Robust Prognostic Signature Research Articles

Identification and Validation of a Diagnostic and Prognostic Multi-Gene Biomarker Panel for Pancreatic Ductal Adenocarcinoma.

Late diagnosis and systemic dissemination essentially contribute to the invariably poor prognosis of pancreatic ductal adenocarcinoma (PDAC). Therefore, the development of diagnostic biomarkers for PDAC are urgently needed to improve patient stratification and outcome in the clinic. By studying the transcriptomes of independent PDAC patient cohorts of tumor and non-tumor tissues, we identified 81 robustly regulated genes, through a novel, generally applicable meta-analysis. Using consensus clustering on co-expression values revealed four distinct clusters with genes originating from exocrine/endocrine pancreas, stromal and tumor cells. Three clusters were strongly associated with survival of PDAC patients based on TCGA database underlining the prognostic potential of the identified genes. With the added information of impact of survival and the robustness within the meta-analysis, we extracted a 17-gene subset for further validation. We show that it did not only discriminate PDAC from non-tumor tissue and stroma in fresh-frozen as well as formalin-fixed paraffin embedded samples, but also detected pancreatic precursor lesions and singled out pancreatitis samples. Moreover, the classifier discriminated PDAC from other cancers in the TCGA database. In addition, we experimentally validated the classifier in PDAC patients on transcript level using qPCR and exemplify the usage on protein level for three proteins (AHNAK2, LAMC2, TFF1) using immunohistochemistry and for two secreted proteins (TFF1, SERPINB5) using ELISA-based protein detection in blood-plasma. In conclusion, we present a novel robust diagnostic and prognostic gene signature for PDAC with future potential applicability in the clinic.

Decentralized Learning Framework of Meta-Survival Analysis for Developing Robust Prognostic Signatures.

PurposeA significant hurdle in developing reliable gene expression–based prognostic models has been the limited sample size, which can cause overfitting and false discovery. Combining data from multiple studies can enhance statistical power and reduce spurious findings, but how to address the biologic heterogeneity across different datasets remains a major challenge. Better meta-survival analysis approaches are needed.Material and MethodsWe presented a decentralized learning framework for meta-survival analysis without the need for data aggregation. Our method consisted of a series of proposals that together alleviated the influence of data heterogeneity and improved the performance of survival prediction. First, we transformed the gene expression profile of every sample into normalized percentile ranks to obtain platform-agnostic features. Second, we used Stouffer’s meta-z approach in combination with Harrell’s concordance index to prioritize and select genes to be included in the model. Third, we used survival discordance as a scale-independent model loss function. Instead of generating a merged dataset and training the model therein, we avoided comparing patients across datasets and individually evaluated the loss function on each dataset. Finally, we optimized the model by minimizing the joint loss function.ResultsThrough comprehensive evaluation on 31 public microarray datasets containing 6,724 samples of several cancer types, we demonstrated that the proposed method has outperformed (1) single prognostic genes identified using conventional meta-analysis, (2) multigene signatures trained on single datasets, (3) multigene signatures trained on merged datasets as well as by other existing meta-analysis methods, and (4) clinically applicable, established multigene signatures.ConclusionThe decentralized learning approach can be used to effectively perform meta-analysis of gene expression data and to develop robust multigene prognostic signatures.

Comparative gene set enrichment analysis (GSEA) of the embryonic stem cell (ES) gene signatures in canine and human osteosarcoma

Osteosarcoma (OSA) is one of the most common malignancies in humans (especially children) and dogs. Canine OSA is considered an ideal model of human OSA because OSA in both species shares comparable oncogenic properties—including histopathological features, tumor biology, therapeutic approaches, and molecular mechanisms. In this study, we sought to compare the expression profiles of embryonic stem cell (ES) gene signatures in canine and human OSA tissues and cell lines. The expression patterns of three major ES gene signatures (modules or gene sets)—namely Myc, Core, and PRC modules—were primarily analyzed through the gene set enrichment analysis method in three gene expression datasets. For verification of the primary results, an additional 13 ES gene sets which were categorized into four groups—namely NOS targets, ES expressed, Myc targets, and Polycomb targets—were evaluated in expression datasets. Additionally, the prognostic efficacy of the gene sets with a similar enrichment pattern in humans and dogs was evaluated using the Cox proportional hazard model. Our results revealed that there were similar ES module expression patterns in the human OSA tissue and cell line, where the MYC modules, Myc targets, ES exp1, and NOS target modules were upregulated and the Polycomb target modules were downregulated in both entities. The canine OSA cell line showed ES enrichment patterns more similar to the patterns in its human counterpart, where we were able to detect the upregulation of the MYC modules, Myc targets, ES exp1, and NOS target modules and the downregulation of the Polycomb targets, including H3K27 bound and PRC2 targets. However, the canine OSA tissues presented a similar enrichment pattern at a lower degree than the canine OSA cell line. Furthermore, survival analysis identified the NOS targets as a more robust prognostic gene signature that efficiently predicted survival time in the human and canine OSA samples.

