Isoform-level gene signature improves prognostic stratification and accurately classifies glioblastoma subtypes

Molecular stratification of tumors is essential for developing personalized therapies. Although patient stratification strategies have been successful; computational methods to accurately translate the gene-signature from high-throughput platform to a clinically adaptable low-dimensional platform are currently lacking. Here, we describe PIGExClass (platform-independent isoform-level gene-expression based classification-system), a novel computational approach to derive and then transfer gene-signatures from one analytical platform to another. We applied PIGExClass to design a reverse transcriptase-quantitative polymerase chain reaction (RT-qPCR) based molecular-subtyping assay for glioblastoma multiforme (GBM), the most aggressive primary brain tumors. Unsupervised clustering of TCGA (the Cancer Genome Altas Consortium) GBM samples, based on isoform-level gene-expression profiles, recaptured the four known molecular subgroups but switched the subtype for 19% of the samples, resulting in significant (P = 0.0103) survival differences among the refined subgroups. PIGExClass derived four-class classifier, which requires only 121 transcript-variants, assigns GBM patients’ molecular subtype with 92% accuracy. This classifier was translated to an RT-qPCR assay and validated in an independent cohort of 206 GBM samples. Our results demonstrate the efficacy of PIGExClass in the design of clinically adaptable molecular subtyping assay and have implications for developing robust diagnostic assays for cancer patient stratification.

[1]  Gordon K Smyth,et al.  Statistical Applications in Genetics and Molecular Biology Linear Models and Empirical Bayes Methods for Assessing Differential Expression in Microarray Experiments , 2011 .

[2]  David R. Kelley,et al.  Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks , 2012, Nature Protocols.

[3]  C. Sander,et al.  Integrative Subtype Discovery in Glioblastoma Using iCluster , 2012, PloS one.

[4]  Pablo Tamayo,et al.  Metagenes and molecular pattern discovery using matrix factorization , 2004, Proceedings of the National Academy of Sciences of the United States of America.

[5]  Lior Pachter,et al.  Exon-Level Microarray Analyses Identify Alternative Splicing Programs in Breast Cancer , 2010, Molecular Cancer Research.

[6]  Somnath Datta,et al.  Classification of Breast Cancer versus Normal Samples from Mass Spectrometry Profiles Using Linear Discriminant Analysis of Important Features Selected by Random Forest , 2008, Statistical applications in genetics and molecular biology.

[7]  P. Grabowski.  Alternative splicing takes shape during neuronal development. , 2011, Current opinion in genetics & development.

[8]  William Pao,et al.  Translating genomic information into clinical medicine: Lung cancer as a paradigm , 2012, Genome research.

[9]  S. Gabriel,et al.  Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. , 2010, Cancer cell.

[10]  L. Deangelis,et al.  Glioblastoma: molecular analysis and clinical implications. , 2013, Annual review of medicine.

[11]  Ramón Díaz-Uriarte,et al.  Gene selection and classification of microarray data using random forest , 2006, BMC Bioinformatics.

[12]  D. Botstein,et al.  Gene expression profiling reveals molecularly and clinically distinct subtypes of glioblastoma multiforme. , 2005, Proceedings of the National Academy of Sciences of the United States of America.

[13]  D. Cheresh,et al.  VEGF inhibits tumor cell invasion and mesenchymal transition through a MET/VEGFR2 complex. , 2012, Cancer cell.

[14]  C. Brennan,et al.  Glioblastoma Subclasses Can Be Defined by Activity among Signal Transduction Pathways and Associated Genomic Alterations , 2009, PloS one.

[15]  Howard A. Fine,et al.  Integration and analysis of genome-scale data from gliomas , 2011, Nature Reviews Neurology.

[16]  Z. Meral Özsoyoglu,et al.  Classification of Large microarray Datasets Using Fast Random Forest Construction , 2011, J. Bioinform. Comput. Biol..

