Type 1 neurofibromatosis (NF1 #MIM 162200) is the most common neurocutaneous disorder, affecting approximately one out of 3000 people worldwide [1
]. This autosomal dominant tumor predisposition syndrome is caused by inactivating mutations in NF1
, a tumor suppressor gene which encodes the RAS inhibitor neurofibromin involved in the regulation of cell growth and survival [2
This condition is characterized by wide allelic heterogeneity, with more than 3000 different variants reported so far, owing to the high mutational rate of the NF1
gene and the absence of hotspots [3
]. NF1 exhibits an extremely variable clinical expressivity, with most patients manifesting cutaneous and ocular signs, including café-au-lait spots, inguinal and axillary freckling, iris hamartomas and choroidal nodules by 6 years of age. A portion of NF1 patients develop one or more complications, including learning disabilities that affect up to 60% of children [4
]. The hallmark of this condition are neurofibromas, benign tumors originating from Schwann cells, which typically arise during adulthood, except for plexiform neurofibromas that are congenital [5
The predisposition to develop tumors involves also the central nervous system: the glioma of the optic pathway (OPG) is a relatively frequent complication of NF1 affecting around 20% of patients, is mostly observed during childhood [6
] and is included in the diagnostic criteria [7
Although being usually characterized by an indolent course, a variable portion of patients manifests symptoms, mainly vision loss and other ophthalmological symptoms, but also precocious puberty or neurological manifestations [8
]. The comprehension of the biology of these tumors has improved significantly over the last few years but challenges still remain: (i) the risk assessment in asymptomatic patients remains demanding, owing to the lack of valid biomarkers and the absence of prospective studies that may help in prognosis definition; (ii) the young age of this unique at-risk population and the learning disabilities that frequently coexist complicate the development of an effective OPG screening; (iii) treatments able to prevent or recover vision loss in patients with OPG are still not available.
In this review we will summarize and discuss the clinical features of OPG, the current diagnostic and therapeutic protocols and the most recent advances on its pathophysiology obtained from preclinical models.
4. Preclinical Models of OPG
The pathogenetic mechanisms at the basis of OPG development in NF1 are not completely understood, thus reducing the possibility to develop effective treatments. The main limitations for the study of the molecular pathways leading to OPG have been the paucity of biological samples, given the low rate of surgery, and the difficulty to obtain patient-derived xenograft-models or cell lines from these tumors [85
]. However, over the last few years distinct strategies have been used to investigate gliomagenesis.
No single model has been found to fully recapitulate the human tumor phenotype, but the combination of different research approaches (ranging from the employment of animal models with increasing complexity to the molecular characterization of human brain tumors and the use of engineered patient’s modified iPSC cells) has allowed to generate a fertile ground for future translational research projects. Hopefully, results obtained from these models will be applied in the clinical setting in order to stratify patients, to better define the prognosis and to promote novel clinical trials with biological drugs or other innovative therapies.
A number of preclinical models of NF1 have been developed, but only some of them are characterized by the development of OPG. Notably, most of our understanding of the biology of these tumors of the central nervous system derives from engineered NF1 mice. Since germline ablation of both copies of NF1
is embryonically lethal in rodents [86
P technology has been employed to conditionally inactivate this tumor suppressor gene in specific time windows during embryogenesis and/or in different neuroglial progenitors, yielding a series of distinct engineered mice with NF1 developing OPG [87
These murine models allowed a better comprehension of gliomagenesis: somatic Nf1
loss induced by Cre expression driven by different cell-specific promoters or induced by tamoxifen during embryogenesis, defined neural stem/progenitor cells (NPCs) and the more committed oligodendrocyte precursor cells (OPCs) as potential cell(s) of origin of OPG with latency of tumor development of 3 and 6 months, respectively [87
]. Furthermore, these engineered models of NF1 were crucial to confirming the role of RAS hyperactivation in the development of OPG, providing the biological rationale to test a series of compounds able to counteract the effectors downstream to RAS: the RAS-MEK-ERK and the PI3K-AKT-mTOR signaling pathways are up-regulated in tumor cells [91
], whereas the adenylyl cyclase-mediated cyclic AMP is associated with decreased levels of cAMP [94
]. However, some of the drugs acting on these pathways (including the mTOR inhibitor Rapamycin) are active in mice only during therapy, with recovery of tumor growth after treatment suspension [85
Additionally, murine models of NF1-associated OPG have provided relevant information regarding the role of microenvironment on tumor growth, and particularly of microglia, immune system-like non-neoplastic cells which, unlike glioma cells that undergo loss of heterozygosity (NF1−/−
), maintain the heterozygous state (NF1+/−
). A series of studies elegantly showed that these cells are required to sustain tumor growth, since loss of NF1
in neuroglial progenitor cells alone does not result in gliomagenesis [87
]. Remarkably, interaction between stromal cells (mainly microglia) and tumor cells is mediated by soluble factors that might be targetable: it has been shown that both genetic and pharmacological inhibition of microglia function is able to reduce OPG growth [95
]. Microglial gliomagens include chemokines, growth factors and inflammatory mediators; among a series of candidate molecules, Ccl5 was found enriched in murine OPG and treatment with Ccl5 neutralizing antibodies reduced tumor growth in mice [97
]. Furthermore, Guo and coauthors demonstrated that OPG variability in mouse relies on the ability of the tumor to secrete specific molecules that recruit T lymphocytes and microglia, rather than on the intrinsic growth ability of the tumor [98
]. T lymphocytes and microglia are required to establish a supportive tumor environment, as shown by studies in which their content correlate with tumor grade, but further studies are ongoing to define how they affect tumor growth.
