Search

Glioblastoma (GBM) show significant inter- and intra-tumoral heterogeneity, impacting response to treatment and overall survival time of 12-15 months. To study glioblastoma phenotypic heterogeneity, multi-parametric magnetic resonance images (MRI) of 85 glioblastoma patients from The Cancer Genome Atlas were analyzed to characterize tumor-derived spatial habitats for their relationship with outcome (overall survival) and to identify their molecular correlates (i.e., determine associated tumor signaling pathways correlated with imaging-derived habitat measurements). Tumor sub-regions based on four sequences (fluid attenuated inversion recovery, T1-weighted, post-contrast T1-weighted, and T2-weighted) were defined by automated segmentation. From relative intensity of pixels in the 3-dimensional tumor region, "imaging habitats" were identified and analyzed for their association to clinical and genetic data using survival modeling and Dirichlet regression, respectively. Sixteen distinct tumor sub-regions ("spatial imaging habitats") were derived, and those associated with overall survival (denoted "relevant" habitats) in glioblastoma patients were identified. Dirichlet regression implicated each relevant habitat with unique pathway alterations. Relevant habitats also had some pathways and cellular processes in common, including phosphorylation of STAT-1 and natural killer cell activity, consistent with cancer hallmarks. This work revealed clinical relevance of MRI-derived spatial habitats and their relationship with oncogenic molecular mechanisms in patients with GBM. Characterizing the associations between imaging-derived phenotypic measurements with the genomic and molecular characteristics of tumors can enable insights into tumor biology, further enabling the practice of personalized cancer treatment. The analytical framework and workflow demonstrated in this study are inherently scalable to multiple MR sequences.

This study analyzed magnetic resonance imaging (MRI) scans of Glioblastoma (GB) patients to develop an imaging-derived predictive model for assessing the extent of intratumoral CD3 T-cell infiltration. Pre-surgical T1-weighted post-contrast and T2-weighted Fluid-Attenuated-Inversion-Recovery (FLAIR) MRI scans, with corresponding mRNA expression of CD3D/E/G were obtained through The Cancer Genome Atlas (TCGA) for 79 GB patients. The tumor region was contoured and 86 image-derived features were extracted across the T1-post contrast and FLAIR images. Six imaging features-kurtosis, contrast, small zone size emphasis, low gray level zone size emphasis, high gray level zone size emphasis, small zone high gray level emphasis-were found associated with CD3 activity and used to build a predictive model for CD3 infiltration in an independent data set of 69 GB patients (using a 50-50 split for training and testing). For the training set, the image-based prediction model for CD3 infiltration achieved accuracy of 97.1% and area under the curve (AUC) of 0.993. For the test set, the model achieved accuracy of 76.5% and AUC of 0.847. This suggests a relationship between image-derived textural features and CD3 T-cell infiltration enabling the non-invasive inference of intratumoral CD3 T-cell infiltration in GB patients, with potential value for the radiological assessment of response to immune therapeutics.

Technological innovations have produced large multi-modal datasets that include imaging and multi-platform genomics data. Integrative analyses of such data have the potential to reveal important biological and clinical insights into complex diseases like cancer. In this paper, we present Bayesian approaches for integrative analysis of radiological imaging and multi-platform genomic data, where-in our goals are to simultaneously identify genomic and radiomic, that is, radiology-based imaging markers, along with the latent associations between these two modalities, and to detect the overall prognostic relevance of the combined markers. For this task, we propose Radio-iBAG: Radiomics-based Integrative Bayesian Analysis of Multiplatform Genomic Data, a multi-scale Bayesian hierarchical model that involves several innovative strategies: it incorporates integrative analysis of multi-platform genomic data sets to capture fundamental biological relationships; explores the associations between radiomic markers accompanying genomic information with clinical outcomes; and detects genomic and radiomic markers associated with clinical prognosis. We also introduce the use of sparse Principal Component Analysis (sPCA) to extract a sparse set of approximately orthogonal meta-features each containing information from a set of related individual radiomic features, reducing dimensionality and combining like features. Our methods are motivated by and applied to The Cancer Genome Atlas glioblastoma multiforme data set, where-in we integrate magnetic resonance imaging-based biomarkers along with genomic, epigenomic and transcriptomic data. Our model identifies important magnetic resonance imaging features and the associated genomic platforms that are related with patient survival times.

