The Brain in NF1
NF1 patients are predisposed to CNS manifestations, mainly cognitive deficits, hyperintensities that appear on MRI imaging, and low grade pilocytic astrocytomas. NF1 mutations are also common in high grade glioma in the general population. Modeling of these phenotypes, understanding underlying mechanisms, and ongoing therapeutic trials were discussed.
Kathryn North (Children’s Hospital Westmead, Australia) overviewed cognitive deficits and academic learning difficulties, the most common neurological disability in patients with NF1. She found that developmental deficits can be identified clinically in patients as young as 2 years old [
15]. Some cognitive deficits may be associated with developmental or structural brain anomalies and are therefore unlikely to be reversible; these include increased grey matter volume and increased cross sectional area of the corpus callosum which are associated with lower IQ and learning deficits [
18], and NF1 MRI T2-hyperintensities (non-enhancing bright areas on T2 images of unknown etiology) in the thalamus that are associated with reduced IQ [
9]. Other cognitive deficits may be reversible and amenable to pharmacological therapy. Dr. North also overviewed published studies that used
Nf1 mouse models. She described studies from the Silva group reversing defects in hippocampal based learning with Lovastin and noted that another mouse model, and humans, with abnormal ras activation due to mutations in
SPRED1 (Legius syndrome) also have impaired hippocampal based learning [
5]. She described another
Nf1 mouse model showing abnormal dopamine homeostasis underlying attention deficits that could be reversed by treatment with methylphenidate, suggesting a defect in brain catecholamine homeostasis [
3]. These animal studies provide a strong imperative for pursuing definitive trials of stimulant medication in children with NF1.
Maria T. Acosta (Children’s National Medical Center) found that lovastatin normalizes the brain spontaneous low-frequency fluctuations in children with NF1 [
1]. She presented an initial examination of functional brain connectivity (FC) in Default Network (DN) architecture in 7/24 patients treated with lovastatin as a part of a phase 1 study testing lovastatin as treatment for cognitive deficits in children with NF1 at the Gilbert Neurofibromatosis Institute. Lovastatin administration in this small sample appeared to normalize anterior–posterior and local functional connectivity within the DN. The pattern of results is consistent with normalization of developmental processes, and with apparent benefits in a mouse model. Although interpretation is necessarily tentative because of the sample size, these results suggest that continued examination of the potentially beneficial effects of lovastatin in NF1 is warranted.
The mechanisms underlying the cognitive changes in NF1 remain incompletely understood. Debra Mayes (Cincinnati Children’s Hospital Medical Center) described her studies of how loss of Nf1 affects oligodendrocytes, the myelinating cell of the brain. She examined the optic nerve and corpus callosum white matter tracts in animal models. Phenotypes of Nf1 heterozygous mice, tamoxifen-induced Nf1 loss in mature oligodendrocytes (Plp-CreERT), and a new transgenic model in which the CNPase promoter drives expression of HRasG12V were presented. Expression HRasGV12 and loss of Nf1 (PLPCre;Nf1fl+ & PLPCre; Nf1fl/fl) in oligo-lineage cells displayed similar defects in myelin, brain vasculature and axonal size, in the optic nerves and corpus callosum. These data support the idea that oligodendrocytes, as well as neurons and astrocytes in the CNS might contribute to patient deficits in NF1 disease.
A workshop on gliomas in NF highlighted problems in clinical management and diagnosis that relate to gliomas in the NF population. Anat Stemmer-Rachamimov (Massachusetts General Hospital) reviewed the World Health Organization (WHO) classification of gliomas. In the WHO classification scheme tumors are classified by type (as per type or cell or origin) and then assigned a histological grade. In NF1 patients, the predominant problem is that of grading of astrocytomas; specifically distinguishing between pilocytic astrocytomas (WHO grade I) from diffuse astrocytomas (WHO grades II, III and IV). The classic pilocytic astrocytomas are non-infiltrating, and have unique histological features including piloid cells, myxoid background, Rosenthal fibers and eosinophilic granular bodies. Pilocytic astrocytomas have excellent prognosis. Diffuse astrocytomas are divided into grades II, III and IV based on the degree of cellular density, cytological atypia, mitotic activity, necrosis and microvascular proliferation. Diffuse astrocytomas progress over time and higher grades (III, IV) require treatment with radiation and chemotherapy. The distinction between pilocytic astrocytomas and diffuse astrocytomas in NF1 patients may be difficult as some tumors present histological features that are intermediate.
