Big data visualization identifies the multidimensional molecular landscape of human gliomas

Visualizing genomic characteristics across an RNA-Seq based reference landscape of normal and neoplastic brain

In order to better understand the relationship between normal and neoplastic brain, we combined five publicly available large-scale datasets, correcting for batch effects and applying Uniform Manifold Approximation and Projection (UMAP) to RNA-Seq data. We assembled a reference Brain-UMAP including 702 adult gliomas, 802 pediatric tumors and 1409 healthy normal brain samples, which can be utilized to investigate the wealth of information obtained from combining several publicly available datasets to study a single organ site. Normal brain regions and tumor types create distinct clusters and because the landscape is generated by RNA-Seq, comparative gene expression profiles and gene ontology patterns are readily evident. To our knowledge, this is the first meta-analysis that allows for comparison of gene expression and pathways of interest across adult gliomas, pediatric brain tumors, and normal brain regions. We provide access to this resource via the open source, interactive online tool Oncoscape, where the scientific community can readily visualize clinical metadata, gene expression patterns, gene fusions, mutations, and copy number patterns for individual genes and pathway over this reference landscape.

An RNA seq-based reference landscape of human normal and neoplastic brain

In order to better understand the relationship between normal and neoplastic brain, we combined five publicly available large-scale datasets, correcting for batch effects and applying Uniform Manifold Approximation and Projection (UMAP) to RNA-seq data. We assembled a reference Brain-UMAP including 702 adult gliomas, 802 pediatric tumors and 1409 healthy normal brain samples, which can be utilized to investigate the wealth of information obtained from combining several publicly available datasets to study a single organ site. Normal brain regions and tumor types create distinct clusters and because the landscape is generated by RNA seq, comparative gene expression profiles and gene ontology patterns are readily evident. To our knowledge, this is the first meta-analysis that allows for comparison of gene expression and pathways of interest across adult gliomas, pediatric brain tumors, and normal brain regions. We provide access to this resource via the open source, interactive online tool Oncoscape, where the scientific community can readily visualize clinical metadata, gene expression patterns, gene fusions, mutations, and copy number patterns for individual genes and pathway over this reference landscape.

An RNA seq-based reference landscape of human normal and neoplastic brain

In order to better understand the relationship between normal and neoplastic brain, we combined five publicly available large-scale datasets, correcting for batch effects and applying Uniform Manifold Approximation and Projection (UMAP) to RNA-seq data. We assembled a reference Brain-UMAP including 702 adult gliomas, 802 pediatric tumors and 1409 healthy normal brain samples, which can be utilized to investigate the wealth of information obtained from combining several publicly available datasets to study a single organ site. Normal brain regions and tumor types create distinct clusters and because the landscape is generated by RNA seq, comparative gene expression profiles and gene ontology patterns are readily evident. To our knowledge, this is the first meta-analysis that allows for comparison of gene expression and pathways of interest across adult gliomas, pediatric brain tumors, and normal brain regions. We provide access to this resource via the open source, interactive online tool Oncoscape, where the scientific community can readily visualize clinical metadata, gene expression patterns, gene fusions, mutations, and copy number patterns for individual genes and pathway over this reference landscape.

Machine learning modeling of genome-wide copy number alteration signatures reliably predicts IDH mutational status in adult diffuse glioma

Knowledge of 1p/19q-codeletion and IDH1/2 mutational status is necessary to interpret any investigational study of diffuse gliomas in the modern era. While DNA sequencing is the gold standard for determining IDH mutational status, genome-wide methylation arrays and gene expression profiling have been used for surrogate mutational determination. Previous studies by our group suggest that 1p/19q-codeletion and IDH mutational status can be predicted by genome-wide somatic copy number alteration (SCNA) data alone, however a rigorous model to accomplish this task has yet to be established. In this study, we used SCNA data from 786 adult diffuse gliomas in The Cancer Genome Atlas (TCGA) to develop a two-stage classification system that identifies 1p/19q-codeleted oligodendrogliomas and predicts the IDH mutational status of astrocytic tumors using a machine-learning model. Cross-validated results on TCGA SCNA data showed near perfect classification results. Furthermore, our astrocytic IDH mutation model validated well on four additional datasets (AUC = 0.97, AUC = 0.99, AUC = 0.95, AUC = 0.96) as did our 1p/19q-codeleted oligodendroglioma screen on the two datasets that contained oligodendrogliomas (MCC = 0.97, MCC = 0.97). We then retrained our system using data from these validation sets and applied our system to a cohort of REMBRANDT study subjects for whom SCNA data, but not IDH mutational status, is available. Overall, using genome-wide SCNAs, we successfully developed a system to robustly predict 1p/19q-codeletion and IDH mutational status in diffuse gliomas. This system can assign molecular subtype labels to tumor samples of retrospective diffuse glioma cohorts that lack 1p/19q-codeletion and IDH mutational status, such as the REMBRANDT study, recasting these datasets as validation cohorts for diffuse glioma research.