Differential expression analysis at the individual level reveals a lncRNA prognostic signature for lung adenocarcinoma

BackgroundDeregulations of long non-coding RNAs (lncRNAs) have been implicated in cancer initiation and progression. Current methods can only capture differential expression of lncRNAs at the population level and ignore the heterogeneous expression of lncRNAs in individual patients.MethodsWe propose a method (LncRIndiv) to identify differentially expressed (DE) lncRNAs in individual cancer patients by exploiting the disrupted ordering of expression levels of lncRNAs in each disease sample in comparison with stable normal ordering. LncRIndiv was applied to lncRNA expression profiles of lung adenocarcinoma (LUAD). Based on the expression profile of LUAD individual-level DE lncRNAs, we used a forward selection procedure to identify prognostic signature for stage I-II LUAD patients without adjuvant therapy.ResultsIn both simulated data and real pair-wise cancer and normal sample data, LncRIndiv method showed good performance. Based on the individual-level DE lncRNAs, we developed a robust prognostic signature consisting of two lncRNA (C1orf132 and TMPO-AS1) for stage I-II LUAD patients without adjuvant therapy (P = 3.06 × 10−6, log-rank test), which was confirmed in two independent datasets of GSE50081 (P = 1.82 × 10−2, log-rank test) and GSE31210 (P = 7.43 × 10−4, log-rank test) after adjusting other clinical factors such as smoking status and stages. Pathway analysis showed that TMPO-AS1 and C1orf132 could affect the prognosis of LUAD patients through regulating cell cycle and cell adhesion.ConclusionsLncRIndiv can successfully detect DE lncRNAs in individuals and be applied to identify prognostic signature for LUAD patients.

RNA sequencing-based cell proliferation analysis across 19 cancers identifies a subset of proliferation-informative cancers with a common survival signature.

Despite advances in cancer diagnosis and treatment strategies, robust prognostic signatures remain elusive in most cancers. Cell proliferation has long been recognized as a prognostic marker in cancer, but the generation of comprehensive, publicly available datasets allows examination of the links between cell proliferation and cancer characteristics such as mutation rate, stage, and patient outcomes. Here we explore the role of cell proliferation across 19 cancers (n = 6,581 patients) by using tissue-based RNA sequencing data from The Cancer Genome Atlas Project and calculating a ‘proliferative index’ derived from gene expression associated with Proliferating Cell Nuclear Antigen (PCNA) levels. This proliferative index is significantly associated with patient survival (Cox, p-value < 0.05) in 7 of 19 cancers, which we have defined as “proliferation-informative cancers” (PICs). In PICs, the proliferative index is strongly correlated with tumor stage and nodal invasion. PICs demonstrate reduced baseline expression of proliferation machinery relative to non-PICs. Additionally, we find the proliferative index is significantly associated with gross somatic mutation burden (Spearman, p = 1.76 × 10−23) as well as with mutations in individual driver genes. This analysis provides a comprehensive characterization of tumor proliferation indices and their association with disease progression and prognosis in multiple cancer types and highlights specific cancers that may be particularly susceptible to improved targeting of this classic cancer hallmark.

