The influence of the activation function in a capsule network for brain tumor type classification

Abstract INTRODUCTION We utilized national registry data to evaluate the unique epidemiology of primary adolescent and young adult (AYA) brain tumors according to the WHO2016 classification. METHODS AYA patients (15≤age≤39) presenting between 2004-2017 with a brain tumor were identified by ICD-O-3 coding from the National Cancer Database (comprising >70% of newly-diagnosed cancers in the U.S.), and compared to pediatric and adult populations. Epidemiology and overall survival (estimated by Kaplan-Meier techniques and multivariable Cox regression) were assessed by WHO2016 tumor type. RESULTS 108,705 AYA brain tumor patients were identified (56.9% female), compared to 23,928 pediatric (46.8% female) and 748,272 adult (55.6% female) patients. Among the 69.4% of AYA brain tumors with pathological diagnosis, diffuse gliomas (31.4%), sellar tumors (19.2%), and meningiomas (15.3%) predominated in both sexes. Diffuse glioma (31.4%), sellar (19.2%), cranial nerve (7.3%), and mesenchymal non-meningothelial (4.1%) tumors represented a greater proportion of AYA brain tumors than in either pediatric or adult populations. A majority of all intracranial GCTs (59.2%) and neuronal & mixed neuronal-glial tumors (51.6%) presented during AYA. Although the prevalence of diffuse gliomas was similar between AYAs and adults, AYA gliomas were more likely to be grade 2-3 astrocytomas (38.9% vs 14.3%) and oligodendrogliomas (19.3% vs 4.3%) than in adults. GBMs represented 76.0% of adult diffuse gliomas vs. only 25.7% of AYA diffuse gliomas, but with a similar prevalence of MGMT promoter methylation (40.8% vs 38.4%). Notably, 50.7% of AYA PCNSLs were associated with HIV/AIDS, vs only 7.1% in adults (p< 0.001). CONCLUSIONS The distribution, epidemiology, and survival outcomes of primary brain tumors in the AYA population are distinct from their pediatric and adult counterparts. Notably, AYA infiltrative gliomas were more often of lower grade than adults and AYA PCNSL were far more likely to be associated with HIV/AIDS. Primary brain tumors in AYA patients require specialized management.

Download Full-text

PATH-46. DETECTING MISDIAGNOSED ATYPICAL TERATOID/RHABDOID TUMOR (ATRT) BY DNA METHYLATION-BASED TUMOR CLASSIFICATION

Neuro-Oncology ◽

10.1093/neuonc/noz175.642 ◽

2019 ◽

Vol 21 (Supplement_6) ◽

pp. vi153-vi154 ◽

Cited By ~ 1

Author(s):

Meng-Chang Hsiao ◽

Lauren Holinka ◽

Michael Peracchio ◽

Kevin Kelly ◽

Andrew Hesse ◽

...

Keyword(s):

Dna Methylation ◽

Brain Tumor ◽

Rhabdoid Tumor ◽

Methylation Profile ◽

Histopathological Diagnosis ◽

Tumor Type ◽

Atypical Teratoid Rhabdoid Tumor ◽

Nucleosome Remodeling ◽

Tumor Diagnostics ◽

Atypical Teratoid

Abstract Brain tumor diagnostics is achieved by combining morphology assessment and biomarker identification. However, the inter-observer variability of the histopathological diagnosis and lack of distinctive biomarkers makes these diagnoses particularly challenging. Recently, the potential for methylation-based characterization has been developed for brain tumors. To provide accurate and efficient brain tumor diagnostics, The Jackson Laboratory developed a DNA methylation array (Illumina Infinium HD Methylation EPIC), combining the “Classifier” established by the German Cancer Research Center (DKFZ), which can identify 82 distinct central nervous system tumors. Here we present a pediatric patient referred to us for medulloblastoma subtype classification by the methylation profile. Surprisingly, the array classified the tumor type as atypical teratoid/rhabdoid tumor (ATRT), subclass SHH. Also, the CNV profile indicated the SMARCB1/INI1 heterozygous deletion as part of the chromosome 22q deletion typically seen in ATRT, and absent was the isochromosome 17q commonly associated with medulloblastoma. Immunohistochemistry confirmed the loss of SMARCB1/INI1 expression leading to ATRT. Additionally, our NGS panel found a novel SMARCB1/INI1 frameshift variant c.1150delG (p.Ala384fs) in the last exon (exon 9). Although this variant is not expected to induce nonsense-mediated mRNA decay, it is located in the ATP-dependent nucleosome-remodeling complex domain Sfh1/SNF5 (IPR017393), which is associated with cell proliferation and differentiation. In addition, ClinVar reported several pathogenic/likely pathogenic variants in this domain, further supporting SMARCB1/INI1 c.1150delG which, together with the SMARCB1/INI1 deletion, may contribute to the loss of SMARCB1/INI1 expression, resulting in ATRT. We have identified the ATRT misclassification by the methylation profile and characterized a novel SMARCB1/INI1 variant c.1150delG most likely contributing to ATRT. We further propose that the DNA methylation profile will significantly improve the brain tumor diagnostics for cases with ambiguous, or contradictory histology and molecular profiles.

