Abstract 2979: A web portal of ‘next-generation’ clustered heat maps for user-friendly, interactive exploration of patterns in TCGA data

The Cancer Genome Atlas (TCGA) program is generating comprehensive molecular profiles of more than 30 clinical tumor types, the first 12 of which have been incorporated into a “Pan-Cancer12” project. One bioinformatic challenge is statistical analysis of the resulting profiles; a second is the visual detective work necessary to explore individual genes, pathways and patterns in the data. For that type of detective work, we introduced CHMs in the early 1990s for pharmacogenomic analysis (1) and later for integrated visualization of genomic, transcriptomic, proteomic, pharmacological, and functional data (2). As the ubiquitous first-order way of visualizing omic data, CHMs have appeared in many thousands of publications (3-9), including all of the major publications by the TCGA Research Network. However, a major limitation is that they have been static or only modestly interactive graphics. We have now developed “next-generation” clustered heat maps (NG-CHMs), which use a Google-maps-like tiling technology for extreme zooming and navigation without loss of resolution. NG-CHMs provide pathway and gene ontology information, re-coloring on the fly, tools for reproducibility, high-resolution graphics output, a statistical toolbox, and link-outs to public sources of information on genes, proteins, pathways and drugs. The result is a visually rich, dynamic environment for exploration of the masses of data produced by TCGA. The compendium of TCGA Pan-Cancer NG-CHMs currently includes 667 maps as an initial set, but the numbers will soon rise into the thousands as more data types, tumour types and algorithms are incorporated (at web portal http://bioinformatics.mdanderson.org/TCGA/NGCHMPortal/). As an illustrative example, NG-CHMs proved pivotal as a tool for discovering and analyzing molecular target themes common to multiple types of gynecological cancers and themes that distinguish them from each other.

1. Weinstein JN … Paull KD. Stem Cells 12; 13, 1994.

2. Weinstein JN … Paull KD. Science 275;343, 1997.

3. Myers TG … Weinstein JN. Electrophoresis 18; 467, 1997.

4. Eisen MB … Botstein D. Proc. Natl. Acad. Sci. U.S.A. 14863, 1998.

5. Golub TR … Lander ES. Science 286; 531, 1999.

6. Ross DT … Brown PA. Nature Genetics 24; 227, 2000

7. Scherf U … Weinstein JN. Nature Genetics 24; 236, 2000.

8. Zeeberg BR … Weinstein JN. BMC Bioinformatics 6; 168, 2005.

9. Weinstein JN. Science 319; 1772, 2008.

Supported in part by NCI Grant No. U24CA143883, by a gift from the Mary K. Chapman Foundation, and by a grant from the Michael and Susan Dell Foundation honoring Lorraine Dell.

Note: This abstract was not presented at the meeting.

Citation Format: John N. Weinstein, Rehan Akbani, David W. Kane, James M. Melott, Tod D. Casasent, Rong Yao, Paul L. Roebuck, Gordon B. Mills, Michael C. Ryan, Christopher Wakefield, Bradley M. Broom. A web portal of ‘next-generation’ clustered heat maps for user-friendly, interactive exploration of patterns in TCGA data. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2979. doi:10.1158/1538-7445.AM2015-2979

Abstract 2969: Progress in The Cancer Genome Atlas bladder cancer project

In 2014, as part of The Cancer Genome Atlas (TCGA) Project, we reported molecular profiling of 131 chemotherapy-naive, muscle-invasive urothelial bladder cancers for DNA copy number variants (CNVs), somatic mutations by whole exome sequencing (WES), DNA methylation, mRNA expression, microRNA expression, protein expression and phosphorylation, transcript splicing, gene fusion, viral integration, pathway perturbation, clinical correlates, and histopathology (TCGA Research Network, Nature 507:315, 2014). Since that time, the number of tumors (from 19 tissue source sites) profiled comprehensively has doubled, and the final analyzed cohort will total 412 by February 2015. Unsupervised consensus clustering for the data from miRNAseq on the first 310 tumor samples shows five robust clusters, with miR-99a, miR-100, and miR-200 family members differentially abundant across the clusters. mRNA clusters I-IV remained stable as well. For the 266 tumour samples for which we have both miRNA and mRNA data, potential miRNA-mRNA targeting relationships supported by functional validation publications include miR-100-5p targeting FGFR3. The data also highlight the miR-200 family, miR-29abc, miR-17-92 and 106b-25 complexes, as well as miR-34a, miR-155, and miR-21. At this time, mutation analysis of WES data has been completed for 238 samples (including the first 131). Seven additional significantly mutated genes (using MutSig) have been identified: ASXL2 (11%), HSP90AA1 (7%), PSIP1 (5%), ZFP36L2 (5%), ZNF513 (5%), PTEN (4%), CEBPB (2%) (mutant frequency in parentheses). A sample with a POLE exonuclease domain mutation (P286R) exhibited an ultra-mutant phenotype (mutation rate ∼100 per MB). Many of the 38 (total) SMGs have not previously been described in bladder cancer. Combining copy number variation and somatic mutation data, 69% of tumors harbor one or more potentially actionable targets. Three mutation clusters were identified and characterized as: 1) “Focally amplified” - enriched in focal copy number alterations (e.g., 3p loss/PPARG) and MLL2 mutations; 2) Enriched for TP53 and RB1 mutations, E2F3 amplifications; and 3) Papillary histology, FGFR3 mutant CDKN2A-deficient.

