Quantification of the Heterogeneity of Prognostic Cellular Biomarkers in Ewing Sarcoma Using Automated Image and Random Survival Forest Analysis

Driven by genomic somatic variation, tumour tissues are typically heterogeneous, yet unbiased quantitative methods are rarely used to analyse heterogeneity at the protein level. Motivated by this problem, we developed automated image segmentation of images of multiple biomarkers in Ewing sarcoma to generate distributions of biomarkers between and within tumour cells. We further integrate high dimensional data with patient clinical outcomes utilising random survival forest (RSF) machine learning. Using material from cohorts of genetically diagnosed Ewing sarcoma with EWSR1 chromosomal translocations, confocal images of tissue microarrays were segmented with level sets and watershed algorithms. Each cell nucleus and cytoplasm were identified in relation to DAPI and CD99, respectively, and protein biomarkers (e.g. Ki67, pS6, Foxo3a, EGR1, MAPK) localised relative to nuclear and cytoplasmic regions of each cell in order to generate image feature distributions. The image distribution features were analysed with RSF in relation to known overall patient survival from three separate cohorts (185 informative cases). Variation in pre-analytical processing resulted in elimination of a high number of non-informative images that had poor DAPI localisation or biomarker preservation (67 cases, 36%). The distribution of image features for biomarkers in the remaining high quality material (118 cases, 104 features per case) were analysed by RSF with feature selection, and performance assessed using internal cross-validation, rather than a separate validation cohort. A prognostic classifier for Ewing sarcoma with low cross-validation error rates (0.36) was comprised of multiple features, including the Ki67 proliferative marker and a sub-population of cells with low cytoplasmic/nuclear ratio of CD99. Through elimination of bias, the evaluation of high-dimensionality biomarker distribution within cell populations of a tumour using random forest analysis in quality controlled tumour material could be achieved. Such an automated and integrated methodology has potential application in the identification of prognostic classifiers based on tumour cell heterogeneity.

[1]  D. Reinke,et al.  R1507, a monoclonal antibody to the insulin-like growth factor 1 receptor, in patients with recurrent or refractory Ewing sarcoma family of tumors: results of a phase II Sarcoma Alliance for Research through Collaboration study. , 2011, Journal of clinical oncology : official journal of the American Society of Clinical Oncology.

[2]  Udaya B. Kogalur,et al.  Consistency of Random Survival Forests. , 2008, Statistics & probability letters.

[3]  Carsten Denkert,et al.  Cutoff Finder: A Comprehensive and Straightforward Web Application Enabling Rapid Biomarker Cutoff Optimization , 2012, PloS one.

[4]  R. Casadonte,et al.  Proteomic analysis of formalin-fixed paraffin-embedded tissue by MALDI imaging mass spectrometry , 2011, Nature Protocols.

[5]  O. Delattre,et al.  Mesenchymal stem cell features of Ewing tumors. , 2007, Cancer cell.

[6]  P. A. Futreal,et al.  Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. , 2012, The New England journal of medicine.

[7]  David T. W. Jones,et al.  Signatures of mutational processes in human cancer , 2013, Nature.

[8]  S. Horvath,et al.  Global histone modification patterns predict risk of prostate cancer recurrence , 2005, Nature.

[9]  M. Ishii,et al.  Quantifying cellular interaction dynamics in 3D fluorescence microscopy data , 2009, Nature Protocols.

[10]  T. Holzer,et al.  Ischemic time impacts biological integrity of phospho-proteins in PI3K/Akt, Erk/MAPK, and p38 MAPK signaling networks. , 2011, Anticancer research.

[11]  D. Rimm,et al.  Automated analysis of tissue microarrays. , 2010, Methods in molecular biology.

[12]  G. Gatta,et al.  Rare cancers are not so rare: the rare cancer burden in Europe. , 2011, European journal of cancer.

[13]  Steve Horvath,et al.  Tumor classification by tissue microarray profiling: random forest clustering applied to renal cell carcinoma , 2005, Modern Pathology.

[14]  J. Rittscher.  Characterization of biological processes through automated image analysis. , 2010, Annual review of biomedical engineering.

[15]  Huanming Yang,et al.  Single-Cell Exome Sequencing Reveals Single-Nucleotide Mutation Characteristics of a Kidney Tumor , 2012, Cell.

[16]  Yu Guo,et al.  Sample size and statistical power considerations in high-dimensionality data settings: a comparative study of classification algorithms , 2010, BMC Bioinformatics.

