Unraveling the complexity of the senescence-associated secretory phenotype in adamantinomatous craniopharyngioma using multimodal machine learning analysis.

BACKGROUND: Cellular senescence can have positive and negative effects on the body, including aiding in damage repair and facilitating tumor growth. Adamantinomatous craniopharyngioma (ACP), the most common pediatric sellar/suprasellar brain tumor, poses significant treatment challenges. Recent studies suggest that senescent cells in ACP tumors may contribute to tumor growth and invasion by releasing a senesecence-associated secretory phenotype. However, a detailed analysis of these characteristics has yet to be completed. METHODS: We analyzed primary tissue samples from ACP patients using single-cell, single-nuclei, and spatial RNA sequencing. We performed various analyses, including gene expression clustering, inferred senescence cells from gene expression, and conducted cytokine signaling inference. We utilized LASSO to select essential gene expression pathways associated with senescence. Finally, we validated our findings through immunostaining. RESULTS: We observed significant diversity in gene expression and tissue structure. Key factors such as NFKB, RELA, and SP1 are essential in regulating gene expression, while senescence markers are present throughout the tissue. SPP1 is the most significant cytokine signaling network among ACP cells, while the Wnt signaling pathway predominantly occurs between epithelial and glial cells. Our research has identified links between senescence-associated features and pathways, such as PI3K/Akt/mTOR, MYC, FZD, and Hedgehog, with increased P53 expression associated with senescence in these cells. CONCLUSIONS: A complex interplay between cellular senescence, cytokine signaling, and gene expression pathways underlies ACP development. Further research is crucial to understand how these elements interact to create novel therapeutic approaches for patients with ACP.

Multi-omic approach identifies hypoxic tumor-associated myeloid cells that drive immunobiology of high-risk pediatric ependymoma.

Ependymoma (EPN) is a devastating childhood brain tumor. Single-cell analyses have illustrated the cellular heterogeneity of EPN tumors, identifying multiple neoplastic cell states including a mesenchymal-differentiated subpopulation which characterizes the PFA1 subtype. Here, we characterize the EPN immune environment, in the context of both tumor subtypes and tumor cell subpopulations using single-cell sequencing (scRNAseq, n = 27), deconvolution of bulk tumor gene expression (n = 299), spatial proteomics (n = 54), and single-cell cytokine release assays (n = 12). We identify eight distinct myeloid-derived subpopulations from which a group of cells, termed hypoxia myeloid cells, demonstrate features of myeloid-derived suppressor cells, including IL6/STAT3 pathway activation and wound healing ontologies. In PFA tumors, hypoxia myeloid cells colocalize with mesenchymal-differentiated cells in necrotic and perivascular niches and secrete IL-8, which we hypothesize amplifies the EPN immunosuppressive microenvironment. This myeloid cell-driven immunosuppression will need to be targeted for immunotherapy to be effective in this difficult-to-cure childhood brain tumor.

Uncertainty-Aware Deep Learning Classification of Adamantinomatous Craniopharyngioma from Preoperative MRI.

Diagnosis of adamantinomatous craniopharyngioma (ACP) is predominantly determined through invasive pathological examination of a neurosurgical biopsy specimen. Clinical experts can distinguish ACP from Magnetic Resonance Imaging (MRI) with an accuracy of 86%, and 9% of ACP cases are diagnosed this way. Classification using deep learning (DL) provides a solution to support a non-invasive diagnosis of ACP through neuroimaging, but it is still limited in implementation, a major reason being the lack of predictive uncertainty representation. We trained and tested a DL classifier on preoperative MRI from 86 suprasellar tumor patients across multiple institutions. We then applied a Bayesian DL approach to calibrate our previously published ACP classifier, extending beyond point-estimate predictions to predictive distributions. Our original classifier outperforms random forest and XGBoost models in classifying ACP. The calibrated classifier underperformed our previously published results, indicating that the original model was overfit. Mean values of the predictive distributions were not informative regarding model uncertainty. However, the variance of predictive distributions was indicative of predictive uncertainty. We developed an algorithm to incorporate predicted values and the associated uncertainty to create a classification abstention mechanism. Our model accuracy improved from 80.8% to 95.5%, with a 34.2% abstention rate. We demonstrated that calibration of DL models can be used to estimate predictive uncertainty, which may enable clinical translation of artificial intelligence to support non-invasive diagnosis of brain tumors in the future.