David Gutmann (Washington University, St Louis) reviewed the clinical features of NF1-associated astrocytomas. The overwhelming majority of CNS tumors in NF1 are optic pathway pilocytic astrocytomas (WHO Grade I). However, diffuse low grade and high grade astrocytomas also occur. These tumors often present a diagnostic and treatment challenges as their radiological and pathological features are different from sporadic tumors and there are no correlational large studies to provide guidance for clinical management and treatment.
Joann Ater (MD Anderson Cancer Center) presented data regarding treatment of progressive hypothalamic/optic pathway gliomas in children with NF1. The Children’s Oncology Group (COG) evaluated tumor response in 127 children with or without NF1, and progressive hypothalamic/optic pathway gliomas treated on protocol COG A9952. Children with NF1 tolerated carboplatin and vincristine well and had tumor response rate and EFS that were superior to children without NF1. Second malignant neoplasms occurred only in patients who had progression and received secondary treatment with temozolomide.
Josh Rubin (Washington University St Louis) outlined a novel set of findings regarding the cAMP pathway in low grade Nf1-related brain tumors. Gliomagenesis can be induced by low levels of cAMP in a genetically engineered mouse model of NF1. However, tumor formation displays a significant sex-dependent disparity resulting in disease in 100% of male, but only 20% of female mice. The molecular basis for the difference in tumor rates in male and female mice appears to involve cell-intrinsic differences in cAMP regulation. These exciting studies may provide insight into why astrocytoma is more common in males than females and whether targeting the cAMP pathway in astrocytoma treatment will have equal value for males and females.
Anat Stemmer-Rachamimov (with Cathy Nutt) reviewed the literature on genomic anomalies associated with sporadic gliomas. Genes often altered in sporadic low grade astrocytomas include
IDH1,
IDH2,
TP53 and
PDGFR while
EGFR,
CDKN2A, chromosome 10 loss,
PTEN and
MDM2 alterations are seen in high grade gliomas. Recent large scale studies using gene expression profiling have identified four subclasses of glioblastomas based on transcriptional signatures: proneural, neural, mesenchymal and classic [
21]. The mesenchymal subtype has been found to be associated with
NF1 mutations. Other mutations found in the mesenchymal subgroup included
PTEN,
TP53, and
RB1. This suggests that
NF1 mutations aggregate with specific other mutations and these may affect prognosis and response to specific treatment. A large correlative study in which histology, molecular features and clinical outcome of gliomas in NF1 will be integrated would provide valuable information as to clinical management and treatment of these tumors, and identify prognostic indicators.
Hui Zong (University of Oregon) used his elegant mouse genetic mosaic system MADM (Mosaic Analysis with Double Markers) to identify glioma cell-of-origin by analyzing aberrations in individual brain cell lineages prior to the glioma formation. After initiating
p53/NF1 mutations sporadically in neural stem cells (NSCs), they analyzed mutant NSCs and their progeny at pre-malignant stages. Their data pointed to a non-NSC cell type, the OPC cell, as the cell-of-origin for glioma [
13]. Further transcriptome analysis and genetic evidences confirmed the essential role of OPCs in gliomagenesis. Their findings provide valuable guidance for cell-type specific treatment for malignant glioma.
The peripheral nervous system in NF1
NF1 patients are predisposed to benign Schwann cell tumors called neurofibromas, and to malignant peripheral nerve sheath tumors, an aggressive sarcoma which is a leading cause of death in NF1 adults. The discussion centered continuing improvement of mouse models, identification of mechanisms of neurofibroma formation, preclinical therapeutics in existing mouse models, and ongoing clinical trials.
Alison Lloyd (University College London, UK) reported on two novel mouse models to study neurofibroma formation. The lab has previously shown in in vitro studies that activation of the ERK signaling pathway was sufficient to drive myelinating Schwann cells back to the progenitor-like cells found in nerves following an injury and which resemble the Schwann cells found in neurofibromas. They constructed a transgenic mouse in which ERK signaling could be reversibly induced following tamoxifen injections. These studies confirmed that this pathway was sufficient to drive this switch in cell-state, and identified that signals from the Schwann cells were responsible for inducing the inflammatory response seen following injury and during tumor formation. The links to injury were confirmed in a second mouse-model in which neurofibromas developed from adult myelinating Schwann cells when NF1 loss was induced in concert with an injury.