Download Full-text

PS1 - 176 Where Have All the Fat Cells Gone? A Comparative Analysis of Adiposity Patterns in Childhood Brain Tumor Survivors and Non-Cancer Controls

Canadian Journal of Neurological Sciences / Journal Canadien des Sciences Neurologiques ◽

10.1017/cjn.2016.359 ◽

2016 ◽

Vol 43 (S4) ◽

pp. S11-S11 ◽

Cited By ~ 1

Author(s):

K.W. Wang ◽

E. Kearsley ◽

N. Falzone ◽

A. Fleming ◽

S. Burrow ◽

...

Keyword(s):

Physical Activity ◽

Brain Tumor ◽

Brain Tumors ◽

Survival Rates ◽

Visceral Adiposity ◽

Activity Levels ◽

Tumor Type ◽

Technological Advances ◽

Cardiometabolic Disorders ◽

Fried Foods

Brain tumors are the most common solid tumors in children in Canada. While technological advances have increased their survival rates, survivors of childhood brain tumors (SCBT) often develop obesity, which can reduce lifespan and quality of life. While adiposity is a known factor for cardiometabolic disorders in the general population, adiposity patterns in SCBT have not been determined. This study aims to investigate how adiposity patterns differ between SCBT and non-cancer controls, and how lifestyle and treatment factors may contribute to these patterns. Methods: Fifty-nine SCBT and 108 non-cancer controls were recruited from the clinics at McMaster Children’s Hospital. Sociodemographic and lifestyle details were collected using standardized tools to assess diet, physical activity, and sleep. Brain tumor type, location and treatment details were obtained from medical records. Total and visceral adiposity were determined by total fat mass (FM) as well as waist-to-hip (WHR) and waist-to-height ratio (WHTR). Results: SCBT have higher total and visceral adiposity, while BMI is similar to controls. Female SCBT who received radiotherapy and/or chemotherapy have higher adiposity. A dietary pattern of white bread and fried foods with low dark bread was positively associated with adiposity. Lower physical activity levels, but not sleep durations, were associated with higher adiposity. Conclusion: SCBT have higher visceral and total adiposity than non-cancer controls. Sex, chemoradiotherapy, high fat diet, and physical inactivity, can contribute to these adiposity patterns. These results provide multiple points of entry to design interventions that reduce adiposity, and may improve long-term outcomes in SCBT.

Download Full-text

Cerebrospinal Fluid MicroRNA Signatures as Diagnostic Biomarkers in Brain Tumors

Cancers ◽

10.3390/cancers11101546 ◽

2019 ◽

Vol 11 (10) ◽

pp. 1546 ◽

Cited By ~ 8

Author(s):

Alena Kopkova ◽

Jiri Sana ◽

Tana Machackova ◽

Marek Vecera ◽

Lenka Radova ◽

...

Keyword(s):