Stromal Gene Expression is Predictive for Metastatic Primary Prostate Cancer

BackgroundClinical grading systems using clinical features alongside nomograms lack precision in guiding treatment decisions in prostate cancer (PCa). There is a critical need for identification of biomarkers that can more accurately stratify patients with primary PCa. ObjectiveTo identify a robust prognostic signature to better distinguish indolent from aggressive prostate cancer (PCa). Design, setting, and participantsTo develop the signature, whole-genome and whole-transcriptome sequencing was conducted on five PCa patient-derived xenograft (PDX) models collected from independent foci of a single primary tumor and exhibiting variable metastatic phenotypes. Multiple independent clinical cohorts including an intermediate-risk cohort were used to validate the biomarkers. Outcome measurements and statistical analysisThe outcome measurement defining aggressive PCa was metastasis following radical prostatectomy. A generalized linear model with lasso regularization was used to build a 93-gene stroma-derived metastasis signature (SDMS). The SDMS association with metastasis was assessed using a Wilcoxon rank-sum test. Performance was evaluated using the area under the curve (AUC) for the receiver operating characteristic, and Kaplan-Meier curves. Univariable and multivariable regression models were used to compare the SDMS alongside clinicopathological variables and reported signatures. AUC was assessed to determine if SDMS is additive or synergistic to previously reported signatures. Results and limitationsA close association between stromal gene expression and metastatic phenotype was observed. Accordingly, the SDMS was modeled and validated in multiple independent clinical cohorts. Patients with higher SDMS scores were found to have worse prognosis. Furthermore, SDMS was an independent prognostic factor, can stratify risk in intermediate-risk PCa, and can improve the performance of other previously reported signatures. ConclusionsProfiling of stromal gene expression led to development of an SDMS that was validated as independently prognostic for the metastatic potential of prostate tumors. Patient summaryOur stroma-derived metastasis signature can predict the metastatic potential of early stage disease and will strengthen decisions regarding selection of active surveillance versus surgery and/or radiation therapy for prostate cancer patients. Furthermore, profiling of stroma cells should be more consistent than profiling of diverse cellular populations of heterogeneous tumors.

Transposon insertional mutagenesis in mice identifies human breast cancer susceptibility genes and signatures for stratification

Robust prognostic gene signatures and therapeutic targets are difficult to derive from expression profiling because of the significant heterogeneity within breast cancer (BC) subtypes. Here, we performed forward genetic screening in mice using Sleeping Beauty transposon mutagenesis to identify candidate BC driver genes in an unbiased manner, using a stabilized N-terminal truncated β-catenin gene as a sensitizer. We identified 134 mouse susceptibility genes from 129 common insertion sites within 34 mammary tumors. Of these, 126 genes were orthologous to protein-coding genes in the human genome (hereafter, human BC susceptibility genes, hBCSGs), 70% of which are previously reported cancer-associated genes, and ∼16% are known BC suppressor genes. Network analysis revealed a gene hub consisting of E1A binding protein P300 (EP300), CD44 molecule (CD44), neurofibromin (NF1) and phosphatase and tensin homolog (PTEN), which are linked to a significant number of mutated hBCSGs. From our survival prediction analysis of the expression of human BC genes in 2,333 BC cases, we isolated a six-gene-pair classifier that stratifies BC patients with high confidence into prognostically distinct low-, moderate-, and high-risk subgroups. Furthermore, we proposed prognostic classifiers identifying three basal and three claudin-low tumor subgroups. Intriguingly, our hBCSGs are mostly unrelated to cell cycle/mitosis genes and are distinct from the prognostic signatures currently used for stratifying BC patients. Our findings illustrate the strength and validity of integrating functional mutagenesis screens in mice with human cancer transcriptomic data to identify highly prognostic BC subtyping biomarkers.

Molecular substratification of bladder cancer: moving towards individualized patient management.

Despite advances in surgical techniques, perioperative therapies and postoperative management, outcomes for patients with bladder cancer have largely remained unchanged. Current management of bladder cancer still relies on pathologic staging that does not always reflect the risk for an individual patient. Studies assessing molecular alterations in individual tumors are offering insights into the myriad of cellular pathways that are deregulated in bladder tumorigenesis and progression. Alterations in pathways involved in cell-cycle regulation, apoptosis, cell signaling, angiogenesis and tumor-cell invasion have been shown to influence disease behavior. High-throughput assays are now allowing multiplexed assessment of biomarker alterations, thereby enabling characterization of novel molecular subtypes of bladder cancer. Such approaches have also been used for discovery and validation of robust prognostic molecular signatures. The future of bladder cancer management will rely on the use of validated multimarker panels for risk stratification, optimal surgical management, and theranostic strategies to identify and target specific alterations in individual tumors.