The link between inflammation and neurofibroma development was further explored in the talk by Yuan Zhu (University of Michigan). He reported a detailed temporal analysis of tumor development in his Nf1 mouse model. He showed that Nf1
−/− Schwann cells mostly differentiate normally except there are abnormalities in Remak-bundle formation which resolved as the animals aged. However, at these later ages, non-myelinating Schwann cells appeared to start to dissociate from axons concomitant with mast cell infiltration. These later defects were associated with increased ERK and mTOR signaling. Treatment with Rapamycin blocked most of the Schwann cell dissociation, indicating a potential treatment to block tumor formation at an early stage.
Nancy Ratner (Cincinnati Children’s Hospital Medical Center) emphasized that specific Ras signaling pathways that mediate neurofibroma and MPNST pathology remain unknown. She described cross-species transcriptome analysis to identify molecular pathways altered in NF1 tumors and guide molecular-targeted therapeutic strategies. The analysis identified global negative feedback of Ras-MAPK signaling was prominent in human and mouse. Inhibiting this pathway with a MEK inhibitor diminished human MPNST cell growth in xenografts and decreased neurofibroma volume in 85% of neurofibromas in a mouse model [
22], providing strong rationale for testing MEK inhibitors in neurofibroma and MPNST human clinical trials.
Wade Clapp (Indiana University School of Medicine) described his efforts to develop novel experimental therapeutics for plexiform neurofibromas. He previously reported that a multi- receptor tyrosine kinase inhibitor (imatinib mesylate), a drug that targets c-kit, PDGF, and c-abl, reduces plexiform neurofibromas in a mouse model [
23]. He has screened ten drugs that are in various phases of clinical development of which two passed both short term and long term treatment objectives that included reducing the size of the tumors by 50% and reducing the absolute number of tumors by 25% or more. His current work is aimed at moving these drugs forward into phase 2 clinical trials.
Brigitte Widemann (National Cancer Institute) described previous human trials for plexiform neurofibroma, utilizing agents such as farnesyl-transferase inhibitors, antifibrotic drugs, and immunomodulary agents (interferon) that have been completed with mixed and somewhat disappointing results; these trials were based on some preclinical testing, but not detailed pre-clinical mouse modeling. The interactions between the human and mouse trials have become a two way road, informing both types of studies. The next generation of human clinical trials for plexiform neurofibromas will be heavily based on preclinical modeling. Imatinib (NCT01140360) and sorafenib (NCT00727233) are two trials which have been recently undertaken based on mouse modeling, but these studies point out the complexity of the transition from mouse models to human care.
Dusica Babovic-Vuksanovic (Mayo Clinic) provided an update on results from a CTEP-NCI multi-center phase 2 trial of cediranib for NF1 plexiform and/or paraspinal neurofibromas (NCT00326872). The study enrolled 26 adult patients who were treated with oral cediranib at 30 mg/day over 26 cycles. The primary endpoint of therapy was a volume change of target tumors, based on 3-D MRI, measured by two independent blinded evaluators. A secondary endpoint was symptomatic improvement based on the Brief Pain Inventory. Four patients responded with decrease in tumor volume of >20% and no radiological progression of disease was noted in any of evaluable patients. Brief Pain Index analysis showed reduction in pain or a reduction in the interference caused by pain in treated patients. Results are encouraging but more data is needed to assess efficacy of cediranib in patients with NF1.
Kent Robertson (Indiana University School of Medicine) presented data from a pilot phase 2 study of imatinib in NF1 patients with plexiform neurofibromas (PN) (NCT01140360). The primary endpoint was PN size by MRI cross sectional area and secondary endpoints included symptomatic improvement, expression of 15 plasma cytokines, and safety/toxicity of imatinib. Of 24 evaluable patients, 15 (62%) responded with tumor reduction of one or more PN’s; 9 (38%) were non-responsive (stable or progressive tumors only). Out of 73 tumors scored, 22 (30%) showed reduction while 22 (30%) showed progression, and 29 tumors remained stable (40%). Symptom improvement was reported by 6 of 15 (40%) patients with responsive disease and 1 of 9 (11%) of patients with non-responsive tumors following 6 months of oral imatinib (p = 0.19). A panel of 15 plasma cytokines after 6 months of therapy showed greater than twofold increase in levels in 24% of cytokines measured in patients with non-responsive disease as compared to 2% in patients with responsive tumors (p < 0.001). Thus, a subset of NF1 patients with PN’s treated with imatinib had reductions in tumor size associated with symptom improvement and cytokine expression correlated with tumor response.