Magnetic Resonance Features of Lower-grade Gliomas in Prediction of the Reverse Phase Protein A

OBJECTIVES

The Cancer Genome Atlas Research Network identified 4 novel protein expression-defined subgroups in patients with lower-grade gliomas (LGGs). The RPPA3 subtype had high levels of Epidermal Growth Factor Receptor and Human epidermal growth factor receptor-2, further increasing the chances for targeted therapy. In this study, we aimed to explore the relationships between magnetic resonance features and reverse phase protein array (RPPA) subtypes (R1-R4).

METHODS

Survival estimates for the Cancer Genome Atlas cohort were generated using the Kaplan-Meier method and time-dependent receiver operating characteristic curves. A total of 153 patients with LGG with brain magnetic resonance imaging from The Cancer Imaging Archive were retrospectively analyzed. Least absolute shrinkage and selection operator algorithm was used to reduce the feature dimensions of the RPPA3 subtype.

RESULTS

A total of 51 (33.3%) RPPA1 subtype, 42 (27.4) RPPA2 subtype, 19 (12.4%) RPPA3 subtype, and 38 (24.8%) RPPA4 subtype were identified. On multivariate logistic regression analysis, subventricular zone involvement [odds ratio (OR), 0.370; P = 0.006; 95% confidence interval (CI), 0.181-0.757) was associated with RPPA1 subtype [area under the curve (AUC), 0.598]. Volume of 60 cm3 or greater (OR, 5.174; P < 0.001; 95% CI, 2.182-12.267) was associated with RPPA2 subtype (AUC, 0.684). Proportion contrast-enhanced tumor greater than 5% (OR, 4.722; P = 0.010; 95% CI, 1.456-15.317), extranodular growth (OR, 5.524; P = 0.010; 95% CI, 1.509-20.215), and L/CS ratio equal to or greater than median (OR, 0.132; P = 0.003; 95% CI, 0.035-0.500) were associated with RPPA3 subtype (AUC, 0.825). Proportion contrast-enhanced tumor greater than 5% (OR, 0.206; P = 0.005; 95% CI, 0.068-0.625) was associated with RPPA4 subtype (AUC, 0.638). For the prediction of RPPA3 subtype, the nomogram showed good discrimination, with an AUC of 0.825 (95% CI, 0.711-0.939) and was well calibrated. The RPPA3 subtype was associated with shortest mean overall survival (RPPA3 subtype vs other: 613 vs 873 days; P < 0.05). The time-dependent receiver operating characteristic curves for the RPPA3 subtype was 0.72 (95% CI, 0.60-0.84) for survival at 1 year. Decision curve analysis indicated that prediction for the RPPA3 model was clinically useful.

CONCLUSIONS

The RPPA3 subtype is an unfavorable prognostic biomarker for overall survival in patients with LGG. Radiogenomics analysis of magnetic resonance features can predict the RPPA subtype preoperatively and may be of clinical value in tailoring the management strategies in patients with LGG.

Radiogenomic modeling predicts survival-associated prognostic groups in glioblastoma

Background

Combined whole-exome sequencing (WES) and somatic copy number alteration (SCNA) information can separate isocitrate dehydrogenase (IDH)1/2-wildtype glioblastoma into two prognostic molecular subtypes, which cannot be distinguished by epigenetic or clinical features. The potential for radiographic features to discriminate between these molecular subtypes has yet to be established.

Methods

Radiologic features (n = 35 340) were extracted from 46 multisequence, pre-operative magnetic resonance imaging (MRI) scans of IDH1/2-wildtype glioblastoma patients from The Cancer Imaging Archive (TCIA), all of whom have corresponding WES/SCNA data. We developed a novel feature selection method that leverages the structure of extracted MRI features to mitigate the dimensionality challenge posed by the disparity between a large number of features and the limited patients in our cohort. Six traditional machine learning classifiers were trained to distinguish molecular subtypes using our feature selection method, which was compared to least absolute shrinkage and selection operator (LASSO) feature selection, recursive feature elimination, and variance thresholding.