The modularity and dynamicity of miRNA–mRNA interactions in high-grade serous ovarian carcinomas and the prognostic implication

Ovarian carcinoma is the fifth-leading cause of cancer death among women in the United States. Major reasons for this persistent mortality include the poor understanding of the underlying biology and a lack of reliable biomarkers. Previous studies have shown that aberrantly expressed MicroRNAs (miRNAs) are involved in carcinogenesis and tumor progression by post-transcriptionally regulating gene expression. However, the interference of miRNAs in tumorigenesis is quite complicated and far from being fully understood. In this work, by an integrative analysis of mRNA expression, miRNA expression and clinical data published by The Cancer Genome Atlas (TCGA), we studied the modularity and dynamicity of miRNA–mRNA interactions and the prognostic implications in high-grade serous ovarian carcinomas. With the top transcriptional correlations (Bonferroni-adjusted p-value<0.01) as inputs, we identified five miRNA–mRNA module pairs (MPs), each of which included one positive-connection (correlation) module and one negative-connection (correlation) module. The number of miRNAs or mRNAs in each module varied from 3 to 7 or from 2 to 873. Among the four major negative-connection modules, three fit well with the widely accepted miRNA-mediated post-transcriptional regulation theory. These modules were enriched with the genes relevant to cell cycle and immune response. Moreover, we proposed two novel algorithms to reveal the group or sample specific dynamic regulations between these two RNA classes. The obtained miRNA–mRNA dynamic network contains 3350 interactions captured across different cancer progression stages or tumor grades. We found that those dynamic interactions tended to concentrate on a few miRNAs (e.g. miRNA-936), and were more likely present on the miRNA–mRNA pairs outside the discovered modules. In addition, we also pinpointed a robust prognostic signature consisting of 56 modular protein-coding genes, whose co-expression patterns were predictive for the survival time of ovarian cancer patients in multiple independent cohorts.

Intra-Gene DNA Methylation Variability Is a Clinically Independent Prognostic Marker in Women's Cancers.

We introduce a novel per-gene measure of intra-gene DNA methylation variability (IGV) based on the Illumina Infinium HumanMethylation450 platform, which is prognostic independently of well-known predictors of clinical outcome. Using IGV, we derive a robust gene-panel prognostic signature for ovarian cancer (OC, n = 221), which validates in two independent data sets from Mayo Clinic (n = 198) and TCGA (n = 358), with significance of p = 0.004 in both sets. The OC prognostic signature gene-panel is comprised of four gene groups, which represent distinct biological processes. We show the IGV measurements of these gene groups are most likely a reflection of a mixture of intra-tumour heterogeneity and transcription factor (TF) binding/activity. IGV can be used to predict clinical outcome in patients individually, providing a surrogate read-out of hard-to-measure disease processes.

Nuclear factor, erythroid 2-like 2-associated molecular signature predicts lung cancer survival.

Nuclear factor, erythroid 2-like 2 (NFE2L2), a transcription factor also known as NF-E2-related factor 2 (Nrf2), is a key cytoprotective gene that regulates critical antioxidant and stress-responsive genes. Nrf2 has been demonstrated to be a promising therapeutic target and useful biomarker in malignant disease. We hypothesized that NFE2L2-mediated gene expression would reflect cancer severity and progression. We conducted a meta-analysis of microarray data for 240 NFE2L2-mediated genes that were enriched in tumor tissues. We then developed a risk scoring system based on NFE2L2 gene expression profiling and designated 50 tumor-associated genes as the NFE2L2-associated molecular signature (NAMS). We tested the relationship between this gene expression signature and both recurrence-free survival and overall survival in lung cancer patients. We find that NAMS predicts clinical outcome in the training cohort and in 12 out of 20 validation cohorts. Cox proportional hazard regressions indicate that NAMS is a robust prognostic gene signature, independent of other clinical and pathological factors including patient age, gender, smoking, gene alteration, MYC level, and cancer stage. NAMS is an excellent predictor of recurrence-free survival and overall survival in human lung cancer. This gene signature represents a promising prognostic biomarker in human lung cancer.