The Iterative Design Process of an Explainable AI Application for Non-Invasive Diagnosis of CNS Tumors: A User-Centered Approach.

Artificial Intelligence (AI) is well-suited to help support complex decision-making tasks within clinical medicine, including clinical imaging applications like radiographic differential diagnosis of central nervous system (CNS) tumors. So far, there have been numerous examples of theoretical AI solutions for this space, for example, large-scale corporate efforts like IBM's Watson AI. However, clinical implementation remains limited due to factors related to the alignment of this technology in the clinical setting. User-Centered Design (UCD) is a design philosophy that focuses on developing tailored solutions for specific users or user groups. In this study, we applied UCD to develop an explainable AI tool to support clinicians in our use case. Through four design iterations, starting from basic functionality and visualizations, we progressed to functional prototypes in a realistic testing environment. We discuss our motivation and approach for each iteration, along with key insights gained. This UCD process has advanced our conceptual idea from feasibility testing to interactive functional AI interfaces designed for specific clinical and cognitive tasks. It has also provided us with directions to develop further an AI system for the non-invasive diagnosis of CNS tumors.

Adamantinomatous craniopharyngioma associated with a compromised blood-brain barrier: patient series.

BACKGROUND: Adamantinomatous craniopharyngioma (ACP) is a highly morbid adult and pediatric brain tumor derived from epithelial remnants of the craniopharyngeal canal (Rathke's pouch), which gives rise to the anterior pituitary gland. Standard therapy includes maximal safe resection with or without radiation therapy. Systemic antitumor therapy remains elusive. Immune-related paracrine signaling involving the interleukin-6 receptor (IL-6R) may contribute to ACP pathogenesis. Tocilizumab, a recombinant humanized monoclonal antibody against IL-6R, is approved by the US Food and Drug Administration but does not cross an intact blood-brain barrier. OBSERVATIONS: In a phase 0 trial design, a single dose of tocilizumab was delivered intravenously before clinically indicated surgical intervention in 3 children with ACP. The presence of tocilizumab was assayed in plasma, tumor tissue, tumor cyst fluid, and cerebrospinal fluid (n = 1) using a novel enzyme-linked immunosorbent assay. Tocilizumab reached ACP tumor tissue and/or cyst fluid after one systemic dose in every patient. LESSONS: This finding helps explain extant data that indicate tocilizumab may contribute to ACP therapy. It further indicates that ACP does not reside behind an intact blood-brain barrier, dramatically broadening the range of potential antitumor therapies against this tumor. This has substantial implications for the design of future clinical trials for novel therapies against ACP in both children and adults.

Robust deep learning classification of adamantinomatous craniopharyngioma from limited preoperative radiographic images.

Deep learning (DL) is a widely applied mathematical modeling technique. Classically, DL models utilize large volumes of training data, which are not available in many healthcare contexts. For patients with brain tumors, non-invasive diagnosis would represent a substantial clinical advance, potentially sparing patients from the risks associated with surgical intervention on the brain. Such an approach will depend upon highly accurate models built using the limited datasets that are available. Herein, we present a novel genetic algorithm (GA) that identifies optimal architecture parameters using feature embeddings from state-of-the-art image classification networks to identify the pediatric brain tumor, adamantinomatous craniopharyngioma (ACP). We optimized classification models for preoperative Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and combined CT and MRI datasets with demonstrated test accuracies of 85.3%, 83.3%, and 87.8%, respectively. Notably, our GA improved baseline model performance by up to 38%. This work advances DL and its applications within healthcare by identifying optimized networks in small-scale data contexts. The proposed system is easily implementable and scalable for non-invasive computer-aided diagnosis, even for uncommon diseases.