Gliomas are tumors which develop in the brain, the most aggressive of which is glioblastoma multiforme (GBM). Despite extensive research to better understand the underlying biology of GBM and advancements in the treatment of this disease, it has an extremely poor prognosis. Poor outcomes in GBM are due to its molecular and clinical heterogeneity, as in other solid tumors. As imaging approaches have been taken to comprehensively characterize tumors, radiomics has emerged as the concept of extracting quantitative radiologic features and drawing associations with clinical outcomes. Radiomics has the potential to improve the predictive ability of radiological datasets. Tumor radiographs are segmented through manual, semi-automated or fully-automated procedures. Segmentation is followed by feature extraction from the tumor volume. Data analysis and predictive modeling are used to relate image-derived features with clinical outcomes. Substantial progress has already been made in solving many of the technical hurdles inherent in the radiomics process. Advances in sequencing, gene expression profiling and machine learning have increased the resolution of datasets and improved the sensitivity and specificity of computational methods used to analyze them. Numerous logistical, computational and clinical challenges remain to unlocking the full potential of the radiomics approach. To make the approach useful in clinical practice, improved statistical models are needed which relate GBM imaging features with patient outcomes with high specificity/sensitivity. More studies correlating radiomic features with disease outcomes and molecular attributes are also needed to illuminate the tumor biology which gives rise to imaging features and underlie response to therapy.

Intra-tumor heterogeneity is the fundamental challenge in finding a cure for late-stage cancers. Physical biopsies do not sufficiently cover the diversity of molecular phenotypes within the tumor. Treatments are only effective on a subset of vulnerable tumor cells due to the prevalence of tumor stem-like cells. GBM tumors exemplify these general properties of late-stage cancers, with heterogeneous molecular profiles, histology, and radiology. Radiomics aims to characterize disease phenotypes from radiology scans in order to provide an alternative view of tumor heterogeneity, enabling models built from retrospective analysis of radiology scan data, and their integration with clinical data and molecular profiles. Computational histology (histomics) follows a workflow analogous to that of radiomics, with pre-processing, segmentation, feature extraction and analytics. The goal of histomics is to compute cellular morphometry and heterogeneity features from histology datasets. Genomic traits can potentially be inferred from histologic features by analysis of large, linked pathology-genomic data sets. There is also an active investigation of computer vision and machine learning applications to classify gliomas using radiology and histology images. The potential of radiomics, radiogenomics and histomics studies is to advance personalized cancer treatment by enabling interpretation of biological mechanisms underlying imaging phenotypes. These efforts aim to make personalized therapies more accessible. Results from preliminary imaging could direct administration of precision assays to guide treatment, measure treatment response and identify targetable genetic alterations from image-derived phenotype data, across biological scale. Radiomics and histomics promises to revolutionize the practice of personalized medicine, by providing an important complement to molecular strategies.

BACKGROUND: To determine whether computationally derived image-based textural and volumetric tumor features can predict cognitive impairment in patients with temporal lobe glioma. METHODS: Treatment naive patients with glioma (n = 69; 64% glioblastoma; 74% left temporal) completed neurocognitive testing [WAIS-III Digit Span (DS); Controlled Oral Word Association (COWA); HVLT-R Total Recall; Trail Making Test A & B (TMT)]. Demographically adjusted scores that were at or below 0.74). Volumetric and textural tumor features appear predictive of neurocognitive impairment in patients with glioma, with inclusion of textural indices conveying at least marginal improvement in classification accuracy.