The influence of cancer tissue sampling on the identification of cancer characteristics.

Cancer tissue sampling affects the identification of cancer characteristics. We aimed to clarify the source of differentially expressed genes (DEGs) in macro-dissected cancer tissue and develop a robust prognostic signature against the effects of tissue sampling. For estrogen receptor (ER)+ breast cancer patients, we identified DEGs in macro-dissected cancer tissues, malignant epithelial cells and stromal cells, defined as Macro-Dissected-DEGs, Epithelial-DEGs and Stromal-DEGs, respectively. Comparing Epithelial-DEGs to Stromal-DEGs (false discovery rate (FDR) < 10%), 86% of the overlapping genes exhibited consistent dysregulation (defined as Consistent-DEGs), and the other 14% of genes were dysregulated inconsistently (defined as Inconsistent-DEGs). The consistency score of dysregulation directions between Macro-Dissected-DEGs and Consistent-DEGs was 91% (P-value < 2.2 × 10−16, binomial test), whereas the score was only 52% between Macro-Dissected-DEGs and Inconsistent-DEGs (P-value = 0.9, binomial test). Among the gene ontology (GO) terms significantly enriched in Macro-Dissected-DEGs (FDR < 10%), 18 immune-related terms were enriched in Inconsistent-DEGs. DEGs associated with proliferation could reflect common changes of malignant epithelial and stromal cells; DEGs associated with immune functions are sensitive to the percentage of malignant epithelial cells in macro-dissected tissues. A prognostic signature which was insensitive to the cellular composition of macro-dissected tissues was developed and validated for ER+ breast patients.

Abstract 4346: Prognostic gene signature for intermediate risk prostate cancer

Abstract The over treatment of prostate cancer patients is a significant concern, as recent clinical trials suggest that many patients are over treated which can lead to significant patient morbidity. Although the Gleason score is a powerful predictor of lethal or indolent disease, a significant proportion (∼40%) of men present with early stage Gleason score (GS) 7 tumors, for whom prognosis is variable. The goal of this study is to develop and optimize a robust prognostic gene signature that can be utilized on formalin fixed paraffin embedded (FFPE) core biopsy tumor material to better classify patients with intermediate risk, GS 7 tumors into good and poor outcome groups. Three gene signatures were derived from publicly available gene expression profiles of the Swedish Watchful Waiting cohort. The Genomic Grade Index consisted of the top 25 molecular signatures discriminating between high (8, 9 &amp; 10) and low (≤ 6) GS tumors. The Lethal Gene Score consisted of the top 25 molecular signatures discriminating between lethal and indolent disease within GS 7 tumors only. A network-based gene signature consisted of 88 genes which accurately stratified GS 7 patients into high risk and low risk groups, resembling the survival curves of high GS and low GS patients. The prognostic capacity of the combined gene signature was tested in silico on the gene expression profiles of the Mayo cohort. Results demonstrated the gene signature's highly robust capacity for differentiating low risk and high risk patients within GS 7 patients. The NanoString nCounter System will be used to quantify mRNA from prostate FFPE blocks to assess the expression of the 138 prognostic genes. 156 archived prostate tumor blocks will be collected from intermediate risk, GS 7 patients enrolled in the 2005 PR5 prostate trial, which also collected 12 years of clinical follow-up information. Results will be correlated with biochemical (PSA) failure rates and overall survival. In short, our findings provide proof-of-principle that through the use of gene signatures it is possible to separate prostate cancer patients of intermediate risk into good and poor outcome groups. Furthermore, they also identify multiple gene candidates whose expression could likely be formulated into a clinically applicable assay, the implementation of which could serve to stratify prostate cancer patients with tumors of intermediate risk into more accurate high and low risk groups. Citation Format: Brian Li, Robin Hallet, Ying Wu, Greg Pong, John Hassell, Sebastien Hotte, Mark Levine, Himansu Lukka, Anita Bane. Prognostic gene signature for intermediate risk prostate cancer. [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 4346. doi:10.1158/1538-7445.AM2015-4346

Evaluation of public cancer datasets and signatures identifies TP53 mutant signatures with robust prognostic and predictive value.