Establishment of patient-derived orthotopic xenograft model of 1q+ posterior fossa group A ependymoma.

BACKGROUND: Treatment for pediatric posterior fossa group A (PFA) ependymoma with gain of chromosome 1q (1q+) has not improved over the past decade owing partially to lack of clinically relevant models. We described the first 2 1q+ PFA cell lines, which have significantly enhanced our understanding of PFA tumor biology and provided a tool to identify specific 1q+ PFA therapies. However, cell lines do not accurately replicate the tumor microenvironment. Our present goal is to establish patient-derived xenograft (PDX) mouse models. METHODS: Disaggregated tumors from 2 1q+ PFA patients were injected into the flanks of NSG mice. Flank tumors were then transplanted into the fourth ventricle or lateral ventricle of NSG mice. Characterization of intracranial tumors was performed using imaging, histology, and bioinformatics. RESULTS: MAF-811_XC and MAF-928_XC established intracranially within the fourth ventricle and retained histological, methylomic, and transcriptomic features of primary patient tumors. We tested the feasibility of treating PDX mice with fractionated radiation or chemotherapy. Mice tolerated radiation despite significant tumor burden, and follow-up imaging confirmed radiation can reduce tumor size. Treatment with fluorouracil reduced tumor size but did not appear to prolong survival. CONCLUSIONS: MAF-811_XC and MAF-928_XC are novel, authentic, and reliable models for studying 1q+ PFA in vivo. Given the successful response to radiation, these models will be advantageous for testing clinically relevant combination therapies to develop future clinical trials for this high-risk subgroup of pediatric ependymoma.

Preclinical analysis of MTOR complex 1/2 inhibition in diffuse intrinsic pontine glioma.

Diffuse intrinsic pontine glioma (DIPG) is an incurable childhood brain tumor. The mechanistic target of rapamycin (MTOR), a key oncogene, functions as two distinct signaling complexes, MTORC1 and MTORC2. We set out to determine the preclinical efficacy and mechanism of action of MTOR inhibitors in DIPG. We evaluated the MTORC1 inhibitor everolimus and the MTORC1/2 inhibitor AZD2014 in three patient-derived DIPG cell lines using cell culture models. We created dose-response curves for both compounds. We measured phenotypic effects on cell self-renewal, apoptosis, cell cycle, differentiation, senescence, and autophagy. We assessed the effects of each compound on the AKT pathway. Finally, we measured the efficacy of AZD2014 in combination with radiation therapy (RT) and a panel of FDA-approved chemotherapy drugs. While everolimus showed minimal antitumor efficacy, AZD2014 revealed IC50 levels of 410-552 nM and IC90 levels of 1.30-8.86 µM in the three cell lines. AZD2014 demonstrated increased inhibition of cell self-renewal compared to everolimus. AZD2014 decreased expression of phospho-AKT, while no such effect was noted with everolimus. Direct AKT inhibition showed similar efficacy to AZD2014, and induction of constitutive AKT activity rescued DIPG cells from the effects of AZD2014. AZD2014 exhibited synergistic relationships with both RT and various chemotherapy agents across classes, including the multikinase inhibitor ponatinib. MTORC1/2 inhibition shows antitumor activity in cell culture models of DIPG due to the effect of MTORC2 inhibition on AKT. This strategy should be further assessed for potential incorporation into combinatorial approaches to the treatment of DIPG.

Characterization of 2 Novel Ependymoma Cell Lines With Chromosome 1q Gain Derived From Posterior Fossa Tumors of Childhood.