Although translation from mouse models to patients remains challenging, these preclinical studies provide the bases to outline clinical trials in NF1 children. An effective treatment undoubtedly requires a better definition of tumor molecular signatures and probably needs a multiple-stage intervention, targeting both glioma cells and stromal non-neoplastic cells.
Besides, preclinical models have been fundamental to elucidating that vision loss due to NF1-associated OPG depends on retinal ganglion cell (RGC) death. Although no clinical trials addressing this pathogenetic pathways are currently available, strategies promoting RGC survival are under investigation in mouse [99
Recently, other mammalian models of NF1 have been generated, including the minipig that develops OPG in a proportion similar to humans. Considering the higher similarities between humans and pigs compared to rodents in many physiological functions (including drug metabolism), this new model of NF1 offers an unprecedented opportunity to study the natural history of these tumors and to test novel treatments [100
Despite the progress in our understanding of OPG pathogenesis, only few molecules proficiently tested in mouse models (including selumetinib) have been found to be effective in clinical trials, highlighting the considerable differences among species, as for instance the lower microglia content in rodents compared with humans. To overcome these limitations, mouse models are now supported by the use of induced pluripotent stem cells (iPSCs) derived from NF1 patients [101
]: human microglia-like cells have been successfully reprogrammed from NF1 iPSCs [102
], providing the bases for generating in vitro co-cultures of the two key players of OPG development.
Beyond its involvement in NF1, the NF1
gene behaves as a tumor suppressor in a number of sporadic malignancies [103
]. A recent study compared the molecular signatures of NF1-associated glioma (including OPG) with their sporadic counterpart, allowing to cluster low- and high-grade lesions and pinpointing distinct biological features that might have relevant implications for the development of novel treatments [104
]. Although this study obviously analyzed only those neoplasms requiring surgery (that is not usually the choice of treatment for these tumors) and included only a small proportion of OPGs, it highlights the following important issues: first, the heterogeneity of the molecular landscape of NF1-associated gliomas; second, the presence of lymphocytic infiltrates in a fraction of these tumors, thus indicating a potential application of immunotherapy.
5. Oncologic Treatment of Children with NF1-OPG
5.1. Standard Treatment
Systemic cytotoxic chemotherapy has been the mainstay of treatment in children with progressive NF1-OPG, with the combination of vincristine/carboplatin (VC) as the most used, and effective regimen, leading to an excellent overall survival, and satisfactory long-term tumor control.
However, in view of the unpredictable nature of NF1-OPG, anti-neoplastic treatment is only recommended, after an initial watch and wait
period, in case of documented clinical and/or radiological progression, with the former criteria defined as visual acuity loss of at least 0.2 logMAR units or new visual field loss appearance. [33
Other risk factors, as pre-existing threat to bilateral vision, as well as chiasmatic/post-chiasmatic OPG involvement, or progressive tumor growth, influence the overall decision, in favor of starting treatment. It is less clear whether young age at OPG diagnosis may constitute an independent prognostic factor for OPG progression, but young children (<1–2 years), may also need treatment, especially in the presence of abnormal vision for age, or unreliable visual testing [33
Treatment duration may vary between 12 and 18 months, according to two largest prospective trials in US (12 in COG A9952), and Europe (18 in SIOP-LGG2004), but with similar results in terms of 5-yr progression-free survival (PFS) of 69% ± 4%, and 71% ± 6%, respectively [20
Overall, VC is a well-tolerated regimen, with manageable hematological toxicity, and peripheral neuropathy and autonomic neuropathy, are the most frequent side effects with vincristine. Carboplatin related renal or ototoxicity are rare and mild, but need a regular monitoring throughout the treatment. Allergy to carboplatin can also occur with increased frequency, especially after the first months of treatment, and requires longer infusion, pre-medication, or change to alternative regimens, in case of severe or repeated reactions. There are very few, if any, long-term side effects with this regimen. However, chemotherapy-induced structural changes in cerebral white matter have been recently described in children with NF1-associated glioma treated with low-intensity chemotherapies [107
]. These neuroradiological findings require long-term follow-up studies to clarify their functional significance. Despite evidence of good response to chemotherapy, children with NF1-OPG have a variable functional outcome, particularly in term of vision, with lack of a clear clinic-radiologic correlation. In fact, ≤1/3 of children with NF1-OPG improve their vision, while ≥1/3 have a stable, and ≈1/3 a worse visual acuity (VA) during or after chemotherapy. Predictive factors for visual outcome are not defined yet, but age after 2 years and no post-chiasmatic involvement are often associated with VA improvement with chemo [19
]. Moreover, despite chemotherapy being able to stabilize vision in about 2/3 of patient, it is of further concern that about half of these patients started treatment with an already compromised vision in one or both eyes.
