Clinical Trials Targeting Neurofibromatoses-associated Tumors: A Systematic Review

Background There is a paucity of literature that comprehensively analyzes previous and current clinical trials targeting neurofibromatoses-related tumors. This article aims to provide readers of drug development efforts targeting these tumors by analyzing translational and clinical findings. Methods This systematic review was written according to the PRISMA guidelines. Inclusion criteria were clinical trials involving patients with neurofibromatosis type 1, type 2, or schwannomatosis that were treated with therapies targeting neurofibromatoses-associated tumors and that were registered on clinicaltrials.gov. In addition, a search was performed in PubMed, Web of Science, Google Scholar, and Embase European for articles fully describing these clinical trials. Results A total of 265 clinical trials were registered and screened for eligibility. Ninety-two were included in this systematic review involving approximately 4,636 participants. The number of therapies analyzed was more than 50. Drugs under investigation mainly act on the MAPK/ERK and PI3K/AKT/mTOR pathways, tumor microenvironment, or aberrantly over-expressed cell surface receptors. Selumetinib was the most effective medication for treating a neurofibromatosis type 1-associated tumor with approximately 68-71% partial response for inoperable or progressive plexiform neurofibromas in children 2 years of age and older and bevacizumab for a neurofibromatosis type 2-related tumor with approximately 36-41% partial response for vestibular schwannomas in patients 12 years of age and older. Conclusions This systematic review presents the results of previous clinical investigations and those under development for neurofibromatoses-associated tumors. Clinicians may use this information to strategize patients to appropriate clinical trials.

Irradiation of the Subventricular Zone and Subgranular Zone in High- and Low-Grade Glioma Patients: an Atlas-based Analysis on Overall Survival

Background Neural stem cells in the subventricular- (SVZ) and subgranular zone (SGZ) are hypothesized to support growth of glioma. Therefore, irradiation of the SVZ and SGZ might reduce tumor growth and might improve overall survival (OS). However, it may also inhibit the repair capacity of brain tissue. The aim of this retrospective cohort study is to assess the impact of SVZ and SGZ radiotherapy doses on OS of patients with high-grade (HGG) or low-grade (LGG) glioma. Methods We included 273 glioma patients who received radiotherapy. We created an SVZ atlas, shared openly with this work, while SGZ labels were taken from the CoBRA atlas. Next, SVZ and SGZ regions were automatically delineated on T1 MR-images. Dose and OS correlations were investigated with Cox regression and Kaplan-Meier analysis. Results Cox regression analyses showed significant hazard ratios for SVZ dose (univariate: 1.029/Gy, p<0.001; multivariate: 1.103/Gy, p = 0.002) and SGZ dose (univariate: 1.023/Gy, p<0.001; multivariate: 1.055/Gy, p<0.001) in HGG patients. Kaplan-Meier analysis showed significant correlations between OS and high/low dose groups for HGG patients (SVZ: respectively 10.7 months (>30.33 Gy) vs 14.0 months (<30.33 Gy) median OS, p = 0.011; SGZ: respectively 10.7 months (>29.11 Gy) vs 15.5 months (<29.11 Gy) median OS, p<0.001). No correlations between dose and OS were found for LGG patients. Conclusion Irradiation doses on neurogenic areas correlate negatively with OS in patients with HGG. Whether sparing of the SVZ and SGZ during radiotherapy improves OS, should be subject of prospective studies.

Long-term effectiveness of Gliadel ® implant for malignant glioma and prognostic factors for survival: 3-year results of a post-marketing surveillance in Japan

Background Adjuvant treatment with Gliadel wafers may prolong overall survival (OS) for malignant glioma patients without increasing toxicity. In Japan, the long-term OS of these patients treated with Gliadel 7.7 mg implants has not been studied. We evaluated OS and prognostic factors that might affect OS in Japanese patients with malignant glioma who received the Gliadel 7.7 mg implant. Methods This observational, long-term, post-marketing surveillance was an extension of a previous surveillance. Data were collected through case report forms at 2 and 3 years after Gliadel implant. Up to 8 Gliadel wafers (61.6 mg of carmustine) were placed over the tumor resection site. Primary endpoints were OS and prognostic factors that may influence OS. Results Among the 506 patients analyzed, 62.6% had newly diagnosed disease, and 37.4% had recurrent disease; 79.1% had glioblastoma histological type and 79.6% had World Health Organization Grade IV disease. Patients received a median of 8 wafers. The median OS was 18.0 months; OS rates were 39.8% and 31.5% at 2 and 3 years, respectively. Age ≥65 years (hazard ratio [HR]: 1.456; P = 0.002), lower resection rate (HR: 1.206; P < 0.001), recurrence (HR: 2.418; P < 0.001), and concomitant radiotherapy (HR: 0.588; P < 0.001) were identified as significant prognostic factors. Conclusions This study confirmed the 2- and 3-year OS of Japanese malignant glioma patients with varied backgrounds after Gliadel implant. With a careful interpretation of indirect comparisons with previously reported data, the results suggest that prognosis could be improved with Gliadel implants.