[17]  Piero Carninci,et al.  A comprehensive promoter landscape identifies a novel promoter for CD133 in restricted tissues, cancers, and stem cells , 2013, Front. Genet..

[18]  T. Triche,et al.  CD133 expression in chemo-resistant Ewing sarcoma cells , 2010, BMC Cancer.

[19]  U. Dirksen,et al.  Cyclophosphamide compared with ifosfamide in consolidation treatment of standard-risk Ewing sarcoma: results of the randomized noninferiority Euro-EWING99-R1 trial. , 2014, Journal of clinical oncology : official journal of the American Society of Clinical Oncology.

[20]  D. Rimm,et al.  Automated subcellular localization and quantification of protein expression in tissue microarrays , 2002, Nature Medicine.

[21]  Yuval Kluger,et al.  Quantitative assessment of effect of preanalytic cold ischemic time on protein expression in breast cancer tissues. , 2012, Journal of the National Cancer Institute.

[22]  Adel Hafiane,et al.  Histopathology tissue segmentation by combining fuzzy clustering with multiphase vector level sets. , 2011, Advances in experimental medicine and biology.

[23]  Michele De Palma,et al.  The biology of personalized cancer medicine: Facing individual complexities underlying hallmark capabilities , 2012, Molecular oncology.

[24]  R. Caprioli,et al.  Lung cancer diagnosis from proteomic analysis of preinvasive lesions. , 2011, Cancer research.

[25]  Hemant Ishwaran,et al.  Random Survival Forests , 2008, Wiley StatsRef: Statistics Reference Online.

[26]  Liam A. McDonnell,et al.  Multiple Statistical Analysis Techniques Corroborate Intratumor Heterogeneity in Imaging Mass Spectrometry Datasets of Myxofibrosarcoma , 2011, PloS one.

[27]  D. Rimm,et al.  Long-term preservation of antigenicity on tissue microarrays , 2004, Laboratory Investigation.

[28]  H. Kovar,et al.  Ewing's sarcoma family of tumors: current management. , 2006, The oncologist.

[29]  E. Álava,et al.  1q gain and CDT2 overexpression underlie an aggressive and highly proliferative form of Ewing sarcoma , 2012, Oncogene.

[30]  Fiona Ginty,et al.  The Relative Distribution of Membranous and Cytoplasmic Met Is a Prognostic Indicator in Stage I and II Colon Cancer , 2008, Clinical Cancer Research.

[31]  Nicolò Riggi,et al.  A TARBP2-dependent miRNA expression profile underlies cancer stem cell properties and provides candidate therapeutic reagents in Ewing sarcoma. , 2012, Cancer cell.

[32]  Klaus Kayser,et al.  Introduction of virtual microscopy in routine surgical pathology - a hypothesis and personal view from Europe , 2012, Diagnostic Pathology.

[33]  F. Klauschen,et al.  Computational reconstruction of cell and tissue surfaces for modeling and data analysis , 2009, Nature Protocols.

[34]  David L Rimm,et al.  Delay to formalin fixation ‘cold ischemia time’: effect on ERBB2 detection by in-situ hybridization and immunohistochemistry , 2013, Modern Pathology.

[35]  Andrew H. Beck,et al.  Systematic Analysis of Breast Cancer Morphology Uncovers Stromal Features Associated with Survival , 2011, Science Translational Medicine.

[36]  A. B. Hassan,et al.  The clinical use of biomarkers as prognostic factors in Ewing sarcoma , 2012, Clinical Sarcoma Research.

[37]  M. Nowak,et al.  Distant Metastasis Occurs Late during the Genetic Evolution of Pancreatic Cancer , 2010, Nature.

[38]  O. Myklebost,et al.  Molecular characterization of commonly used cell lines for bone tumor research: A trans‐European EuroBoNet effort , 2010, Genes, chromosomes & cancer.

[39]  N. Rosenfeld,et al.  Non-invasive analysis of acquired resistance to cancer therapy by sequencing of plasma DNA , 2013, Nature.

[40]  Charles Swanton,et al.  Intratumor Heterogeneity: Seeing the Wood for the Trees , 2012, Science Translational Medicine.

[41]  Anne E Carpenter,et al.  Workflow and Metrics for Image Quality Control in Large-Scale High-Content Screens , 2012, Journal of biomolecular screening.