BackgroundSystematic analysis of cancer gene-expression patterns using high-throughput transcriptional profiling technologies has led to the discovery and publication of hundreds of gene-expression signatures. However, few public signature values have been cross-validated over multiple studies for the prediction of cancer prognosis and chemosensitivity in the neoadjuvant setting.MethodsTo analyze the prognostic and predictive values of publicly available signatures, we have implemented a systematic method for high-throughput and efficient validation of a large number of datasets and gene-expression signatures. Using this method, we performed a meta-analysis including 351 publicly available signatures, 37,000 random signatures, and 31 breast cancer datasets. Survival analyses and pathologic responses were used to assess prediction of prognosis, chemoresponsiveness, and chemo-drug sensitivity.ResultsAmong 31 breast cancer datasets and 351 public signatures, we identified 22 validation datasets, two robust prognostic signatures (BRmet50 and PMID18271932Sig33) in breast cancer and one signature (PMID20813035Sig137) specific for prognosis prediction in patients with ER-negative tumors. The 22 validation datasets demonstrated enhanced ability to distinguish cancer gene profiles from random gene profiles. Both prognostic signatures are composed of genes associated with TP53 mutations and were able to stratify the good and poor prognostic groups successfully in 82%and 68% of the 22 validation datasets, respectively. We then assessed the abilities of the two signatures to predict treatment responses of breast cancer patients treated with commonly used chemotherapeutic regimens. Both BRmet50 and PMID18271932Sig33 retrospectively identified those patients with an insensitive response to neoadjuvant chemotherapy (mean positive predictive values 85%-88%). Among those patients predicted to be treatment sensitive, distant relapse-free survival (DRFS) was improved (negative predictive values 87%-88%). BRmet50 was further shown to prospectively predict taxane-anthracycline sensitivity in patients with HER2-negative (HER2-) breast cancer.ConclusionsWe have developed and applied a high-throughput screening method for public cancer signature validation. Using this method, we identified appropriate datasets for cross-validation and two robust signatures that differentiate TP53 mutation status and have prognostic and predictive value for breast cancer patients.Electronic supplementary materialThe online version of this article (doi:10.1186/s12885-015-1102-7) contains supplementary material, which is available to authorized users.

Abstract A72: Do prognostic gene signatures exist in Ewing sarcoma? A report from the Children's Oncology Group

Abstract Background &amp; Objectives: Relapse of Ewing sarcoma (ES) can occur months or years after initial remission, and salvage therapy for relapsed disease is usually ineffective. There is great need to develop biomarkers that can predict which patients are at risk for relapse, so that therapy and post-therapy evaluation can be adjusted accordingly. In the current study we sought to identify prognostic gene expression signatures in patients diagnosed with ES. Methods: Specimens were obtained from the Children's Oncology Group (COG) Biorepository. All samples were prospectively acquired from patients treated on clinical trials INT-0154 and AEWS0031. Criteria for inclusion included confirmation of localized ES, registration on a clinical trial and availability of frozen tumor. Total RNA was processed for whole genome expression profiling using Affymetrix GeneChip Human Exon 1.0 ST arrays. Affymetrix Power Tools (APT) were used to generate normalized gene-level signal intensity estimates from raw CEL file data. Differentially expressed genes with false discovery rate (FDR) of &amp;lt; 0.2 and absolute fold-change &amp;gt; 1.3 were considered to be significantly associated with outcome (survivor vs. non-survivor, relapse vs. no-relapse). Differentially expressed genes were used to create prognostic gene signatures. An independent group of ES samples from the Euro-Ewing tumor Biorepository in Muenster, Germany was similarly analyzed as a validation cohort. Results: Frozen tumor tissue was available from 254 COG patients. Following RNA and sample QC, Affymetrix CEL files were successfully generated from 56 unique patients with localized disease for whom outcome data were available. Analysis of processed gene expression data led to exclusion of 10 cases from further analysis due to issues of batch effect and outlier status. The demographic, event-free (EFS), and overall (OS) characteristics of the remaining 46 cases were shown to be representative of the INT-0154 and AEWS0031 studies as a whole. Unsupervised clustering of transcript-level data failed to demonstrate segregation of the 46 tumors into groups based on relapse or survival status. Supervised analysis of survivors vs. non-survivors identified a small number of differentially expressed genes and several statistically significant gene signatures, but none of these candidate classifiers could be internally validated. Interestingly, gene set enrichment analysis (GSEA) reproducibly revealed that integrin and chemokine receptor pathways were significantly over-represented as discriminators of EFS and OS (FDR&amp;lt;0.25; nominal p&amp;lt;0.01). However, the significance of these pathways was limited to tumor samples that contained at least 30% stromal cell contamination. Analysis of an independent cohort of 39 tumors from the Euro-Ewing biorepository (all with &amp;lt;30% stromal content) failed to identify prognostic genes, and no pathways reached statistical significance upon GSEA analysis. Conclusions: Robust prognostic gene signatures that pass internal and external cross-validation were not identified in this study. These findings demonstrate the profound degree of molecular heterogeneity in ES and suggest that the heterogeneous nature of these tumors is likely to preclude identification of prognostic signatures that will be clinically useful for all cases. Whether prognostic gene signatures will be useful for differentiating among subsets of ES cases is unknown and will require larger cohort analyses. Nevertheless, these studies support recent laboratory-based discoveries implicating cell adhesion and chemokine receptor signaling in ES aggression. Further investigation of these pathways in the context of tumor cell:tumor stroma interactions is warranted. Citation Format: Samuel L. Volchenboum, Jorge Andrade, Donald A. Barkauskas, Mark Krailo, Richard Sposto, Andreas Ranft, Jenny Potratz, Uta Dirksen, Timothy J. Triche, Elizabeth Lawlor. Do prognostic gene signatures exist in Ewing sarcoma? A report from the Children's Oncology Group. [abstract]. In: Proceedings of the AACR Special Conference on Pediatric Cancer at the Crossroads: Translating Discovery into Improved Outcomes; Nov 3-6, 2013; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2013;74(20 Suppl):Abstract nr A72.