Improved prediction of postoperative paediatric cerebellar mutism syndrome using an artificial neural network

Background Postoperative paediatric cerebellar mutism syndrome (pCMS) is a common but severe complication which may arise following the resection of posterior fossa tumours in children. Two previous studies have aimed to preoperatively predict pCMS, with varying results. In this work, we examine the generalisation of these models and determine if pCMS can be predicted more accurately using an artificial neural network (ANN). Methods An overview of reviews was performed to identify risk factors for pCMS, and a retrospective dataset collected as per these defined risk factors from children undergoing resection of primary posterior fossa tumours. The ANN was trained on this dataset and its performance evaluated in comparison to logistic regression and other predictive indices via analysis of receiver operator characteristic curves. Area under the curve (AUC) and accuracy were calculated and compared using a Wilcoxon signed rank test, with p<0.05 considered statistically significant. Results 204 children were included, of whom 80 developed pCMS. The performance of the ANN (AUC 0.949; accuracy 90.9%) exceeded that of logistic regression (p<0.05) and both external models (p<0.001). Conclusion Using an ANN, we show improved prediction of pCMS in comparison to previous models and conventional methods.

Multivariate Analysis of Associations between Clinical Sequencing and Outcome in Glioblastoma

Background Many factors impact survival in patients with glioblastoma, including age, Karnofsky Performance Status, post-operative chemoradiation, IDH1/2 mutation status, MGMT promoter methylation status, and extent of resection. High-throughput next generation sequencing is a widely available diagnostic tool, but the independent impact of tumors harboring specific mutant genes on survival and the efficacy of extent of resection are not clear. Methods We utilized a widely available diagnostic platform (FoundationOne CDx) to perform high-throughput next generation sequencing on 185 patients with newly diagnosed glioblastoma in our tertiary care center. We performed multivariate analysis to control for clinical parameters with known impact on survival to elucidate the independent prognostic value of prevalent mutant genes and the independent impact of gross total resection. Results When controlling for factors with known prognostic significance including IDH1/2 mutation and after multiple comparisons analysis, CDKN2B and EGFR mutations were associated with reduced overall survival while PTEN mutation was associated with improved overall survival. Gross total resection, compared to other extent of resection, was associated with improved overall survival in patients with tumors harboring mutations in CDKN2A, CDKN2B, EGFR, PTEN, TERT promoter, and TP53. All patients possessed at least one of these six mutant genes. Conclusions This study verifies the independent prognostic value of several mutant genes in glioblastoma. Six commonly found mutant genes were associated with improved survival when gross total resection was achieved. Thus, even when accounting for known predictors of survival and multiple mutant gene comparisons, extent of resection continues to be strongly associated with survival.

Radiation-induced neuroinflammation – a potential protective role for PARP (poly ADP ribose polymerase) inhibitors?

Radiotherapy (RT) plays a fundamental role in the treatment of glioblastoma (GBM). GBM are notoriously invasive and harbour a subpopulation of cells with stem-like features which exhibit upregulation of the DNA damage response and are radioresistant. High radiation doses are therefore delivered to large brain volumes and are known to extend survival but also cause delayed toxicity with 50-90% of patients developing neurocognitive dysfunction. Emerging evidence identifies neuroinflammation as a critical mediator of the adverse effects of RT on cognitive function. In addition to its well-established role in promoting repair of radiation induced DNA damage, activation of poly(ADP-ribose) polymerase (PARP) can exacerbate neuroinflammation by promoting secretion of inflammatory mediators. Therefore, PARP represents an intriguing mechanistic link between radiation-induced activation of the DNA damage response and subsequent neuroinflammation. PARP inhibitors have emerged as promising new agents for GBM when given in combination with RT, with multiple preclinical studies demonstrating radiosensitizing effects and at least three compounds being evaluated in clinical trials. We propose that concomitant use of PARP inhibitors could reduce radiation-induced neuroinflammation and reduce the severity of radiation-induced cognitive dysfunction while at the same time improving tumour control by enhancing radiosensitivity.