Other chemotherapy regimens might have comparable efficacy, including weekly vinblastine (5-yr PFS = 70–75%), a cheap and widely available drug, which is now offered as standard of care in some countries, either upfront or at relapse, in view of its favorable toxicity profile, compared to the VC regimen [112
]. The efficacy and toxicity of vinblastine, alone or in combination with non-cytotoxic agents as nilotinib (a multi TKI possibly targeting tumor microenviroment) or anti-VEGF (Bevacizumab), are being tested in two randomized phase II trials in children with progressive newly diagnosed NF1-LGG. Results are expected soon and might possibly extend the treatment armamentarium for these children.
Consistent data from various retrospective series indicate that Bevacizumab, alone or in combination, can lead to visual improvement as well as radiological response in a significant proportion (50–70%) of children with multiply relapsed OPG with or without NF1 [114
While bevacizumab has proven to be effective and with a reasonable toxicity including systemic hypertension, and proteinuria, when treatment duration extends beyond 12 months [115
], more than 40% children recur within few months after drug stop. Its effect is not durable and its role needs to be reconsidered in view of the high costs.
5.2. MEK Inhibitors
Recent data on preclinical models have confirmed the constitutional activation of the MAPK/ERK/MEK pathway as a direct consequence of NF1-associated loss of neurofibromin function, prompting recent clinical trials employing inhibitors of mitogen-activated protein kinase (MEK) for the treatment children with NF1-associated tumors, including OPG [92
]. Selumetinib (AZD6244) is a potent, selective, orally available, non-ATP competitive small molecule inhibitor of MEK-1/2, which has shown in vivo significant growth inhibition in NF1-associated plexiform neurofibromas [116
]. A recently published phase II clinical trial with the use of selumetinib in relapsed LGG, including NF1-associated OPG, showed that ten (40%) of 25 NF1-LGG patients achieved a sustained partial response after a median follow-up of 48 months, and only one patient progressed while on therapy. The 2-year PFS was 96 ± 4% [115
A prospective randomized phase 3 study, with the aim of comparing the efficacy of selumetinib vs. carboplatin-vincristine in terms of event-free survival (EFS) and visual function (primarily assessed by Teller Acuity Cards), in previously untreated progressive NF1-OPG, might be soon launched as a result of a joint collaboration between pharma, US (COG), as well as European (SIOP) brain tumor experts.
Furthermore, a phase I/II trial (INSPECT) will further assess dosing, toxicity, and preliminary efficacy of an intermittent on/off schedule of selumetinib in children with relapsed NF1-OPG in selected centers.
While selumetinib is the most advanced drug in terms of efficacy and toxicity data in the NF1 population, other MEK inhibitors (MEKi) such as trametinib, cobimetinib, or binimetinib, are currently being tested in early phase clinical trials, or as compassionate use, in children with OPG, with or without NF1, and may be also considered.
Skin toxicity is the most commonly observed side effect from MEKi, and it is more frequent in adolescent and post pubertal children, as acneiform, follicular rash or paronychia. While toxicity is usually mild and reversible, it requires prompt diagnosis and timely treatment to avoid the need for therapy interruption. Prevention of acneiform rash includes the regular use of thick emollient creams. Other less frequent side effects include diarrhea, edema, fatigue, hypertension, constipation, pain, pneumonitis, as well as asymptomatic CPK elevation. Retinal findings (e.g., retinal pigment epithelial detachment (RPED) due to subretinal fluid accumulation or retinal vein occlusion (RVO) are both associated with MEKi [117
]. If RPED is diagnosed, treatment with MEKi should be discontinued until resolution, or permanently if RVO occurs. Both RPED and RVO should be monitored with fundoscopic exam and/or OCT.
Although MEKi proved to be effective in NF1-associated tumors including OPG, several questions remain unanswered, including the possible occurrence of resistance mechanisms, the effect durability, how to overcome and most importantly, the long term side effects, especially in comparison with current chemotherapy.
5.3. Other Potential Treatments for NF1-OPG
A randomized trial from Falsini et al. [119
], showed that the regular use of nerve growth factor (NGF) eyedrops could possibly improve visual fields, VEPs and/or other electrophysiological parameters in 13 patients with NF1-OPG who received experimental NGF after the end of chemotherapy. A phase II trial will hopefully confirm the potential benefit of this innovative approach, with potentially very few, if any, systemic side effects.
Genetically engineered mouse (GEM) models with NF1 associated tumors showed that pharmacologic elevation of cAMP, by the use of PDE4 inhibitors (rolipram) resulted in a dramatically reduction of optic glioma growth. Unfortunately, these exciting preclinical data have not undergone further translation into clinical trials yet [120