This work was supported by the following grants from the United States National Institutes of Health: U54HG003273, U54 HG003067, U54 HG003079, U24 CA143799, U24 CA143835, U24CA143840, U24 CA143843, U24 CA143845, U24 CA143848, U24 CA143858, U24CA143866, U24 CA143867, U24 CA143882, U24 CA143883, U24 CA144025 and P01 CA120964, as well as multiple other funding sources for the individual authors.

Citation Format: John N. Weinstein, Jaegil Kim, Chad J. Creighton, Rehan Akbani, Katherine A. Hoadley, William Y. Kim, Margaret B. Morgan, Toshinori Hinoue, Andrew Cherniack, Xiaoping Su, Andrew J. Mungall, Michael C. Ryan, Dean F. Bajorin, Jonathan E. Rosenberg, Bogdan Czerniak, Donna Hansel, Victor E. Reuter, Brian D. Robinson, Hikmat A. Al-Ahmadie, Jeffrey S. Damrauer, Wei Zhang, Yuexin Liu, Dmitry R. Gordenin, Joshua M. Stuart, Nikolaus Schultz, Gordon Robertson, Steven JM Jones, Raju R. Kucherlapati, David J. McConkey, Peter W. Laird, Gordon B. Mills, David J. Kwiatkowski, Seth P. Lerner, TCGA Bladder Cancer Working Group, TCGA Research Network. Progress in The Cancer Genome Atlas bladder cancer project. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2969. doi:10.1158/1538-7445.AM2015-2969

Abstract 968: Co-occurring genomic alterations define major subsets of KRAS-mutant lung adenocarcinoma (LUAC) with distinct biology and therapeutic vulnerabilities

Introduction

The development of more effective treatment strategies for LUAC bearing activating mutations in KRAS is hampered by the biological heterogeneity of KRAS-mutant tumors. The molecular underpinnings that drive this process are poorly characterized. Here, we implemented an integrated approach to the discovery of biologically distinct subsets of KRAS-mutant LUAC and explored their molecular vulnerabilities.

Methods

Our datasets consisted of 68 KRAS-mutant tumors from TCGA, 88 additional chemo-naive KRAS-mutant LUACs (PROSPECT and Chitale datasets) and 36 platinum-refractory LUACs from the BATTLE-2 clinical trial. Non-negative matrix factorization (NMF) consensus clustering was applied to RNASeq data as previously described. Signature enrichment was assessed using Gene Set Enrichment Analysis (GSEA).

Results

NMF consensus clustering identified three robust subsets of KRAS-mutant LUAC that were reproducible across diverse clinical datasets of early-stage, chemotherapy-naive and metastatic, chemo-refractory tumors. Distinct KRAS-mutant alleles were not differentially represented in the three subgroups (P = 0.3). In contrast, the subgroups were dominated, respectively, by co-occurring genetic events in STK11/LKB1 (termed the KL subgroup) (P = 1.03e−05), TP53 (KP) (P = 3.8e-06) and low expression of TTF1 coupled with frequent CDKN2A/B inactivation (KC) (P = 0.004 and P = 0.002).

Distinct patterns of intracellular signaling were detected in the three subsets. KL tumors showed evidence of LKB1-AMPK pathway inactivation and adaptation to energetic, proteotoxic and oxidative stress, the latter exemplified by near ubiquitous inactivation of KEAP1 and up-regulation of a NRF2-driven antioxidant signature. KP tumors carried a higher somatic mutation load and were characterized by prominent inflammation and up-regulation of several immune checkpoint effector molecules, including PD-L1. KC tumors frequently displayed a GI-like differentiation program, suppression of MTORC1 signaling and elevated wild-type p53 transcriptional output.

Using a large panel of KRAS-mutant NSCLC cell lines we detected co-mutation-dependent patterns of drug sensitivity. Specifically, KL cell lines showed enhanced sensitivity to several structurally distinct HSP90 inhibitors. These results were confirmed in panels of isogenic cell lines. Mechanistically, treatment with ganetespib resulted in concurrent degradation of several molecules with established role in supporting the fitness of LKB1-deficient cells.

Conclusions

Our work identifies three major subsets of KRAS-mutant LUAC - dominated by co-occurring genetic events - with distinct biology and therapeutic vulnerabilities.

Citation Format: Ferdinandos Skoulidis, Lauren Byers, Lixia Diao, Vassiliki Papadimitrakopoulou, Pan Tong, Julie Izzo, Carmen Behrens, Humam Kadara, Edwin R. Parra, Jaime Rodriguez-Canales, Jianjun Zhang, Uma Giri, Jayanthi Gudikote, Maria Angelica Cortez, Chao Yang, You Hong Fan, Michael Peyton, Luc Girard, Kevin R. Coombes, Carlo Toniatti, Timothy P. Heffernan, Murim Choi, Garrett M. Frampton, Vincent Miller, John N. Weinstein, Roy S. Herbst, Kwok-Kin Wong, Jianhua Zhang, Padmanee Sharma, Gordon M. Mills, Waun Ki Hong, John D. Minna, James P. Allison, Andrew Futreal, Jing Wang, Ignacio Wistuba, John V. Heymach. Co-occurring genomic alterations define major subsets of KRAS-mutant lung adenocarcinoma (LUAC) with distinct biology and therapeutic vulnerabilities. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 968. doi:10.1158/1538-7445.AM2015-968

Abstract 161: Long noncoding RNA subtype classification of human prostate adenocarcinomas

Background: Long non-coding RNAs (lncRNAs) play a key role in regulating normal cell physiology as well as cancer progression. Although the role of several lncRNAs has been established in prostate cancer progression, the subtype classification of lncRNAs in human prostate cancer adenocarcinomas remains fully uncovered.