Results

We were able to classify glioblastomas into two prognostic subgroups with a cross-validated area under the curve score of 0.80 (±0.03) using ridge logistic regression on the 15-dimensional principle component analysis (PCA) embedding of the features selected by our novel feature selection method. An interrogation of the selected features suggested that features describing contours in the T2 signal abnormality region on the T2-weighted fluid-attenuated inversion recovery (FLAIR) MRI sequence may best distinguish these two groups from one another.

Conclusions

We successfully trained a machine learning model that allows for relevant targeted feature extraction from standard MRI to accurately predict molecularly-defined risk-stratifying IDH1/2-wildtype glioblastoma patient groups.

Clustering Analysis Supports the Detection of Biological Processes Related to Autism Spectrum Disorder

Genome sequencing has identified a large number of putative autism spectrum disorder (ASD) risk genes, revealing possible disrupted biological pathways; however, the genetic and environmental underpinnings of ASD remain mostly unanswered. The presented methodology aimed to identify genetically related clusters of ASD individuals. By using the VariCarta dataset, which contains data retrieved from 13,069 people with ASD, we compared patients pairwise to build “patient similarity matrices”. Hierarchical-agglomerative-clustering and heatmapping were performed, followed by enrichment analysis (EA). We analyzed whole-genome sequencing retrieved from 2062 individuals, and isolated 11,609 genetic variants shared by at least two people. The analysis yielded three clusters, composed, respectively, by 574 (27.8%), 507 (24.6%), and 650 (31.5%) individuals. Overall, 4187 variants (36.1%) were common to the three clusters. The EA revealed that the biological processes related to the shared genetic variants were mainly involved in neuron projection guidance and morphogenesis, cell junctions, synapse assembly, and in observational, imitative, and vocal learning. The study highlighted genetic networks, which were more frequent in a sample of people with ASD, compared to the overall population. We suggest that itemizing not only single variants, but also gene networks, might support ASD etiopathology research. Future work on larger databases will have to ascertain the reproducibility of this methodology.

Advancing computer-aided drug discovery (CADD) by big data and data-driven machine learning modeling

Advancing a new drug to market requires substantial investments in time as well as financial resources. Crucial bioactivities for drug candidates, including their efficacy, pharmacokinetics (PK), and adverse effects, need to be investigated during drug development. With advancements in chemical synthesis and biological screening technologies over the past decade, a large amount of biological data points for millions of small molecules have been generated and are stored in various databases. These accumulated data, combined with new machine learning (ML) approaches, such as deep learning, have shown great potential to provide insights into relevant chemical structures to predict in vitro, in vivo, and clinical outcomes, thereby advancing drug discovery and development in the big data era.

A Panel of Synapse-Related Genes as a Biomarker for Gliomas

Gliomas are the most common primary brain cancers. In recent years, IDH mutation and 1p/19q codeletion have been suggested as biomarkers for the diagnosis, treatment, and prognosis of gliomas. However, these biomarkers are only effective for a part of glioma patients, and thus more biomarkers are still emergently needed. Recently, an electrochemical communication between normal neurons and glioma cells by neuro-glioma synapse has been reported. Moreover, it was discovered that breast-to-brain metastasis tumor cells have pseudo synapses with neurons, and these synapses were indicated to promote tumor progression and metastasis. Based on the above observations, we first curated a panel of 17 synapse-related genes and then proposed a metric, synapse score to quantify the "stemness" for each sample of 12 glioma gene expression datasets from TCGA, CGGA, and GEO. Strikingly, synapse score showed excellent predictive ability for the prognosis, diagnosis, and grading of gliomas. Moreover, being compared with the two established biomarkers, IDH mutation and 1p/19q codeletion, synapse score demonstrated independent and better predictive performance. In conclusion, this study proposed a quantitative method, synapse score, as an efficient biomarker for monitoring gliomas.