[42]  Carolyn Tregidgo,et al.  Genome Sequencing and Analysis of the Tasmanian Devil and Its Transmissible Cancer , 2012, Cell.

[43]  Allen R. Tannenbaum,et al.  Localizing Region-Based Active Contours , 2008, IEEE Transactions on Image Processing.

[44]  Joel H. Saltz,et al.  Machine-Based Morphologic Analysis of Glioblastoma Using Whole-Slide Pathology Images Uncovers Clinically Relevant Molecular Correlates , 2013, PloS one.

[45]  M.,et al.  Statistical and Structural Approaches to Texture , 2022 .

[46]  R. Halaban,et al.  Automated Quantitative Analysis of HDM2 Expression in Malignant Melanoma Shows Association with Early-Stage Disease and Improved Outcome , 2004, Cancer Research.

[47]  Polina Golland,et al.  Voronoi-Based Segmentation of Cells on Image Manifolds , 2005, CVBIA.

[48]  D. Rimm,et al.  Quantitative Assessment Shows Loss of Antigenic Epitopes as a Function of Pre-analytic Variables , 2011, Laboratory Investigation.

[49]  Nicolò Riggi,et al.  Identification of cancer stem cells in Ewing's sarcoma. , 2009, Cancer research.

[50]  W. Schubert,et al.  Analyzing proteome topology and function by automated multidimensional fluorescence microscopy , 2006, Nature Biotechnology.

[51]  J. Troge,et al.  Tumour evolution inferred by single-cell sequencing , 2011, Nature.

[52]  D. Hanahan,et al.  Hallmarks of Cancer: The Next Generation , 2011, Cell.

[53]  F. Harrell,et al.  Evaluating the yield of medical tests. , 1982, JAMA.

[54]  Huanming Yang,et al.  Single-Cell Exome Sequencing and Monoclonal Evolution of a JAK2-Negative Myeloproliferative Neoplasm , 2012, Cell.

[55]  Polina Golland,et al.  Scoring diverse cellular morphologies in image-based screens with iterative feedback and machine learning , 2009, Proceedings of the National Academy of Sciences.

[56]  Richard M Caprioli,et al.  MALDI imaging mass spectrometry of human tissue: method challenges and clinical perspectives. , 2011, Trends in biotechnology.

[57]  Nicolò Riggi,et al.  EWS-FLI-1 modulates miRNA145 and SOX2 expression to initiate mesenchymal stem cell reprogramming toward Ewing sarcoma cancer stem cells. , 2010, Genes & development.

[58]  Xi Chen,et al.  An Integrative Pathway-based Clinical-genomic Model for Cancer Survival Prediction. , 2010, Statistics & probability letters.

[59]  A. Llombart‐Bosch,et al.  Clinicopathological significance of cell cycle regulation markers in a large series of genetically confirmed Ewing's Sarcoma Family of Tumors , 2011, International journal of cancer.

[60]  Walter Schubert,et al.  Fluorescence detection of protein clusters in individual cells and tissue sections by using toponome imaging system: sample preparation and measuring procedures , 2007, Nature Protocols.

[61]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[62]  M. Ladanyi,et al.  Prognostic impact of P53 status in Ewing sarcoma , 2000, Cancer.

[63]  Corbin E. Meacham,et al.  Tumour heterogeneity and cancer cell plasticity , 2013, Nature.

[64]  U. Dirksen,et al.  Preliminary efficacy of the anti-insulin-like growth factor type 1 receptor antibody figitumumab in patients with refractory Ewing sarcoma. , 2011, Journal of clinical oncology : official journal of the American Society of Clinical Oncology.

[65]  Olaf Hellwich,et al.  Determining similarity in histological images using graph-theoretic description and matching methods for content-based image retrieval in medical diagnostics , 2012, Diagnostic Pathology.

[66]  M. Tyers,et al.  From genomics to proteomics , 2003, Nature.

[67]  Friedrich Feuerhake,et al.  Effects of cold ischemia and inflammatory tumor microenvironment on detection of PI3K/AKT and MAPK pathway activation patterns in clinical cancer samples , 2012, International journal of cancer.

[68]  F. Markowetz,et al.  Quantitative Image Analysis of Cellular Heterogeneity in Breast Tumors Complements Genomic Profiling , 2012, Science Translational Medicine.

[69]  N. Otsu.  A threshold selection method from gray level histograms , 1979 .