Cerebrospinal Fluid ◽

Brain Tumor ◽

Brain Tumors ◽

Low Grade ◽

Tumor Type ◽

Tumor Patients ◽

Primary Tumors ◽

Mirna Profiling ◽

Tumor Types ◽

Cns Malignancies

Central nervous system (CNS) malignancies include primary tumors that originate within the CNS as well as secondary tumors that develop as a result of metastatic spread. Circulating microRNAs (miRNAs) were found in almost all human body fluids including cerebrospinal fluid (CSF), and they seem to be highly stable and resistant to even extreme conditions. The overall aim of our study was to identify specific CSF miRNA patterns that could differentiate among brain tumors. These new biomarkers could potentially aid borderline or uncertain imaging results onto diagnosis of CNS malignancies, avoiding most invasive procedures such as stereotactic biopsy or biopsy. In total, 175 brain tumor patients (glioblastomas, low-grade gliomas, meningiomas and brain metastases), and 40 non-tumor patients with hydrocephalus as controls were included in this prospective monocentric study. Firstly, we performed high-throughput miRNA profiling (Illumina small RNA sequencing) on a discovery cohort of 70 patients and 19 controls and identified specific miRNA signatures of all brain tumor types tested. Secondly, validation of 9 candidate miRNAs was carried out on an independent cohort of 105 brain tumor patients and 21 controls using qRT-PCR. Based on the successful results of validation and various combination patterns of only 5 miRNA levels (miR-30e, miR-140, let-7b, mR-10a and miR-21-3p) we proposed CSF-diagnostic scores for each tumor type which enabled to distinguish them from healthy donors and other tumor types tested. In addition to this primary diagnostic tool, we described the prognostic potential of the combination of miR-10b and miR-196b levels in CSF of glioblastoma patients. In conclusion, we performed the largest study so far focused on CSF miRNA profiling in patients with brain tumors, and we believe that this new class of biomarkers have a strong potential as a diagnostic and prognostic tool in these patients.

Download Full-text

3D extracellular matrix microenvironment in bioengineered tissue models of primary pediatric and adult brain tumors

Nature Communications ◽

10.1038/s41467-019-12420-1 ◽

2019 ◽

Vol 10 (1) ◽

Cited By ~ 18

Author(s):

Disha Sood ◽

Min Tang-Schomer ◽

Dimitra Pouli ◽

Craig Mizzoni ◽

Nicole Raia ◽

...

Keyword(s):

Extracellular Matrix ◽

Brain Tumor ◽

Brain Tumors ◽

Adult Brain ◽

Metabolic Imaging ◽

Tumor Type ◽

Label Free ◽

Pediatric Ependymoma ◽

Extracellular Lipid ◽

Tumor Cell Behavior

Abstract Dynamic alterations in the unique brain extracellular matrix (ECM) are involved in malignant brain tumors. Yet studies of brain ECM roles in tumor cell behavior have been difficult due to lack of access to the human brain. We present a tunable 3D bioengineered brain tissue platform by integrating microenvironmental cues of native brain-derived ECMs and live imaging to systematically evaluate patient-derived brain tumor responses. Using pediatric ependymoma and adult glioblastoma as examples, the 3D brain ECM-containing microenvironment with a balance of cell-cell and cell-matrix interactions supports distinctive phenotypes associated with tumor type-specific and ECM-dependent patterns in the tumor cells’ transcriptomic and release profiles. Label-free metabolic imaging of the composite model structure identifies metabolically distinct sub-populations within a tumor type and captures extracellular lipid-containing droplets with potential implications in drug response. The versatile bioengineered 3D tumor tissue system sets the stage for mechanistic studies deciphering microenvironmental role in brain tumor progression.

Download Full-text

Brain Tumor Classification of MRI Images Using Deep Convolutional Neural Network

Traitement du signal ◽

10.18280/ts.380428 ◽

2021 ◽

Vol 38 (4) ◽

pp. 1171-1179

Author(s):

Swaraja Kuraparthi ◽

Madhavi K. Reddy ◽

C.N. Sujatha ◽

Himabindu Valiveti ◽

Chaitanya Duggineni ◽

...

Keyword(s):

Neural Network ◽

Brain Tumor ◽

Magnetic Resonance ◽

Data Augmentation ◽

Magnetic Resonance Images ◽

Tumor Classification ◽

Support Vector ◽

Svm Classifier ◽

Tumor Type

Manual tumor diagnosis from magnetic resonance images (MRIs) is a time-consuming procedure that may lead to human errors and may lead to false detection and classification of the tumor type. Therefore, to automatize the complex medical processes, a deep learning framework is proposed for brain tumor classification to ease the task of doctors for medical diagnosis. Publicly available datasets such as Kaggle and Brats are used for the analysis of brain images. The proposed model is implemented on three pre-trained Deep Convolution Neural Network architectures (DCNN) such as AlexNet, VGG16, and ResNet50. These architectures are the transfer learning methods used to extract the features from the pre-trained DCNN architecture, and the extracted features are classified by using the Support Vector Machine (SVM) classifier. Data augmentation methods are applied on Magnetic Resonance images (MRI) to avoid the network from overfitting. The proposed methodology achieves an overall accuracy of 98.28% and 97.87% without data augmentation and 99.0% and 98.86% with data augmentation for Kaggle and Brat's datasets, respectively. The Area Under Curve (AUC) for Receiver Operator Characteristic (ROC) is 0.9978 and 0.9850 for the same datasets. The result shows that ResNet50 performs best in the classification of brain tumors when compared with the other two networks.