Identification of relevant genetic alterations in cancer using topological data analysis

Large-scale cancer genomic studies enable the systematic identification of mutations that lead to the genesis and progression of tumors, uncovering the underlying molecular mechanisms and potential therapies. While some such mutations are recurrently found in many tumors, many others exist solely within a few samples, precluding detection by conventional recurrence-based statistical approaches. Integrated analysis of somatic mutations and RNA expression data across 12 tumor types reveals that mutations of cancer genes are usually accompanied by substantial changes in expression. We use topological data analysis to leverage this observation and uncover 38 elusive candidate cancer-associated genes, including inactivating mutations of the metalloproteinase ADAMTS12 in lung adenocarcinoma. We show that ADAMTS12-/- mice have a five-fold increase in the susceptibility to develop lung tumors, confirming the role of ADAMTS12 as a tumor suppressor gene. Our results demonstrate that data integration through topological techniques can increase our ability to identify previously unreported cancer-related alterations.

Coding of Glioblastoma Progression and Therapy Resistance through Long Noncoding RNAs

Glioblastoma is the most aggressive and lethal primary brain malignancy, with an average patient survival from diagnosis of 14 months. Glioblastoma also usually progresses as a more invasive phenotype after initial treatment. A major step forward in our understanding of the nature of glioblastoma was achieved with large-scale expression analysis. However, due to genomic complexity and heterogeneity, transcriptomics alone is not enough to define the glioblastoma “fingerprint”, so epigenetic mechanisms are being examined, including the noncoding genome. On the basis of their tissue specificity, long noncoding RNAs (lncRNAs) are being explored as new diagnostic and therapeutic targets. In addition, growing evidence indicates that lncRNAs have various roles in resistance to glioblastoma therapies (e.g., MALAT1, H19) and in glioblastoma progression (e.g., CRNDE, HOTAIRM1, ASLNC22381, ASLNC20819). Investigations have also focused on the prognostic value of lncRNAs, as well as the definition of the molecular signatures of glioma, to provide more precise tumor classification. This review discusses the potential that lncRNAs hold for the development of novel diagnostic and, hopefully, therapeutic targets that can contribute to prolonged survival and improved quality of life for patients with glioblastoma.

Cellular Plasticity and Tumor Microenvironment in Gliomas: The Struggle to Hit a Moving Target

Brain tumors encompass a diverse group of neoplasias arising from different cell lineages. Tumors of glial origin have been the subject of intense research because of their rapid and fatal progression. From a clinical point of view, complete surgical resection of gliomas is highly difficult. Moreover, the remaining tumor cells are resistant to traditional therapies such as radio- or chemotherapy and tumors always recur. Here we have revised the new genetic and epigenetic classification of gliomas and the description of the different transcriptional subtypes. In order to understand the progression of the different gliomas we have focused on the interaction of the plastic tumor cells with their vasculature-rich microenvironment and with their distinct immune system. We believe that a comprehensive characterization of the glioma microenvironment will shed some light into why these tumors behave differently from other cancers. Furthermore, a novel classification of gliomas that could integrate the genetic background and the cellular ecosystems could have profound implications in the efficiency of current therapies as well as in the development of new treatments.

A kinase-deficient NTRK2 splice variant predominates in glioma and amplifies several oncogenic signaling pathways

Independent scientific achievements have led to the discovery of aberrant splicing patterns in oncogenesis, while more recent advances have uncovered novel gene fusions involving neurotrophic tyrosine receptor kinases (NTRKs) in gliomas. The exploration of NTRK splice variants in normal and neoplastic brain provides an intersection of these two rapidly evolving fields. Tropomyosin receptor kinase B (TrkB), encoded NTRK2, is known for critical roles in neuronal survival, differentiation, molecular properties associated with memory, and exhibits intricate splicing patterns and post-translational modifications. Here, we show a role for a truncated NTRK2 splice variant, TrkB.T1, in human glioma. TrkB.T1 enhances PDGF-driven gliomas in vivo, augments PDGF-induced Akt and STAT3 signaling in vitro, while next generation sequencing broadly implicates TrkB.T1 in the PI3K signaling cascades in a ligand-independent fashion. These TrkB.T1 findings highlight the importance of expanding upon whole gene and gene fusion analyses to include splice variants in basic and translational neuro-oncology research.