PURPOSE: To determine which computationally derived imaging features are correlated with immune gene-signature activity in GBM. Recently, we showed that immune-related genes are enriched within molecular subtypes of GBMs. The significance of immune system response to the tumor is becoming relevant as a therapeutic strategy. In this study, we investigated an imaging-based method for determining the immune system response to GBM. METHODS AND MATERIALS: The study was conducted on pre-surgical Contrast enhanced T1-post and T2-FLAIR MRI images from 83 GBM patients from The Cancer Genome Atlas(TCGA) database. The tumors were segmented semi-automatically (MITK toolkit). Image heterogeneity features were extracted for 3D volumes using in-house MatlabTM scripts . Features include three-dimensional statistical, transform-based and model-based features, pertaining to pixel gray-level heterogeneity measures such as energy, entropy, correlationGenetic programming-based models are used to discriminate subjects according to up- or down-regulation of six immune signatures (measured using Gene Set Enrichment Analysis): immune effector, immune suppression, immune effector process, regulation of the immune effector process, positive regulation of immune system process and negative regulation of immune system process. Receiver Operating Characteristic curves (ROC), and true positive/false positive rates (TPR/FPR) were used to assess the performance of the immune signature status classifier. RESULTS: Based on six different gene-sets associated with immune response, we found that image-derived features are capable of accurately predicting immune signature status in GBMs. ROC analysis reveals that image-features are capable of determining up- or down regulation of diverse immune signatures. The true positive rates (TPR) for each of the six immune signatures are 89%, 83%, 80%, 78%, 72% and 69% respectively (with FPR less than 20%). CONCLUSION: This study presents preliminary evidence that MRI image-derived volumetric and texture features are predictive of immune activity in GBM at the molecular level.

One of the most common and aggressive malignant brain tumors is Glioblastoma multiforme. Despite the multimodality treatment such as radiation therapy and chemotherapy (temozolomide: TMZ), the median survival rate of glioblastoma patient is less than 15 months. In this study, we investigated the association between measures of spatial diversity derived from spatial point pattern analysis of multiparametric magnetic resonance imaging (MRI) data with molecular status as well as 12-month survival in glioblastoma. We obtained 27 measures of spatial proximity (diversity) via spatial point pattern analysis of multiparametric T1 post-contrast and T2 fluid-attenuated inversion recovery MRI data. These measures were used to predict 12-month survival status (≤12 or >12 months) in 74 glioblastoma patients. Kaplan-Meier with receiver operating characteristic analyses was used to assess the relationship between derived spatial features and 12-month survival status as well as molecular subtype status in patients with glioblastoma. Kaplan-Meier survival analysis revealed that 14 spatial features were capable of stratifying overall survival in a statistically significant manner. For prediction of 12-month survival status based on these diversity indices, sensitivity and specificity were 0.86 and 0.64, respectively. The area under the receiver operating characteristic curve and the accuracy were 0.76 and 0.75, respectively. For prediction of molecular subtype status, proneural subtype shows highest accuracy of 0.93 among all molecular subtypes based on receiver operating characteristic analysis. We find that measures of spatial diversity from point pattern analysis of intensity habitats from T1 post-contrast and T2 fluid-attenuated inversion recovery images are associated with both tumor subtype status and 12-month survival status and may therefore be useful indicators of patient prognosis, in addition to providing potential guidance for molecularly-targeted therapies in Glioblastoma multiforme.

We analyzed the spatial diversity of tumor habitats, regions with distinctly different intensity characteristics of a tumor, using various measurements of habitat diversity within tumor regions. These features were then used for investigating the association with a 12-month survival status in glioblastoma (GBM) patients and for the identification of epidermal growth factor receptor (EGFR)-driven tumors. T1 postcontrast and T2 fluid attenuated inversion recovery images from 65 GBM patients were analyzed in this study. A total of 36 spatial diversity features were obtained based on pixel abundances within regions of interest. Performance in both the classification tasks was assessed using receiver operating characteristic (ROC) analysis. For association with 12-month overall survival, area under the ROC curve was 0.74 with confidence intervals [0.630 to 0.858]. The sensitivity and specificity at the optimal operating point ([Formula: see text]) on the ROC were 0.59 and 0.75, respectively. For the identification of EGFR-driven tumors, the area under the ROC curve (AUC) was 0.85 with confidence intervals [0.750 to 0.945]. The sensitivity and specificity at the optimal operating point ([Formula: see text]) on the ROC were 0.76 and 0.83, respectively. Our findings suggest that these spatial habitat diversity features are associated with these clinical characteristics and could be a useful prognostic tool for magnetic resonance imaging studies of patients with GBM.