Download Full-text

Expression and clinical importance of a newly discovered alternative splice variant of the gene for acrosin binding protein found in human brain tumors

Asian Biomedicine ◽

10.1515/abm-2020-0033 ◽

2020 ◽

Vol 14 (6) ◽

pp. 243-252

Author(s):

Baolong Zheng

Keyword(s):

Brain Tumor ◽

Brain Tumors ◽

Splice Variant ◽

Binding Protein ◽

Clinical Importance ◽

Seminiferous Tubules ◽

Tumor Type ◽

Splicing Variant ◽

Tumor Tissues ◽

Significant Difference

AbstractBackgroundAcrosin binding protein (ACRBP) is a member of the cancer–testis antigen (CTA) family. Normally, ACRBP mRNA is expressed only in seminiferous tubules, while abnormally it is expressed in various types of cancers in tumor tissues, such as brain tumor.ObjectivesTo determine the expression and clinical impact of a newly discovered splice variant of ACRBP in brain tumor.MethodsTotal RNA was extracted and reverse transcribed from 92 tumor specimens and 3 cell lines. Primers were designed to determine the expression of the new splice variant in all the samples. Quantitative real-time PCR (qPCR) was conducted for samples positive in reverse transcriptase-PCR. Association of the expression of ACRBP with the clinicopathological features of the various brain tumors was assessed statistically.ResultsThe primers identified a newly discovered splice variant of ACRBP named ACRBP-V5a. The proportions of samples of the various brain tumor types positive for the ACRBP-V5a splicing variant were as follows: astrocytoma 10/33 (30%), glioblastoma 10/30 (33%), medulloblastoma 14/29 (48%), all tumors 34/92 (37%). Although we did not find a significant difference in the proportions of samples of various types of brain tumor tissues positive for the new splice variant (P > 0.05), levels of expression of the ACRBP-V5a splice variant were significantly different for tumor grade (P = 0.01) and tumor type (P = 0.02).ConclusionsA newly discovered splice variant, ACRBP-V5a, is present in brain tumor. The new splicing variant may have discriminative value and potential importance in molecular-targeted therapy for brain tumors.

Download Full-text

OMIC-04. IDENTIFICATION AND CHARACTERIZATION OF CIRCULATING RNAS (CODING AND NONCODING) AND METABOLITES IN CEREBROSPINAL FLUID IN MEDULLOBLASTOMA PATIENTS

Neuro-Oncology ◽

10.1093/neuonc/noab090.151 ◽

2021 ◽

Vol 23 (Supplement_1) ◽

pp. i37-i37

Author(s):

Bongyong Lee ◽

Stacie Stapleton ◽

Rudramani Pokhrel ◽

Chetan Bettegowda ◽

George Jallo ◽

...

Keyword(s):

Cerebrospinal Fluid ◽

Brain Tumor ◽

Metabolic Pathways ◽

Tumor Tissue ◽

Tumor Type ◽

Molecular Signatures ◽

Rna Seq ◽

Neuro Imaging ◽

Identification And Characterization

Abstract Medulloblastoma (MB) is the most common malignant brain tumor in children, and monitoring patients for treatment response and recurrence can be challenging with available current technologies in neuro-imaging and performing a biopsy to confirm response or recurrence carries risks, whereas cerebrospinal fluid (CSF) can be obtained with a little invasiveness. MB has altered cellular metabolism due to changes in gene expression, therefore, we hypothesized that any changes in MB cells lead to changes in cell-free transcripts and metabolites in CSF. To test this, we applied RNA-sequencing and mass spectrometry to analyze transcripts and metabolites including lipid in CSF from patients with different sub-groups of MB tumors (i.e., WNT, SHH, G3/4, G4, and unknown) and compared them to non-cancerous CSF. Tumor and sub-group specific transcriptomic and metabolic signatures were shown by unsupervised hierarchical clustering facilitating tumor type differentiation. By comparison with previously published tumor tissue RNA-seq data, we were able to identify a group of upregulated molecular signatures in both tumor tissue and CSF. We also identified a group of lipids that differentiate each MB sub-group from normal CSF, and Pathway analysis confirmed alterations in multiple metabolic pathways. Finally, we attempted to integrate RNA-seq data with lipidomics data, and results depict that the combinatorial analysis of CSF RNAs and metabolites can be useful in diagnosing and monitoring patients with MB tumors. (This research was conducted using samples made available by The Children’s Brain Tumor Network.)

Download Full-text