Next Generation Sequencing and Machine Learning Technologies Are Painting the Epigenetic Portrait of Glioblastoma

Even with a rare occurrence of only 1.35% of cancer cases in the United States of America, brain tumors are considered as one of the most lethal malignancies. The most aggressive and invasive type of brain tumor, glioblastoma, accounts for 60–70% of all gliomas and presents with life expectancy of only 12–18 months. Despite trimodal treatment and advances in diagnostic and therapeutic methods, there are no significant changes in patient outcome. Our understanding of glioblastoma was significantly improved with the introduction of next generation sequencing technologies. This led to the identification of different genetic and molecular subtypes, which greatly improve glioblastoma diagnosis. Still, because of the poor life expectancy, novel diagnostic, and treatment methods are broadly explored. Epigenetic modifications like methylation and changes in histone acetylation are such examples. Recently, in addition to genetic and molecular characteristics, epigenetic profiling of glioblastomas is also used for sample classification. Further advancement of next generation sequencing technologies is expected to identify in detail the epigenetic signature of glioblastoma that can open up new therapeutic opportunities for glioblastoma patients. This should be complemented with the use of computational power i.e., machine and deep learning algorithms for objective diagnostics and design of individualized therapies. Using a combination of phenotypic, genotypic, and epigenetic parameters in glioblastoma diagnostics will bring us closer to precision medicine where therapies will be tailored to suit the genetic profile and epigenetic signature of the tumor, which will grant longer life expectancy and better quality of life. Still, a number of obstacles including potential bias, availability of data for minorities in heterogeneous populations, data protection, and validation and independent testing of the learning algorithms have to be overcome on the way.

Tumor Cell Invasion in Glioblastoma

Glioblastoma (GBM) is a particularly devastating tumor with a median survival of about 16 months. Recent research has revealed novel insights into the outstanding heterogeneity of this type of brain cancer. However, all GBM subtypes share the hallmark feature of aggressive invasion into the surrounding tissue. Invasive glioblastoma cells escape surgery and focal therapies and thus represent a major obstacle for curative therapy. This review aims to provide a comprehensive understanding of glioma invasion mechanisms with respect to tumor-cell-intrinsic properties as well as cues provided by the microenvironment. We discuss genetic programs that may influence the dissemination and plasticity of GBM cells as well as their different invasion patterns. We also review how tumor cells shape their microenvironment and how, vice versa, components of the extracellular matrix and factors from non-neoplastic cells influence tumor cell motility. We further discuss different research platforms for modeling invasion. Finally, we highlight the importance of accounting for the complex interplay between tumor cell invasion and treatment resistance in glioblastoma when considering new therapeutic approaches.

Translating Basic Science Discoveries into Improved Outcomes for Glioblastoma

Members of the scientific and clinical neuro-oncology community met in April 2019 to discuss the current challenges and opportunities associated with translating basic science discoveries in glioblastoma for improved survival of patients. A summary of key points of these discussions is presented in this article.

Copy number profiling across glioblastoma populations has implications for clinical trial design

Background

Copy number alterations form prognostic molecular subtypes of glioblastoma with clear differences in median overall survival. In this study, we leverage molecular data from several glioblastoma cohorts to define the distribution of copy number subtypes across random cohorts as well as cohorts with selection biases for patients with inherently better outcome.

Methods

Copy number subtype frequency was established for 4 glioblastoma patient cohorts. Two randomly selected cohorts include The Cancer Genome Atlas (TCGA) and the German Glioma Network (GGN). Two more selective cohorts include the phase II trial ARTE in elderly patients with newly diagnosed glioblastoma and a multi-institutional cohort focused on paired resected initial/recurrent glioblastoma. The paired initial/recurrent cohort also had exome data available, which allowed for evaluation of multidimensional scaling analysis.

Results

Smaller selective glioblastoma cohorts are enriched for copy number subtypes that are associated with better survival, reflecting the selection of patients who do well enough to enter a clinical trial or who are deemed well enough to undergo resection at recurrence. Adding exome data to copy number data provides additional data reflective of outcome.

Conclusions

The overall outcome for diffuse glioma patients is predicted by DNA structure at initial tumor resection. Molecular signature shifts across glioblastoma populations reflect the inherent bias of patient selection toward longer survival in clinical trials. Therefore it may be important to include molecular profiling, including copy number, when enrolling patients for clinical trials in order to balance arms and extrapolate relevance to the general glioblastoma population.

Layer-based visualization and biomedical information exploration of multi-channel large histological data

BACKGROUND AND OBJECTIVE

Modern microscopes can acquire multi-channel large histological data from tissues of human beings or animals, which contain rich biomedical information for disease diagnosis and biological feature analysis. However, due to the large size, fuzzy tissue structure, and complicated multiple elements integrated in the image color space, it is still a challenge for current software systems to effectively calculate histological data, show the inner tissue structures and unveil hidden biomedical information. Therefore, we developed new algorithms and a software platform to address this issue.