Ependymoma (EPN) is a common brain tumor of childhood that, despite standard surgery and radiation therapy, has a relapse rate of 50%. Clinical trials have been unsuccessful in improving outcome by addition of chemotherapy, and identification of novel therapeutics has been hampered by a lack of in vitro and in vivo models. We describe 2 unique EPN cell lines (811 and 928) derived from recurrent intracranial metastases. Both cell lines harbor the high-risk chromosome 1q gain (1q+) and a derivative chromosome 6, and both are classified as molecular group A according to transcriptomic analysis. Transcriptional enrichment of extracellular matrix-related genes was a common signature of corresponding primary tumors and cell lines in both monolayer and 3D formats. EPN cell lines, when cultured in 3D format, clustered closer to the primary tumors with better fidelity of EPN-specific transcripts than when grown as a monolayer. Additionally, 3D culture revealed ependymal rosette formation and cilia-related ontologies, similar to in situ tumors. Our data confirm the validity of the 811 and 928 cell lines as representative models of intracranial, posterior fossa 1q+ EPN, which holds potential to advance translational science for patients affected by this tumor.

Polo-like Kinase 1 as a potential therapeutic target in Diffuse Intrinsic Pontine Glioma.

BACKGROUND: Diffuse intrinsic pontine gliomas (DIPGs) are highly aggressive, fatal, childhood tumors that arise in the brainstem. DIPGs have no effective treatment, and their location and diffuse nature render them inoperable. Radiation therapy remains the only standard of care for this devastating disease. New therapeutic targets are needed to develop novel therapy for DIPG. METHODS: We examined the expression of PLK1 mRNA in DIPG tumor samples through microarray analysis and found it to be up regulated versus normal pons. Using the DIPG tumor cells, we inhibited PLK1 using a clinically relevant specific inhibitor BI 6727 and evaluated the effects on, proliferation, apoptosis, induction of DNA damage and radio sensitization of the DIPG tumor cells. RESULTS: Treatment of DIPG cell lines with BI 6727, a new generation, highly selective inhibitor of PLK1, resulted in decreased cell proliferation and a marked increase in cellular apoptosis. Cell cycle analysis showed a significant arrest in G2-M phase and a substantial increase in cell death. Treatment also resulted in an increased γH2AX expression, indicating induction of DNA damage. PLK1 inhibition resulted in radiosensitization of DIPG cells. CONCLUSION: These findings suggest that targeting PLK1 with small-molecule inhibitors, in combination with radiation therapy, will hold a novel strategy in the treatment of DIPG that warrants further investigation.

Checkpoint kinase 1 expression is an adverse prognostic marker and therapeutic target in MYC-driven medulloblastoma.

Checkpoint kinase 1 (CHK1) is an integral component of the cell cycle as well as the DNA Damage Response (DDR) pathway. Previous work has demonstrated the effectiveness of inhibiting CHK1 with small-molecule inhibitors, but the role of CHK1 mediated DDR in medulloblastoma is unknown. CHK1, both at the mRNA and protein level, is highly expressed in medulloblastoma and elevated CHK1 expression in Group3 medulloblastoma is an adverse prognostic marker. CHK1 inhibition with the small-molecule drug AZD7762, results in decreased cell growth, increased DNA damage and cell apoptosis. Furthermore, AZD7762 acts in synergy with cisplatin in reducing cell proliferation in medulloblastoma. Similar phenotypic changes were observed with another CHK1 inhibitor, PF477736, as well as genetic knockdown using siRNA against CHK1. Treatments with small-molecule inhibitors of CHK1 profoundly modulated the expression of both upstream and downstream target proteins within the CHK1 signaling pathways. This suggests the presence of a feedback loop in activating CHK1. Overall, our results demonstrate that small-molecule inhibition of CHK1 in combination with, cisplatin, is more advantageous than either treatment alone, especially for Group 3 medulloblastoma, and therefore this combined therapeutic approach serves as an avenue for further investigation.