Material and Methods: We performed a genomic analysis of GENCODE lncRNAs in prostate adenocarcinoma (PRAD), using The Cancer Genome Atlas (TCGA) RNA-Seq profiles of 297 primary tumors. Furthermore, we described global correlations between lncRNAs and expression of cis-acting genes, and established lncRNAs expression-based subtypes with distinct signatures. In addition, lncRNAs expression-based subtypes were correlated with copy-number alterations and somatic mutations.

Results: Using stringent criteria, we identified 1,596 lncRNAs and predicted those that are potential drivers for cancer progression through integrative analysis. The expression of 885 (55.4%) lncRNAs showed a significantly positive correlation with the mRNAs expression of their neighboring genes, while only 29 (1.8%) lncRNAs showed a significantly negative correlation. 511 lncRNAs were significantly differentially expressed between PRAD and normal prostate; GREAT analysis showed that those lncRNAs are cis-acting on genes involved in the establishment of apical/basal cell polarity (p = 3.5×10−6). Unsupervised clustering of cancer differentially expressed lncRNAs revealed three robust subtypes, which were highly related with ERG gene fusions status (p = 2.2×10-16). While C2 cluster (n = 125) was composed of 76.3% of PRAD with ERG gene fusions cluster (n = 127), C1 cluster was composed in majority of PRAD without ERG gene fusions (83.5%). Cluster C3 (n = 45) stood out as a unique cluster with only 35% of PRAD samples bearing ERG fusions. Of note, those 3 subtypes were not different according to TNM stage, Gleason grade and age. Intriguingly, only SPOP somatic mutation was found to be enriched in C1 cluster (p = 7.9×10-6). Using copy number changes, we identified 13 regions which were differentially deleted between the three subgroups, including chr21q22.3 and chr8p21.3. As expected, Gene set enrichment analysis (GSEA) revealed enrichment of C2 cluster with prostate cancer TMPRSS2_ERG fusion signature, while C1 cluster was enriched for PPAR (p<10−12) and breast cancer ESR1 signatures (p<10−12). Importantly, Cluster 3 was enriched for EZH2 targets (p = 0.004) suggesting efficacy of EZH2 inhibitors in this subgroup. Finally, C3 cluster showed activation of UXT (Androgen receptor Trapped clone 27 protein) pathway (p = 2.73×10−3).

Conclusion: This study characterizes the spectrum of lncRNAs, which may be important in normal prostate as well as in prostate adenocarcinomas. Furthermore, we establish the foundation of lncRNAs expression-based subtype classification in human prostate adenocarcinomas.

Citation Format: Gabriel G. Malouf, Jianping Zhang, Jean-Philippe Spano, Eva Compérat, Nizar M. Tannir, Erika K. Thompson, John N. Weinstein, Debasish Tripathy, David Khayat, Xiaoping Su. Long noncoding RNA subtype classification of human prostate adenocarcinomas. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 161. doi:10.1158/1538-7445.AM2015-161

TANRIC: An Interactive Open Platform to Explore the Function of lncRNAs in Cancer

Long noncoding RNAs (lncRNA) have emerged as essential players in cancer biology. Using recent large-scale RNA-seq datasets, especially those from The Cancer Genome Atlas (TCGA), we have developed "The Atlas of Noncoding RNAs in Cancer" (TANRIC; http://bioinformatics.mdanderson.org/main/TANRIC:Overview), a user-friendly, open-access web resource for interactive exploration of lncRNAs in cancer. It characterizes the expression profiles of lncRNAs in large patient cohorts of 20 cancer types, including TCGA and independent datasets (>8,000 samples overall). TANRIC enables researchers to rapidly and intuitively analyze lncRNAs of interest (annotated lncRNAs or any user-defined ones) in the context of clinical and other molecular data, both within and across tumor types. Using TANRIC, we have identified a large number of lncRNAs with potential biomedical significance, many of which show strong correlations with established therapeutic targets and biomarkers across tumor types or with drug sensitivity across cell lines. TANRIC represents a valuable tool for investigating the function and clinical relevance of lncRNAs in cancer, greatly facilitating lncRNA-related biologic discoveries and clinical applications.

Tissue-specific isoform switch and DNA hypomethylation of the pyruvate kinase PKM gene in human cancers

The M2 isoform of pyruvate kinase (PKM2) plays an important role in aerobic glycolysis and is a mediator of the Warburg effect in tumors. It was previously thought that tumor cells switch expression of PKM from normal tissue-expressed PKM1 to tumor-specific PKM2 via an alternative splicing mechanism. This view was challenged by a recent report demonstrating that PKM2 is already the major PKM isoform expressed in many differentiated normal tissues. Here, through analyses on sixteen tumor types using the cancer genome atlas RNA-Seq and exon array datasets, we confirmed that isoform switch from PKM1 to PKM2 occurred in glioblastomas but not in other tumor types examined. Despite lacking of isoform switches, PKM2 expression was found to be increased in all cancer types examined, and correlated strongly to poor prognosis in head and neck cancers. We further demonstrated that elevated PKM2 expression correlated well with the hypomethylation status of intron 1 of the PKM gene in multiple cancer types, suggesting epigenetic regulation by DNA methylation as a major mechanism in controlling PKM transcription in tumors. Our study suggests that isoform switch of PKM1 to PKM2 in cancers is tissue-specific and targeting PKM2 activity in tumors remains a promising approach for clinical intervention of multiple cancer types.