METHODS

This paper presents a multi-channel biomedical data computing and visualization system that can efficiently process large 3D histological images acquired from high-resolution microscopes. A novelty of our system is that it can dynamically display a volume of interest and extract tissue information using a layer-based data navigation scheme. During the data exploring process, the actual resolution of the loaded data can be dynamically determined and updated, and data rendering is synchronized in four display windows at each data layer, where 2D textures are extracted from the imaging volume and mapped onto the displayed clipping planes in 3D space.

RESULTS

To test the efficiency and scalability of this system, we performed extensive evaluations using several different hardware systems and large histological color datasets acquired from a CryoViz 3D digital system. The experimental results demonstrated that our system can deliver interactive data navigation speed and display detailed imaging information in real time, which is beyond the capability of commonly available biomedical data exploration software platforms.

CONCLUSION

Taking advantage of both CPU (central processing unit) main memory and GPU (graphics processing unit) graphics memory, the presented software platform can efficiently compute, process and visualize very large biomedical data and enhance data information. The performance of this system can satisfactorily address the challenges of navigating and interrogating volumetric multi-spectral large histological image at multiple resolution levels.

A pan-cancer analysis of driver gene mutations, DNA methylation and gene expressions reveals that chromatin remodeling is a major mechanism inducing global changes in cancer epigenomes

Background

Recent large-scale cancer sequencing studies have discovered many novel cancer driver genes (CDGs) in human cancers. Some studies also suggest that CDG mutations contribute to cancer-associated epigenomic and transcriptomic alterations across many cancer types. Here we aim to improve our understanding of the connections between CDG mutations and altered cancer cell epigenomes and transcriptomes on pan-cancer level and how these connections contribute to the known association between epigenome and transcriptome.

Method

Using multi-omics data including somatic mutation, DNA methylation, and gene expression data of 20 cancer types from The Cancer Genome Atlas (TCGA) project, we conducted a pan-cancer analysis to identify CDGs, when mutated, have strong associations with genome-wide methylation or expression changes across cancer types, which we refer as methylation driver genes (MDGs) or expression driver genes (EDGs), respectively.

Results

We identified 32 MDGs, among which, eight are known chromatin modification or remodeling genes. Many of the remaining 24 MDGs are connected to chromatin regulators through either regulating their transcription or physically interacting with them as potential co-factors. We identified 29 EDGs, 26 of which are also MDGs. Further investigation on target genes’ promoters methylation and expression alteration patterns of these 26 overlapping driver genes shows that hyper-methylation of target genes’ promoters are significantly associated with down-regulation of the same target genes and hypo-methylation of target genes’ promoters are significantly associated with up-regulation of the same target genes.

Conclusion

This finding suggests a pivotal role for genetically driven changes in chromatin remodeling in shaping DNA methylation and gene expression patterns during tumor development.

Electronic supplementary material

The online version of this article (10.1186/s12920-018-0425-z) contains supplementary material, which is available to authorized users.

Next-Generation in vivo Modeling of Human Cancers

Animal models of human cancers played a major role in our current understanding of tumor biology. In pre-clinical oncology, animal models empowered drug target and biomarker discovery and validation. In turn, this resulted in improved care for cancer patients. In the quest for understanding and treating a diverse spectrum of cancer types, technological breakthroughs in genetic engineering and single cell "omics" offer tremendous potential to enhance the informative value of pre-clinical models. Here, I review the state-of-the-art in modeling human cancers with focus on animal models for human malignant gliomas. The review highlights the use of glioma models in dissecting mechanisms of tumor initiation, in the retrospective identification of tumor cell-of-origin, in understanding tumor heterogeneity and in testing the potential of immuno-oncology. I build on the deep review of glioma models as a basis for a more general discussion of the potential ways in which transformative technologies may shape the next-generation of pre-clinical models. I argue that refining animal models along the proposed lines will benefit the success rate of translation for pre-clinical research in oncology.

Using genomics to guide treatment for glioblastoma

Glioblastoma has been shown to have many different genetic mutations found both within and between tumor samples. Molecular testing and genomic sequencing has helped to classify diagnoses and clarify difficult to interpret histopathological specimens. Genomic information also plays a critical role in prognostication for patients, with IDH mutations and MGMT methylation having significant impact of the response to chemotherapy and overall survival of patients. Unfortunately, personalized medicine and targeted therapy against specific mutations have not been shown to improve patient outcomes. As technology continues to improve, exome and RNA sequencing will play a role in the design of clinical trials, classification of patient subgroups and identification of rare mutations that can be targeted by small-molecule inhibitors and biologic agents.