[17]  S. Horvath,et al.  Gene expression analysis of glioblastomas identifies the major molecular basis for the prognostic benefit of younger age , 2008, BMC Medical Genomics.

[18]  Yingtao Bi,et al.  Isoform level expression profiles provide better cancer signatures than gene level expression profiles , 2013, Genome Medicine.

[19]  B. Ebert,et al.  Mutations in RNA splicing machinery in human cancers. , 2011, New England Journal of Medicine.

[20]  Jun Z. Li,et al.  Genomic Estimates of Aneuploid Content in Glioblastoma Multiforme and Improved Classification , 2012, Clinical Cancer Research.

[21]  P. Rousseeuw.  Silhouettes: a graphical aid to the interpretation and validation of cluster analysis , 1987 .

[22]  G. K. Gray,et al.  Activation of the NF-κB Pathway by the STAT3 Inhibitor JSI-124 in Human Glioblastoma Cells , 2013, Molecular Cancer Research.

[23]  Yuan Qi,et al.  Integrated Genomic Analysis Identifies Clinically Relevant Subtypes of Glioblastoma Characterized by Abnormalities in PDGFRA , IDH 1 , EGFR , and NF 1 Citation Verhaak , 2010 .

[24]  D. Louis,et al.  Diagnostic and therapeutic avenues for glioblastoma: no longer a dead end? , 2013, Nature Reviews Clinical Oncology.

[25]  R. Wilson,et al.  Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. , 2010, Cancer cell.

[26]  J. Uhm,et al.  The transcriptional network for mesenchymal transformation of brain tumours , 2010 .

[27]  T. Wakabayashi,et al.  Current Trends in Targeted Therapies for Glioblastoma Multiforme , 2012, Neurology research international.

[28]  Q. Gao,et al.  Therapeutic targeting of EGFR-activated metabolic pathways in glioblastoma , 2013, Expert opinion on investigational drugs.

[29]  Luis Souhami,et al.  Age as an independent prognostic factor in patients with glioblastoma: a radiation therapy oncology group and American College of Surgeons National Cancer Data Base comparison , 2011, Journal of Neuro-Oncology.

[30]  Huan Liu,et al.  Discretization: An Enabling Technique , 2002, Data Mining and Knowledge Discovery.

[31]  Lynda Chin,et al.  Emerging insights into the molecular and cellular basis of glioblastoma. , 2012, Genes & development.

[32]  R. Davuluri,et al.  Alternative transcription and alternative splicing in cancer. , 2012, Pharmacology & therapeutics.

[33]  David T. W. Jones,et al.  Hotspot mutations in H3F3A and IDH1 define distinct epigenetic and biological subgroups of glioblastoma. , 2012, Cancer cell.

[34]  B. Blencowe,et al.  Global Profiling and Molecular Characterization of Alternative Splicing Events Misregulated in Lung Cancer , 2010, Molecular and Cellular Biology.

[35]  A. Dreher.  Modeling Survival Data Extending The Cox Model , 2016 .

[36]  Hyunsoo Kim,et al.  Alternative transcription exceeds alternative splicing in generating the transcriptome diversity of cerebellar development. , 2011, Genome research.

[37]  S. Knudsen,et al.  A new non-linear normalization method for reducing variability in DNA microarray experiments , 2002, Genome Biology.

[38]  Yuri Kotliarov,et al.  Unsupervised analysis of transcriptomic profiles reveals six glioma subtypes. , 2009, Cancer research.

[39]  Tom Misteli,et al.  RAF inhibitor resistance is mediated by dimerization of aberrantly spliced BRAF(V600E) , 2011, Nature.

[40]  Michael R Stratton,et al.  Genomics and the continuum of cancer care. , 2011, The New England journal of medicine.

[41]  Thomas D. Wu,et al.  Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. , 2006, Cancer cell.

[42]  Alex Lewin,et al.  MMBGX: a method for estimating expression at the isoform level and detecting differential splicing using whole-transcript Affymetrix arrays , 2009, Nucleic acids research.