Comprehensive, Integrative Genomic Analysis of Diffuse Lower-Grade Gliomas

BACKGROUND

Diffuse low-grade and intermediate-grade gliomas (which together make up the lower-grade gliomas, World Health Organization grades II and III) have highly variable clinical behavior that is not adequately predicted on the basis of histologic class. Some are indolent; others quickly progress to glioblastoma. The uncertainty is compounded by interobserver variability in histologic diagnosis. Mutations in IDH, TP53, and ATRX and codeletion of chromosome arms 1p and 19q (1p/19q codeletion) have been implicated as clinically relevant markers of lower-grade gliomas.

METHODS

We performed genomewide analyses of 293 lower-grade gliomas from adults, incorporating exome sequence, DNA copy number, DNA methylation, messenger RNA expression, microRNA expression, and targeted protein expression. These data were integrated and tested for correlation with clinical outcomes.

RESULTS

Unsupervised clustering of mutations and data from RNA, DNA-copy-number, and DNA-methylation platforms uncovered concordant classification of three robust, nonoverlapping, prognostically significant subtypes of lower-grade glioma that were captured more accurately by IDH, 1p/19q, and TP53 status than by histologic class. Patients who had lower-grade gliomas with an IDH mutation and 1p/19q codeletion had the most favorable clinical outcomes. Their gliomas harbored mutations in CIC, FUBP1, NOTCH1, and the TERT promoter. Nearly all lower-grade gliomas with IDH mutations and no 1p/19q codeletion had mutations in TP53 (94%) and ATRX inactivation (86%). The large majority of lower-grade gliomas without an IDH mutation had genomic aberrations and clinical behavior strikingly similar to those found in primary glioblastoma.

CONCLUSIONS

The integration of genomewide data from multiple platforms delineated three molecular classes of lower-grade gliomas that were more concordant with IDH, 1p/19q, and TP53 status than with histologic class. Lower-grade gliomas with an IDH mutation either had 1p/19q codeletion or carried a TP53 mutation. Most lower-grade gliomas without an IDH mutation were molecularly and clinically similar to glioblastoma. (Funded by the National Institutes of Health.).

Whole-genome and multisector exome sequencing of primary and post-treatment glioblastoma reveals patterns of tumor evolution

Glioblastoma (GBM) is a prototypical heterogeneous brain tumor refractory to conventional therapy. A small residual population of cells escapes surgery and chemoradiation, resulting in a typically fatal tumor recurrence ∼7 mo after diagnosis. Understanding the molecular architecture of this residual population is critical for the development of successful therapies. We used whole-genome sequencing and whole-exome sequencing of multiple sectors from primary and paired recurrent GBM tumors to reconstruct the genomic profile of residual, therapy resistant tumor initiating cells. We found that genetic alteration of the p53 pathway is a primary molecular event predictive of a high number of subclonal mutations in glioblastoma. The genomic road leading to recurrence is highly idiosyncratic but can be broadly classified into linear recurrences that share extensive genetic similarity with the primary tumor and can be directly traced to one of its specific sectors, and divergent recurrences that share few genetic alterations with the primary tumor and originate from cells that branched off early during tumorigenesis. Our study provides mechanistic insights into how genetic alterations in primary tumors impact the ensuing evolution of tumor cells and the emergence of subclonal heterogeneity.

Genes suppressed by DNA methylation in non-small cell lung cancer reveal the epigenetics of epithelial-mesenchymal transition

Background

DNA methylation is associated with aberrant gene expression in cancer, and has been shown to correlate with therapeutic response and disease prognosis in some types of cancer. We sought to investigate the biological significance of DNA methylation in lung cancer.

Results

We integrated the gene expression profiles and data of gene promoter methylation for a large panel of non-small cell lung cancer cell lines, and identified 578 candidate genes with expression levels that were inversely correlated to the degree of DNA methylation. We found these candidate genes to be differentially methylated in normal lung tissue versus non-small cell lung cancer tumors, and segregated by histologic and tumor subtypes. We used gene set enrichment analysis of the genes ranked by the degree of correlation between gene expression and DNA methylation to identify gene sets involved in cellular migration and metastasis. Our unsupervised hierarchical clustering of the candidate genes segregated cell lines according to the epithelial-to-mesenchymal transition phenotype. Genes related to the epithelial-to-mesenchymal transition, such as AXL, ESRP1, HoxB4, and SPINT1/2, were among the nearly 20% of the candidate genes that were differentially methylated between epithelial and mesenchymal cells. Greater numbers of genes were methylated in the mesenchymal cells and their expressions were upregulated by 5-azacytidine treatment. Methylation of the candidate genes was associated with erlotinib resistance in wild-type EGFR cell lines. The expression profiles of the candidate genes were associated with 8-week disease control in patients with wild-type EGFR who had unresectable non-small cell lung cancer treated with erlotinib, but not in patients treated with sorafenib.

Conclusions

Our results demonstrate that the underlying biology of genes regulated by DNA methylation may have predictive value in lung cancer that can be exploited therapeutically.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-1079) contains supplementary material, which is available to authorized users.

Development of a robust classifier for quality control of reverse-phase protein arrays

MOTIVATION

High-throughput reverse-phase protein array (RPPA) technology allows for the parallel measurement of protein expression levels in approximately 1000 samples. However, the many steps required in the complex protocol (sample lysate preparation, slide printing, hybridization, washing and amplified detection) may create substantial variability in data quality. We are not aware of any other quality control algorithm that is tuned to the special characteristics of RPPAs.