Integrating morphologic and molecular histopathological features through whole slide image registration and deep learning

Background Modern molecular pathology workflows in neuro-oncology heavily rely on the integration of morphologic and immunohistochemical patterns for analysis, classification, and prognostication. However, despite the recent emergence of digital pathology platforms and artificial intelligence-driven computational image analysis tools, automating the integration of histomorphologic information found across these multiple studies is challenged by large files sizes of whole slide images (WSIs) and shifts/rotations in tissue sections introduced during slide preparation. Methods To address this, we develop a workflow that couples different computer vision tools including scale-invariant feature transform (SIFT) and deep learning to efficiently align and integrate histopathological information found across multiple independent studies. We highlight the utility and automation potential of this workflow in the molecular subclassification and discovery of previously unappreciated spatial patterns in diffuse gliomas. Results First, we show how a SIFT-driven computer vision workflow was effective at automated WSI alignment in a cohort of 107 randomly selected surgical neuropathology cases (97/107 (91%) showing appropriate matches, AUC = 0.96). This alignment allows our AI-driven diagnostic workflow to not only differentiate different brain tumor types, but also integrate and carry out molecular subclassification of diffuse gliomas using relevant immunohistochemical biomarkers (IDH1-R132H, ATRX). To highlight the discovery potential of this workflow, we also examined spatial distributions of tumors showing heterogenous expression of the proliferation marker MIB1 and Olig2. This analysis helped uncovered an interesting and unappreciated association of Olig2 positive and proliferative areas in some gliomas (r = 0.62). Conclusion This efficient neuropathologist-inspired workflow provides a generalizable approach to help automate a variety of advanced immunohistochemically compatible diagnostic and discovery exercises in surgical neuropathology and neuro-oncology.

Mitochondrial DNA copy number as a prognostic marker is age-dependent in adult glioblastoma

Background Glioblastoma (GBM) is the most common and aggressive form of glioma. GBM frequently displays chromosome (chr) 7 gain, chr 10 loss and/or EGFR amplification (chr7+/chr10-/EGFRamp). Overall survival (OS) is 15 months after treatment. In young adults, IDH1/2 mutations are associated with longer survival. In children, histone H3 mutations portend a dismal prognosis. Novel reliable prognostic markers are needed in GBM. We assessed the prognostic value of mitochondrial DNA (mtDNA) copy number in adult GBM. Methods mtDNA copy number was assessed using real-time quantitative PCR in 232 primary GBM. Methylation of POLG and TFAM genes, involved in mtDNA replication, was assessed by bisulfite-pyrosequencing in 44 and 51 cases, respectively. Results Median age at diagnosis was 56.6 years-old and median OS, 13.3 months. 153/232 GBM (66 %) displayed chr7+/chr10-/EGFRamp, 23 (9.9 %) IDH1/2 mutation, 3 (1.3 %) H3 mutation and 53 (22.8 %) no key genetic alterations. GBM were divided into two groups, “Low” (n = 116) and “High” (n = 116), according to the median mtDNA/nuclear DNA ratio (237.7). There was no significant difference in OS between the two groups. By dividing the whole cohort according to the median age at diagnosis, OS was longer in the “High” vs “Low” subgroup (27.3 vs 15 months, p = 0.0203) in young adult GBM (n = 117) and longer in the “Low” vs “High” subgroup (14.5 vs 10.2 months, p = 0.0116) in older adult GBM (n = 115). POLG was highly methylated, whereas TFAM remained unmethylated. Conclusion mtDNA copy number may be a novel prognostic biomarker in GBM, its impact depending on age.

Immune deconvolution and temporal mapping identifies stromal targets and developmental intervals for abrogating murine low-grade optic glioma formation

Background Brain tumor formation and progression are dictated by cooperative interactions between neoplastic and non-neoplastic cells. This stromal dependence is nicely illustrated by tumors arising in the Neurofibromatosis type 1 (NF1) cancer predisposition syndrome, where children develop low-grade optic pathway gliomas (OPGs). Using several authenticated Nf1-OPG murine models, we previously demonstrated that murine Nf1-OPG growth is regulated by T cell function and microglia Ccl5 production, such that their inhibition reduces tumor proliferation in vivo. While these interactions are critical for established Nf1-OPG tumor growth, their importance in tumor formation has not been explored. Methods A combination of bulk and single cell RNA mouse optic nerve sequencing, immunohistochemistry, T cell assays, and pharmacologic and antibody-mediated inhibition methods were used in these experiments. Results We show that T cells and microglia are the main non-neoplastic immune cell populations in both murine and human LGGs. Moreover, we demonstrate that CD8 + T cells, the predominant LGG-infiltrating lymphocyte population, are selectively recruited through increased Ccl2 receptor (Ccr4) expression in CD8 +, but not CD4 +, T cells, in a NF1/RAS-dependent manner. Finally, we identify the times during gliomagenesis when microglia Ccl5 production (3-6 weeks of age) and Ccl2-mediated T cell infiltration (7-10 weeks of age) occur, such that temporally-restricted Ccl2 or Ccl5 inhibition abrogates tumor formation >3.5 months following the cessation of treatment. Conclusions Collectively, these findings provide proof-of-concept demonstrations that targeting stromal support during early gliomagenesis durably blocks murine LGG formation.