Fawn Leigh (Massachusetts General Hospital/Harvard Medical School) presented data related to efforts to identify genetic modifiers of cutaneous neurofibroma tumor burden in NF1. The MGH group carried out a genome-wide association analysis (GWA) in 300 subjects with extreme burden of cutaneous tumors, using the Affymetrix GeneChip 6.0 platform, providing 909, 622 single nucleotide polymorphism (SNP) markers. While their preliminary data indicate that there is no single common modifier responsible for the variation in tumor burden, genotyping in an additional cohort of extreme subjects to confirm or refute its modifier status is ongoing. Copy number variant (CNV) analysis in their initial cohort is also in process. Importantly, the authors solicit collaboration from all NF Clinics interested in a collaborative community-wide effort to address the cooperative identification of genetic modifiers of NF1.
David Largaespada (University of Minnesota)’s laboratory uses insertional mutagenesis for cancer gene discovery in the mouse. By crossing mice harboring mutagenic Sleeping Beauty (SB) transposons to mice expressing SB transposase in specific cell-types they are able to identify novel driver genes for tumor formation in specific tissues. Using this approach to target Schwann cells, Largaespada reported the formation of a large number of SB-accelerated neurofibromas and malignant peripheral nerve sheath tumors. Analysis of the SB transposon insertion sites resulted in the identification of Nf1 and Pten genes as recurrent insertional sites as well as a large number of other genes not previously identified as having a role in Schwann cell tumor formation. Interestingly, Pten and Nf1 inactivation often occurred in the same tumors arguing that NF1 loss may not be sufficient to achieve the levels of PI3-kinase signaling required for tumor formation.
Luis F. Parada (University of Texas Southwestern Medical Center) presented exciting results from a screen for novel drugs to treat glioblastomas and MPNSTs for which current therapeutics are currently ineffective. He took the novel approach of isolating cancer stem cells from glioblastomas and MPNSTs from his mouse-tumor models. Growing these cells under more physiological conditions, he pooled cells from multiple tumors for each tumor type, to provide a target for a screen that should hit multiple independent tumors. He then screened a 200K chemical library for small molecule drugs that differentially killed tumor cells versus normal cells. Interestingly, in this assay, a number of drugs that have been used therapeutically but with little success were ineffective. These screens have resulted in the identification of a number of new potential drugs that function at low concentrations and with some specificity, and thus are exciting candidates for future therapeutics.
Eduard Serra (Institut de Medicina Predictiva I Personalitzada del Cancer, Barcelona) correlated transcriptional profiles generated by the NF1 Microarray Consortium with SNP data from neurofibromas and MPNSTs. His group identified clusters of up-regulated or down-regulated genes that mapped in specific genomic regions (termed Transcriptional Imbalances, TIs), present in MPNSTs and not neurofibromas. Many TIs correlated with somatic copy number alterations (gains and losses of genetic material), with the best correlation with genomic gains rather than losses. Dr. Serra’s group suggested that long range epigenetic silencing (LRES) could be the mechanism responsible for the generation of TIs representing clusters of down-regulated genes that were not explained by losses of genetic material, since there was a good overlapping between the precise chromosomal position of these TIs and LRES previously identified in prostate cancer.
Karen Cichowski (Brigham and Women’s Hospital, Harvard Medical School) reviewed the finding that inhibition of the mTOR pathway alone is insufficient to cause MPNST cell death, and proposed a more comprehensive, cytotoxic approach. She described a new potential combination therapy for these tumors [
6]. Specifically, she showed that agents that enhance proteotoxic stress, including HSP90 inhibitors, induce rapid tumor regression in an aggressive genetically engineered mouse cancer model, but only when combined with the mTOR inhibitor, rapamycin. Her lab found that these agents synergize by promoting irresolvable proteotoxic stress and inducing catastrophic endoplasmic reticulum and mitochondrial damage. Moreover, Dr. Cichowski found that this combination therapy is also effective in a genetically engineered mouse model of K-Ras mutant lung cancer. These studies have defined a specific and promising drug combination that can be assessed in patients with MPNSTs.