RESULTS

We have developed a novel classifier for quality control of RPPA experiments using a generalized linear model and logistic function. The outcome of the classifier, ranging from 0 to 1, is defined as the probability that a slide is of good quality. After training, we tested the classifier using two independent validation datasets. We conclude that the classifier can distinguish RPPA slides of good quality from those of poor quality sufficiently well such that normalization schemes, protein expression patterns and advanced biological analyses will not be drastically impacted by erroneous measurements or systematic variations.

AVAILABILITY AND IMPLEMENTATION

The classifier, implemented in the "SuperCurve" R package, can be freely downloaded at http://bioinformatics.mdanderson.org/main/OOMPA:Overview or http://r-forge.r-project.org/projects/supercurve/. The data used to develop and validate the classifier are available at http://bioinformatics.mdanderson.org/MOAR.

Abstract 1541: Co-mutations define major subsets of KRAS-driven lung adenocarcinoma with implications for targeted cancer therapy

Introduction

The molecular underpinnings that beget the considerable biological and clinical heterogeneity of lung adenocarcinoma (AC) driven by oncogenic KRAS are poorly characterized. We hypothesized that significantly co-occurring mutations in components of key cellular tumor suppressor pathways are major determinants of signalling diversification downstream of mutant KRAS.

Methods

We applied un-supervised hierarchical analysis and classical multi-dimensional scaling to determine patterns of somatic (co)-mutations within a curated set of significantly mutated genes (MutSigCV p value <0.001) among 99 KRAS-mutant lung ACs from the TCGA dataset. Independently, we derived expression clusters by applying un-supervised hierarchical analysis to a subset of 68 KRAS-driven lung ACs with available RNASeq data. Non-negative matrix factorization (NMF) was used to determine cluster stability.

Results

Clustering of KRAS-driven lung ACs based on patterns of somatic mutations revealed six major mutation clades. Five clades could be defined by combinations of somatic mutations in only four genes : KRAS, TP53, STK11/LKB1 and KEAP1, whereas a sixth clade appeared more heterogenous and was associated with a higher overall rate of somatic mutation.

Co-mutations in STK11/LKB1 and TP53 accounted for higher branches of the cluster dendrogram and were largely non-overlapping in the context of early-stage disease.

Integration of co-mutation analysis with unsupervised hierarchical analysis of RNA Seq differential expression data identified 3 major expression clusters, that were significantly enriched for the KRAS;TP53 (KP), KRAS;LKB1 (KL) and KRAS only (K) somatic mutation clades, as well as sub-clusters characterized by the presence of mutations in KEAP1. On the contrary, the distribution of the main KRAS amino acid substitutions (G12C, G12V and G12D) did not differ significantly between the expression clusters, indicating that co-mutations rather than distinct KRAS codon 12 base substitutions are the major driver of biological heterogeneity within KRAS-mutant lung ACs.

Re-analysis of publicly available large scale drug sensitivity data based on the proposed sub-classification revealed novel genotype-specific therapeutic vulnerabilities for the KL and KP subgroups including selective sensitivity of KL AC cell lines to MK-2206 ( a pan-AKT inhibitor), BX-795 (a PDK1 and TBK1/IKKϵ inhibitor) and the type II topoisomerase inhibitor etoposide. Selected hits were subsequently validated using pairs of isogenic cell lines.

Conclusions

Co-mutations in a small number of key tumor suppressor genes define biologically distinct and therapeutically relevant subgroups of KRAS-driven lung AC. Efforts to target KRAS signalling in the context of specific co-mutations may provide a more rational and personalized approach to therapy for this challenging group of lung cancer patients.

Citation Format: Ferdinandos Skoulidis, Kevin R. Coombes, Lixia Diao, Pan Tong, Maria A. Cortez, Uma Giri, Chao Yang, You Hong Fan, John N. Weinstein, Vassiliki Papadimitrakopoulou, John D. Minna, Jing Wang, Lauren A. Byers, John V. Heymach. Co-mutations define major subsets of KRAS-driven lung adenocarcinoma with implications for targeted cancer therapy. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 1541. doi:10.1158/1538-7445.AM2014-1541

Abstract 984: Alteration of the p53 pathway is associated with subclonal tumor progression in glioblastoma

To evaluate evolutionary patterns in progression and therapy-resistance of GBM, we analyzed the genomic profiles of 252 GBM samples from The Cancer Genome Atlas (TCGA)1, including 48 multi-sector and recurrent tumor biopsies taken from 17 pairs of pre- and post-treatment GBMs, to understand 1) the intratumoral heterogeneity of GBM and 2) how GBM responds to therapeutic intervention.

We integrated variant allele fraction, DNA copy number and genotype information to determine clonality of all mutations and found that 69.5% of mutations (median across samples 70.1% ± 19.6%) were classified as clonal and 30.5% as subclonal. To verify our classification approach, we classified mutations detected in two non-overlapping biopsies from 11 tumors into clonal and subclonal categories. Of mutations detected in both tumor sectors 86.2% were classified as clonal and 45.2% of sample-private mutations were categorized as clonal, which was a strongly significant difference (P = 1.8×10-87).

Separating patients into discrete age groups by an interval of 10 years, we found a significant linear correlation between clonal mutations and age (P < 0.001). This observation supports the notion that clonal mutations predominantly accumulated over the life span of the cell population that gave rise to the cell of origin before neoplastic onset2. No correlation with age was found for subclonal mutations. In contrast, the frequency of subclonal mutations was strongly correlated with the presence of alterations related to the p53 pathway (P < 0.001), particularly TP53 point mutation or amplification of MDM2. This observation raises the possibility that p53 pathway alterations stimulate subclonal evolution, possibly by providing greater tolerance to DNA damage and/or suppression of apoptosis3.