Entropy-based consensus clustering for patient stratification

Motivation

Patient stratification or disease subtyping is crucial for precision medicine and personalized treatment of complex diseases. The increasing availability of high-throughput molecular data provides a great opportunity for patient stratification. Many clustering methods have been employed to tackle this problem in a purely data-driven manner. Yet, existing methods leveraging high-throughput molecular data often suffers from various limitations, e.g. noise, data heterogeneity, high dimensionality or poor interpretability.

Results

Here we introduced an Entropy-based Consensus Clustering (ECC) method that overcomes those limitations all together. Our ECC method employs an entropy-based utility function to fuse many basic partitions to a consensus one that agrees with the basic ones as much as possible. Maximizing the utility function in ECC has a much more meaningful interpretation than any other consensus clustering methods. Moreover, we exactly map the complex utility maximization problem to the classic K -means clustering problem, which can then be efficiently solved with linear time and space complexity. Our ECC method can also naturally integrate multiple molecular data types measured from the same set of subjects, and easily handle missing values without any imputation. We applied ECC to 110 synthetic and 48 real datasets, including 35 cancer gene expression benchmark datasets and 13 cancer types with four molecular data types from The Cancer Genome Atlas. We found that ECC shows superior performance against existing clustering methods. Our results clearly demonstrate the power of ECC in clinically relevant patient stratification.

Availability and implementation

The Matlab package is available at http://scholar.harvard.edu/yyl/ecc .

Contact

yunfu@ece.neu.edu or yyl@channing.harvard.edu.

Supplementary information

Supplementary data are available at Bioinformatics online.

Patient similarity for precision medicine: A systematic review

Evidence-based medicine is the most prevalent paradigm adopted by physicians. Clinical practice guidelines typically define a set of recommendations together with eligibility criteria that restrict their applicability to a specific group of patients. The ever-growing size and availability of health-related data is currently challenging the broad definitions of guideline-defined patient groups. Precision medicine leverages on genetic, phenotypic, or psychosocial characteristics to provide precise identification of patient subsets for treatment targeting. Defining a patient similarity measure is thus an essential step to allow stratification of patients into clinically-meaningful subgroups. The present review investigates the use of patient similarity as a tool to enable precision medicine. 279 articles were analyzed along four dimensions: data types considered, clinical domains of application, data analysis methods, and translational stage of findings. Cancer-related research employing molecular profiling and standard data analysis techniques such as clustering constitute the majority of the retrieved studies. Chronic and psychiatric diseases follow as the second most represented clinical domains. Interestingly, almost one quarter of the studies analyzed presented a novel methodology, with the most advanced employing data integration strategies and being portable to different clinical domains. Integration of such techniques into decision support systems constitutes and interesting trend for future research.

TumorMap: Exploring the Molecular Similarities of Cancer Samples in an Interactive Portal

Vast amounts of molecular data are being collected on tumor samples, which provide unique opportunities for discovering trends within and between cancer subtypes. Such cross-cancer analyses require computational methods that enable intuitive and interactive browsing of thousands of samples based on their molecular similarity. We created a portal called TumorMap to assist in exploration and statistical interrogation of high-dimensional complex "omics" data in an interactive and easily interpretable way. In the TumorMap, samples are arranged on a hexagonal grid based on their similarity to one another in the original genomic space and are rendered with Google's Map technology. While the important feature of this public portal is the ability for the users to build maps from their own data, we pre-built genomic maps from several previously published projects. We demonstrate the utility of this portal by presenting results obtained from The Cancer Genome Atlas project data. Cancer Res; 77(21); e111-4. ©2017 AACR.

Visualizing omics and clinical data: Which challenges for dealing with their variety?

Life sciences are currently going through a great number of transformations raised by the in-going revolution in high-throughput technologies for the acquisition of data. The integration of their high dimensionality, ranging from omics to clinical data, is becoming one of the most challenging stages. It involves inter-disciplinary developments with the aim to move towards an enhanced understanding of human physiology for caring purposes. Biologists, bioinformaticians, physicians and other experts related to the healthcare domain have to accompany each step of the analysis process in order to investigate and expertise these various data. In this perspective, methods related to information visualization are gaining increasing attention within life sciences. The softwares based on these methods are now well recognized to facilitate expert users' success in carrying out their data analysis tasks. This article aims at reviewing the current methods and techniques dedicated to information visualisation and their current use in software development related to omics or/and clinical data.