PURPOSE: To characterize the biological and clinical correlates of radiologically-defined tumor composition. Further, to investigate if tumor volumetric and heterogeneity features can predict up- and down-regulation of biologically relevant gene-sets. METHODS: Genetic (mRNA expression) and radiographic scans (T1,T2,T1-post Gd and FLAIR) for 45 GBM cases was extracted from TCGA (Cancer Genome Atlas) and The Cancer Imaging Archive. Volumetric segmentation of the tumor region into 4 sub-compartments(necrosis, edema, enhancing and non-enhancing regions) was performed with an automated segmentation tool. Tumor-heterogeneity characteristics were measured from segmented post-contrast T1 and T2-FLAIR images. Information about biological pathways (including inflammation-associated gene signatures) were obtained from the Broad-website (MSigDb). Pathway up- or down-regulation scores were obtained using Gene Set Enrichment Analysis (GSEA). Multiple-response regression was used to identify pathways associated with the volumetric composition (percentage amounts of necrosis, edema, enhancing and non-enhancing regions) of the tumor. Four-dimensional volumetric composition was used to predict survival as well as status of key inflammatory pathways via a classification approach. True-positive (TPR) and false-positive rates (FPR) were obtained using cross-validation. RESULTS: Percentage of edema, necrosis and enhancing regions in the tumor are significantly associated with mesenchymal-subtype (p-value < 0.05). Examination of the molecular correlates underlying the volumetric composition of the tumor reveals several key pathways such as nitric-oxide biosynthesis, cAMP metabolism, microenvironmental remodeling and cellular differentiation are associated with(p-value < 0.05) and likely drive tumor composition. Further, prediction of 12 month survival using volumetrics and texture features is feasible: TPR of ∼90% and FPR of ∼20%. This study also reveals that volumetric features can be used for prediction of gene-set status (immune/inflammatory gene signatures in this case); with TPR and FPR of 80% and 10% respectively. CONCLUSION: Here, we characterized the molecular correlates of the radiologically-derived volumetric composition of glioblastoma. We show that combining image-derived volumetric and tumor heterogeneity features can predict gene-set status (specifically inflammation status) and overall survival.

Tumor heterogeneity is a crucial area of cancer research wherein inter- and intra-tumor differences are investigated to assess and monitor disease development and progression, especially in cancer. The proliferation of imaging and linked genomic data has enabled us to evaluate tumor heterogeneity on multiple levels. In this work, we examine magnetic resonance imaging (MRI) in patients with brain cancer to assess image-based tumor heterogeneity. Standard approaches to this problem use scalar summary measures (e.g., intensity-based histogram statistics) that do not adequately capture the complete and finer scale information in the voxel-level data. In this paper, we introduce a novel technique, DEMARCATE (DEnsity-based MAgnetic Resonance image Clustering for Assessing Tumor hEterogeneity) to explore the entire tumor heterogeneity density profiles (THDPs) obtained from the full tumor voxel space. THDPs are smoothed representations of the probability density function of the tumor images. We develop tools for analyzing such objects under the Fisher–Rao Riemannian framework that allows us to construct metrics for THDP comparisons across patients, which can be used in conjunction with standard clustering approaches. Our analyses of The Cancer Genome Atlas (TCGA) based Glioblastoma dataset reveal two significant clusters of patients with marked differences in tumor morphology, genomic characteristics and prognostic clinical outcomes. In addition, we see enrichment of image-based clusters with known molecular subtypes of glioblastoma multiforme, which further validates our representation of tumor heterogeneity and subsequent clustering techniques., Highlights • Tumor heterogeneity in glioblastoma is represented by a probability density function. • A technique for clustering, based on these density functions is introduced. • Geometry of the space on which these density functions lie is utilized. • Clusters show difference in tumor characteristics and prognostic clinical outcomes. • Clusters also show enrichment with subtypes of tumor and genomic covariates.