To evaluate the effects of clonal heterogeneity on disease recurrence, we analyzed matched pairs of primary and recurrent GBM, including five pairs with mutated TP53 and nine pairs with wildtype TP53. Recurrent TP53 mutant GBM showed a further and significant increase in the subclonal mutation frequency. In contrast, TP53 wildtype tumors showed an increase in the frequency of clonal mutations compared to their matched primary tumor. These data suggest that TP53 mutant GBM became increasingly clonally complex at time of recurrence, whereas TP53 wildtype GBM showed a reduced level of intratumoral heterogeneity. We observed an ultramutator phenotype in three recurrent tumors, associated with temozolomide treatment.

Our results suggest that mutations in the p53 pathway affect the response to therapy by supporting greater intratumoral heterogeneity. Paradoxically we observe a trend towards improved event free survival in samples with high subclonal mutation frequency. Further research is needed to assess whether the level of intratumoral heterogeneity is a reflection of the molecular portrait of GBM progression.

Citation Format: Hoon Kim, Siyuan Zheng, Seyed S. Amini, Selene M. Virk, Tom Mikkelsen, Daniel J. Brat, Jonna Grimsby, Carrie Sougnez, Andrew E. Sloan, Mark L. Cohen, Erwin G. Van Meir, Lisa Scarpace, The Cancer Genome Research Network, Peter W. Laird, John N. Weinstein, Eric Lander, Stacey Gabriel, Gad Getz, Matthew Meyerson, Lynda Chin, Jill S. Barnholtz-Sloan, Roel G.w. Verhaak. Alteration of the p53 pathway is associated with subclonal tumor progression in glioblastoma. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 984. doi:10.1158/1538-7445.AM2014-984

Abstract 3788: Drug mechanisms of action: Triangulating with cultured cancer cells

The clinical trial is irreplaceable as a scientific test bed for cancer therapeutics, animal models are becoming ever more sophisticated for preclinical testing, and 3D cell cultures can mimic some aspects of tumor structure and microenvironment. So what are the proper roles of 2D cell cultures in research on mechanisms of drug action? A partial answer is historical: most of our information about mechanisms of drug sensitivity and resistance has come from such cultures. A second is that cells in culture can be engineered readily by selective pressure or genetic manipulation to express or silence any chosen drug target or modulator of drug activity. A third is provided by the emergence of cancer cell panels for use in large-scale drug discovery programs, at the same time generating massive amounts of data that can be used to support or refute hypotheses about molecular mechanism. Cell line assays are not in general predictive of clinical efficacy, but patterns of response among lines and under different conditions can certainly predict mechanism of action, often quite incisively (see, e.g., Weinstein, Science 258;447, 1992).

The first large-scale cell-line screen used to illuminate mechanisms of action was the NCI-60, a diverse set of 60 human lines used to test >100,000 compounds and natural products since 1988 (Shoemaker, Prog. Clin. Biol. Res. 2:361, 1988). Because we and others have profiled the same lines at the DNA, RNA, protein, chromosomal, and epigenomic levels (see, e.g., Weinstein, Science 258;447, 1997), it has been possible to correlate molecular aberrations in the cells with their sensitivity to the agents tested. More recently, the Cancer Cell Line Encyclopedia (CCLE) (Barretina, Nature 483;603, 2012) and Cancer Genome Project (CGP) (Garnett, Nature 483;570, 2012) were published, describing molecular profiles and drug response assays for 1,036 cell lines and 24 drugs, and 727 cell lines and 138 drugs, respectively. The three cell panel assays complement each other: the NCI-60 provides a relatively small number of cell types but many tested agents; the CCLE and CGP provide many more cell types but a much smaller collection of tested compounds. Because the assays and experimental conditions being used in the three systems differ (Weinstein and Lorenzi, Nature, doi: 10.1038/nature12839, 2013), we are able to use the similarities and differences in response to ‘triangulate’ for more incisive and extensive information on the robustness of molecular mechanisms suggested by each separately. Even if single pharmacological measurements for a given cell/drug pair show considerable variability across different assay types (Haibe-Kains et al. Nature doi: 10.1038/nature12831, 2013), we find that integrated analysis of the patterns of response across the 60 cell lines and the hundreds of cell lines are often pathognomonic for particular genomic drivers or particular functional pathways.

Citation Format: John N. Weinstein, Bo Peng, William C. Reinhold, Yves Pommier, Philip L. Lorenzi. Drug mechanisms of action: Triangulating with cultured cancer cells. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 3788. doi:10.1158/1538-7445.AM2014-3788

Abstract 987: Comprehensive characterization of urothelial bladder cancer: a TCGA Project update

Urothelial carcinoma (UC) is a major cause of morbidity and mortality for which there are no approved molecularly targeted agents and few good treatment options beyond cisplatin-based chemotherapy. As part of The Cancer Genome Atlas (TCGA) Project, we analyzed 131 chemotherapy-naive, muscle-invasive UC tumors for somatic mutations, DNA copy number variants (CNVs), mRNA and microRNA expression, protein expression and phosphorylation, DNA methylation, transcript splicing, gene fusion, viral integration, pathway perturbation, clinical correlates, and histopathology (TCGA Research Network, Nature, in press). Whole-exome sequencing showed 29 recurrently mutated genes. Potential therapeutic targets include altered PIK3CA, ERBB2, FGFR3, TSC1, and ERBB3, plus mutated chromatin-regulating genes MLL, MLL2, MLL3, CREBBP, CHD7, SRCAP, ARID1A, KDM6A (UTX), and EP300. There were 22 arm-level CNVs and 27 focally amplified or deleted regions. CDKN2A was deleted in 47%. Low-pass whole genome sequencing identified FGFR3-TACC3 fusions. Viral DNA was identified in 6% (CMV, HHV6B, HPV16, BK polyoma), and viral transcripts were identified in 4% (CMV, BK polyoma, HPV16).

