumatoid arthritis, and diabetes, may provide insights into potential immunomodulatory therapies for neurological conditions. Immunological therapies, such as checkpoint inhibitors, may be uniquely beneficial to patients with NF1-LGG in a high-immune subgroup. In future studies, it will be necessary to establish immunocompetent mouse models and investigate the molecular and clinical implications of the immune microenvironment of patients at the signal cell level.¹⁵² Furthermore, previous studies have confirmed that in sporadic PAs, BRAF activation in astrocytes, or neural stem cells results in oncogene-induced senescence (OIS) for growth arrest caused by MAPK/ERK hyperactivation.^183,184 It was demonstrated that OIS expressed a senescence-associated secretory phenotype (SASP), which constituted a complex inflammatory network to further promote OIS.¹⁸⁵ It may be connected to the negative function of the tumor microenvironment in NF1-OPGs if this OIS/SASP circuit is present. The mechanism might explain the intriguing phenomenon that some children with NF1-OPGs experienced spontaneous regression or even complete disappearance documented by MRI.¹⁸⁶ As a result, therapies that mediate senescence and promote OIS may also prove promising in gliomas in the future.¹⁸⁷ 

Because human specimens are rarely available and the difficulty of maintaining in vitro or developing as xenografts in vivo, mouse optic glioma models have been widely recognized as classic research tools to investigate the molecular and cellular mechanisms of human NF1-OPGs. Murine models have many advantages,^{71,89,124,188} but they still have a large gap in comparison to human NF1-OPGs. First, GEMs have lower microglia levels and lack several classical features, such as eosinophilic granular bodies and Rosenthal fibers.^189,190 Second, gliomas appear later than normally seen in patients with NF1-OPGs (according to equivalent species ages) and do not affect the optic tracts or radiations.¹⁸⁹ Furthermore, mice depend more on olfaction for survival than vision and have a smaller vision system with fewer RGC subtypes.^12,189 Hence, due to their tractability and accessibility, minipigs have been successfully used as complementary NF1-OPG models.¹⁹¹ There are striking similarities between swine and humans in terms of visual anatomy and physiological functions (such as drug metabolism), which can provide a unique opportunity to research the natural history and drug efficacy of tumors.^192,193 In addition, considering species differences, NF1 individual-derived iPSCs also have the potential to serve as models.¹⁹⁴ Currently, NF1 iPSCs can be reprogrammed to generate human microglia-like cells¹⁹⁵ or engineered with 7 different patient-derived NF1 mutations to identify mutational equivalency in tumors.¹²⁶ 

Acknowledgments

Supported by Innovative Research Team of High-level Local Universities in Shanghai (SHSMU-ZDCX20210900 and SHSMU-ZDCX20210902). These figures were created using the BioRender software. 

Disclosure: Y. Chen, None; J. Yu, None; S. Ge, None; R. Jia, None; X. Song, None; Y. Wang, None; X. Fan, None