Dina Lev (MD Anderson) is studying the impact of drug-induced autophagy in cancer. While several drugs promote tumor cell death via a possible autophagic mechanism and/or apoptosis, such drugs may also promote the unintentional survival of cancer cells, indicating a likely role for drug-induced autophagy as a component of tumor chemoresistance. Her group has demonstrated that blockade of the PI3K/mTOR pathway and inhibtion of HDAC induce autophagy in human MPNST cells, both in vitro and in xenografts [
14]. Furthermore, such genetic and chemical blockade of drug-induced autophagy significantly enhanced therapeutic apoptotic MPNST cell death, a clear indication that drug-induced autophagy represents a survival mechanism in MPNST that may identify a new focus of therapy development to prevent MPNSTs from becoming resistant to chemotherapy.
NF1 manifestations outside the nervous system
Children with NF1 are predisposed to myoproliferative disease, to bone abnormalities, and to vascular abnormalities.
Kevin Shannon (UCSF) presented a talk on the effects of MEK inhibitors in mouse models of acute myeloid leukemia and myeloproliferative neoplasms, including juvenile myelomonocytic leukemia. Interestingly, his data suggest that Raf/MEK/ERK signaling plays a central role in the pathologic growth of hematopoietic tumors that have inactivated NF1. However, he found that there are disease stage-specific effects of treatment with MEK inhibitors in that advanced cancers such as AML initiated by NF1 inactivation are highly dependent on Raf/MEK/ERK signaling and sensitive to MEK inhibitors [
11]. By contrast, early stage neoplasms in which NF1 inactivation is the dominant (or only) cell autonomous genetic lesions are less “addicted” to Raf/MEK/ERK signaling, but respond to MEK inhibition by modulating aberrant growth. Importantly, this treatment does not eradicate mutant cells. These data have implications for treating benign neoplasms and fully malignant cancers in persons with NF1.
David Ingram (Indiana University School of Medicine) addressed inflammation in the adventitia of the vessel wall and identified a central role for macrophages. It is notable that the recruitment of bone marrow macrophages takes months. Potential treatment regimens to interrupt recruitment of Nf1 heterozygous macrophages to inflamed neointima were addressed. Switching topics to assess the possibility that dysregulated Ras signaling could lead to aneurismal vessel disease, Dr. Ingram presented mouse models based on ApoE receptor signaling that when provided a high lipid or high cholesterol diet developed thickened adventitia and elevated oxidative stress. This oxidative stress has been seen in an Nf1 drosophila model, and provides potential in treatment paths for aneurismal vessel disease. Likewise, the ROS pathway is now a target for novel therapeutic intervention.
Kimberly Jett (University of British Columbia, Canada) discussed the effect of vitamin D3 treatment on bone density (BMD) in NF1 patients. They investigated the therapeutic potential of high dose oral vitamin D3 on bone mineral density in NF1 patients with vitamin D3 deficiency. The dose that maintained the serum 25-hydroxyvitamin D3 level >30 μg/l was used. Treated subjects had significantly improved BMD in hip and lumbar spine in comparison to untreated subjects. Further studies are needed to elucidate the mechanisms responsible for reduced bone mineral density in NF1.
Cautionary notes
The NF community is focused on using preclinical studies to guide clinical trials.
Matthias Karajannis (NYU Langone Medical Center) discussed how preclinical trials are useful to inform clinical trials by reviewing the use of cultured primary tumor cells, immortalized/transformed cells, and presumptive cancer stem cells. While useful for cell-based high throughput screens, the challenges associated with using these in vitro models were highlighted, including the development of secondary genetic or epigenetic changes that may arise in culture, as well as the inability to assess tumor-host and host-tumor interactions. As an alternative, he discussed advantages of genetically engineered mouse models for preclinical studies.
Scott Plotkin (Massachusetts General Hospital) highlighted the difficulties in directly comparing the preclinical mouse model to human patients. Variable expressivity of NF1 and NF2, effects of other genes, gene/environment interactions are limits to the ability of mouse models to inform human disease. Perhaps, the greatest limitation of the mouse models is their inability to model the impact of disease on quality of life that is decreased in NF1, NF2 and schwannomatosis patients. Understanding how psychological and social factors influence the experience of living with NF will require human rather than animal studies.