. mRNA-seq identified 4 tumor clusters. Cluster I shows papillary morphology and FGFR3 dysregulation. Clusters I and II express high HER2 (ERBB2) and estrogen receptor beta signaling signature, sharing features with Luminal A breast cancer. Cluster III shows similarities to Basal-like breast and squamous cell head and neck carcinomas.

. Integrated analyses confirm alteration of multiple pathways, including cell cycle regulation (93%), kinase and PI3-K signaling (72%), and epigenetic regulation (histone-modifiers: 89%; SWI/SNF nucleosome remodeling complex: 64%). Recurrent alterations in the PI3-kinase/AKT/mTOR pathway (42%) and RTK/RAS pathway (44%) are potentially actionable.

. Overall, this study and others have identified multiple druggable targets in UC. FGFR3 is activated by mutation, gene fusion, and overexpression, suggesting clinical trials of FGFR3 inhibitors. PI3-kinase/mTOR/AKT/TSC1 pathway alterations are frequent, and mutation in TSC1 has been associated with response to mTOR inhibitors. ERBB2 amplifications and activating mutations may be targetable with agents such as trastuzumab, trastuzumab-DM1, lapatinib, and neratinib. The frequent alterations in epigenetic regulatory pathways suggest trials of agents such as the bromodomain inhibitors. The project is now being updated on the basis of data on 117 additional tumors.

Citation Format: John N. Weinstein, Jaegil Kim, Chad J. Creighton, Rehan Akbani, Katherine A. Hoadley, William Y. Kim, Margaret B. Morgan, Toshinori Hinoue, Andrew Cherniack, Xiaoping Su, Andrew J. Mungall, Michael C. Ryan, Jonathan E. Rosenberg, Dean F. Bajorin, Bogdan Czerniak, Donna Hansel, Victor E. Reuter, Brian D. Robinson, Hikmat A. Al-Ahmadie, Jeffrey S. Damrauer, Wei Zhang, Yuexin Liu, Dmitry Gordenin, Joshua M. Stuart, Nikolaus Schultz, Gordon Robertson, Raju Kucherlapati, Peter W. Laird, Gordon B. Mills, David J. Kwiatkowski, Seth P. Lerner, representing TCGA's Bladder Cancer Working Group. Comprehensive characterization of urothelial bladder cancer: a TCGA Project update. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 987. doi:10.1158/1538-7445.AM2014-987

Multiplatform analysis of 12 cancer types reveals molecular classification within and across tissues of origin

Recent genomic analyses of pathologically defined tumor types identify "within-a-tissue" disease subtypes. However, the extent to which genomic signatures are shared across tissues is still unclear. We performed an integrative analysis using five genome-wide platforms and one proteomic platform on 3,527 specimens from 12 cancer types, revealing a unified classification into 11 major subtypes. Five subtypes were nearly identical to their tissue-of-origin counterparts, but several distinct cancer types were found to converge into common subtypes. Lung squamous, head and neck, and a subset of bladder cancers coalesced into one subtype typified by TP53 alterations, TP63 amplifications, and high expression of immune and proliferation pathway genes. Of note, bladder cancers split into three pan-cancer subtypes. The multiplatform classification, while correlated with tissue-of-origin, provides independent information for predicting clinical outcomes. All data sets are available for data-mining from a unified resource to support further biological discoveries and insights into novel therapeutic strategies.

Squamous cell carcinoma of the oral tongue in young non-smokers is genomically similar to tumors in older smokers

PURPOSE

Epidemiologic studies have identified an increasing incidence of squamous cell carcinoma of the oral tongue (SCCOT) in younger patients.

EXPERIMENTAL DESIGN

DNA isolated from tongue tumors of young (<45 years, non-smokers) and old (>45 years) patients at was subjected to whole-exome sequencing and copy-number analysis. These data were compared with data from similar patients in the TCGA (The Cancer Genome Atlas) project.

RESULTS

In this study, we found that gene-specific mutation and copy-number alteration frequencies were similar between young and old patients with SCCOT in two independent cohorts. Likewise, the types of base changes observed in the young cohort were similar to those in the old cohort even though they differed in smoking history. TCGA data also demonstrate that the genomic effects of smoking are tumor site-specific, and we find that smoking has only a minor impact on the types of mutations observed in SCCOT.

CONCLUSIONS

Overall, tumors from young patients with SCCOT appear genomically similar to those of older patients with SCCOT, and the cause for the increasing incidence of young SCCOT remains unknown. These data indicate that the functional impact of smoking on carcinogenesis in SCCOT is still poorly understood.