Intratumoral heterogeneity and clonal evolution in blood malignancies and solid tumors

This meeting held at the University of Barcelona in March 2017, brought together scientists and clinicians worldwide to discuss current and future clinico-biological implications of intratumoral heterogeneity (ITH) and subclonal evolution in cancer diagnosis, patient stratification, and treatment resistance in diagnosis, treatment and follow-up. There was consensus that both longitudinal and tumor multi-region studies in matched samples are needed to better understand the dynamics of ITH. The contribution of the epigenome and microenvironment to ITH and subclone evolution remains understudied. It was recommended to combine computational, pathology and imaging tools to study the role of the microenvironment in subclone selection/evolution.

Putting Glioblastoma in Its Place: IRF3 Inhibits Invasion

With an unsurpassed capacity for invasion into normal brain tissue, glioblastoma multiforme is the most lethal primary brain tumor. New research suggests that altering a subset of extracellular matrix factors, including interferon regulatory factor (IRF)3 and casein kinase (CK)2, may decrease the migratory potential of these aggressive tumors.

Multidimensional scaling of diffuse gliomas: application to the 2016 World Health Organization classification system with prognostically relevant molecular subtype discovery

Recent updating of the World Health Organization (WHO) classification of central nervous system (CNS) tumors in 2016 demonstrates the first organized effort to restructure brain tumor classification by incorporating histomorphologic features with recurrent molecular alterations. Revised CNS tumor diagnostic criteria also attempt to reduce interobserver variability of histological interpretation and provide more accurate stratification related to clinical outcome. As an example, diffuse gliomas (WHO grades II–IV) are now molecularly stratified based upon isocitrate dehydrogenase 1 or 2 (IDH) mutational status, with gliomas of WHO grades II and III being substratified according to 1p/19q codeletion status. For now, grading of diffuse gliomas is still dependent upon histological parameters. Independent of WHO classification criteria, multidimensional scaling analysis of molecular signatures for diffuse gliomas from The Cancer Genome Atlas (TCGA) has identified distinct molecular subgroups, and allows for their visualization in 2-dimensional (2D) space. Using the web-based platform Oncoscape as a tool, we applied multidimensional scaling-derived molecular groups to the 2D visualization of the 2016 WHO classification of diffuse gliomas. Here we show that molecular multidimensional scaling of TCGA data provides 2D clustering that represents the 2016 WHO classification of diffuse gliomas. Additionally, we used this platform to successfully identify and define novel copy-number alteration-based molecular subtypes, which are independent of WHO grading, as well as predictive of clinical outcome. The prognostic utility of these molecular subtypes was further validated using an independent data set of the German Glioma Network prospective glioblastoma patient cohort.

Mutant IDH1 regulates the tumor-associated immune system in gliomas

Amankulor et al. created a syngeneic pair mouse model for mutant IDH1 (muIDH1) and wild-type IDH1 (wtIDH1) gliomas and demonstrated that IDH1 mutations caused down-regulation of leukocyte chemotaxis, resulting in repression of the tumor-associated immune system.

Gliomas harboring mutations in isocitrate dehydrogenase 1/2 (IDH1/2) have the CpG island methylator phenotype (CIMP) and significantly longer patient survival time than wild-type IDH1/2 (wtIDH1/2) tumors. Although there are many factors underlying the differences in survival between these two tumor types, immune-related differences in cell content are potentially important contributors. In order to investigate the role of IDH mutations in immune response, we created a syngeneic pair mouse model for mutant IDH1 (muIDH1) and wtIDH1 gliomas and demonstrated that muIDH1 mice showed many molecular and clinical similarities to muIDH1 human gliomas, including a 100-fold higher concentration of 2-hydroxygluratate (2-HG), longer survival time, and higher CpG methylation compared with wtIDH1. Also, we showed that IDH1 mutations caused down-regulation of leukocyte chemotaxis, resulting in repression of the tumor-associated immune system. Given that significant infiltration of immune cells such as macrophages, microglia, monocytes, and neutrophils is linked to poor prognosis in many cancer types, these reduced immune infiltrates in muIDH1 glioma tumors may contribute in part to the differences in aggressiveness of the two glioma types.