Assessing the clinical utility of cancer genomic and proteomic data across tumor types

Molecular profiling of tumors promises to advance the clinical management of cancer, but the benefits of integrating molecular data with traditional clinical variables have not been systematically studied. Here we retrospectively predict patient survival using diverse molecular data (somatic copy-number alteration, DNA methylation and mRNA, microRNA and protein expression) from 953 samples of four cancer types from The Cancer Genome Atlas project. We find that incorporating molecular data with clinical variables yields statistically significantly improved predictions (FDR < 0.05) for three cancers but those quantitative gains were limited (2.2-23.9%). Additional analyses revealed little predictive power across tumor types except for one case. In clinically relevant genes, we identified 10,281 somatic alterations across 12 cancer types in 2,928 of 3,277 patients (89.4%), many of which would not be revealed in single-tumor analyses. Our study provides a starting point and resources, including an open-access model evaluation platform, for building reliable prognostic and therapeutic strategies that incorporate molecular data.

An artifact in LC-MS/MS measurement of glutamine and glutamic acid: in-source cyclization to pyroglutamic acid

Advances in metabolomics, particularly for research on cancer, have increased the demand for accurate, highly sensitive methods for measuring glutamine (Gln) and glutamic acid (Glu) in cell cultures and other biological samples. N-terminal Gln and Glu residues in proteins or peptides have been reported to cyclize to pyroglutamic acid (pGlu) during liquid chromatography (LC)-mass spectrometry (MS) analysis, but cyclization of free Gln and Glu to free pGlu during LC-MS analysis has not been well-characterized. Using an LC-MS/MS protocol that we developed to separate Gln, Glu, and pGlu, we found that free Gln and Glu cyclize to pGlu in the electrospray ionization source, revealing a previously uncharacterized artifact in metabolomic studies. Analysis of Gln standards over a concentration range from 0.39 to 200 μM indicated that a minimum of 33% and maximum of almost 100% of Gln was converted to pGlu in the ionization source, with the extent of conversion dependent on fragmentor voltage. We conclude that the sensitivity and accuracy of Gln, Glu, and pGlu quantitation by electrospray ionization-based mass spectrometry can be improved dramatically by using (i) chromatographic conditions that adequately separate the three metabolites, (ii) isotopic internal standards to correct for in-source pGlu formation, and (iii) user-optimized fragmentor voltage for acquisition of the MS spectra. These findings have immediate impact on metabolomics and metabolism research using LC-MS technologies.

A curated census of autophagy-modulating proteins and small molecules: candidate targets for cancer therapy

Autophagy, a programmed process in which cell contents are delivered to lysosomes for degradation, appears to have both tumor-suppressive and tumor-promoting functions; both stimulation and inhibition of autophagy have been reported to induce cancer cell death, and particular genes and proteins have been associated both positively and negatively with autophagy. To provide a basis for incisive analysis of those complexities and ambiguities and to guide development of new autophagy-targeted treatments for cancer, we have compiled a comprehensive, curated inventory of autophagy modulators by integrating information from published siRNA screens, multiple pathway analysis algorithms, and extensive, manually curated text-mining of the literature. The resulting inventory includes 739 proteins and 385 chemicals (including drugs, small molecules, and metabolites). Because autophagy is still at an early stage of investigation, we provide extensive analysis of our sources of information and their complex relationships with each other. We conclude with a discussion of novel strategies that could potentially be used to target autophagy for cancer therapy.

Tumor subtype-specific cancer-testis antigens as potential biomarkers and immunotherapeutic targets for cancers

Cancer-testis (CT) antigens are potential targets for cancer immunotherapy because of their restricted expression in immune-privileged germ cells and various malignancies. Current application of CT-based immunotherapy has been focused on CT expression-rich tumors such as melanoma and lung cancers. In this study, we surveyed CT expression using The Cancer Genome Atlas (TCGA) datasets for ten common cancer types. We show that CT expression is specific and enriched within certain cancer molecular subtypes. For example, HORMAD1, CXorf61, ACTL8, and PRAME are highly enriched in the basal subtype of breast cancer; MAGE and CSAG are most frequently activated in the magnoid subtype of lung adenocarcinoma; and PRAME is highly upregulated in the ccB subtype of clear cell renal cell carcinoma. Analysis of CT gene expression and DNA methylation indicates that some CTs are regulated epigenetically, whereas others are controlled primarily by tissue- and subtype-specific transcription factors. Our results suggest that although for some CT expression is associated with patient outcome, not many are independent prognostic markers. Thus, CTs with shared expression pattern are heterogeneous molecules with distinct activation modes and functional properties in different cancers and cancer subtypes. These data suggest a cancer subtype-orientated application of CT antigen as biomarkers and immunotherapeutic targets.

PRADA: pipeline for RNA sequencing data analysis

SUMMARY

Technological advances in high-throughput sequencing necessitate improved computational tools for processing and analyzing large-scale datasets in a systematic automated manner. For that purpose, we have developed PRADA (Pipeline for RNA-Sequencing Data Analysis), a flexible, modular and highly scalable software platform that provides many different types of information available by multifaceted analysis starting from raw paired-end RNA-seq data: gene expression levels, quality metrics, detection of unsupervised and supervised fusion transcripts, detection of intragenic fusion variants, homology scores and fusion frame classification. PRADA uses a dual-mapping strategy that increases sensitivity and refines the analytical endpoints. PRADA has been used extensively and successfully in the glioblastoma and renal clear cell projects of The Cancer Genome Atlas program.

AVAILABILITY AND IMPLEMENTATION

 http://sourceforge.net/projects/prada/

CONTACT

 gadgetz@broadinstitute.org or rverhaak@mdanderson.org

SUPPLEMENTARY INFORMATION

 Supplementary data are available at Bioinformatics online.