Significance Analysis of Prognostic Signatures

A major goal in translational cancer research is to identify biological signatures driving cancer progression and metastasis. A common technique applied in genomics research is to cluster patients using gene expression data from a candidate prognostic gene set, and if the resulting clusters show statistically significant outcome stratification, to associate the gene set with prognosis, suggesting its biological and clinical importance. Recent work has questioned the validity of this approach by showing in several breast cancer data sets that “random” gene sets tend to cluster patients into prognostically variable subgroups. This work suggests that new rigorous statistical methods are needed to identify biologically informative prognostic gene sets. To address this problem, we developed Significance Analysis of Prognostic Signatures (SAPS) which integrates standard prognostic tests with a new prognostic significance test based on stratifying patients into prognostic subtypes with random gene sets. SAPS ensures that a significant gene set is not only able to stratify patients into prognostically variable groups, but is also enriched for genes showing strong univariate associations with patient prognosis, and performs significantly better than random gene sets. We use SAPS to perform a large meta-analysis (the largest completed to date) of prognostic pathways in breast and ovarian cancer and their molecular subtypes. Our analyses show that only a small subset of the gene sets found statistically significant using standard measures achieve significance by SAPS. We identify new prognostic signatures in breast and ovarian cancer and their corresponding molecular subtypes, and we show that prognostic signatures in ER negative breast cancer are more similar to prognostic signatures in ovarian cancer than to prognostic signatures in ER positive breast cancer. SAPS is a powerful new method for deriving robust prognostic biological signatures from clinically annotated genomic datasets.

A network module-based method for identifying cancer prognostic signatures

Discovering robust prognostic gene signatures as biomarkers using genomics data can be challenging. We have developed a simple but efficient method for discovering prognostic biomarkers in cancer gene expression data sets using modules derived from a highly reliable gene functional interaction network. When applied to breast cancer, we discover a novel 31-gene signature associated with patient survival. The signature replicates across 5 independent gene expression studies, and outperforms 48 published gene signatures. When applied to ovarian cancer, the algorithm identifies a 75-gene signature associated with patient survival. A Cytoscape plugin implementation of the signature discovery method is available at http://wiki.reactome.org/index.php/Reactome_FI_Cytoscape_Plugin

Development and validation of a robust prognostic and predictive signature for colorectal cancer (CRC) patients

4036 Background: Between 25 and 35% of stage II CRC patients will experience a recurrence of their disease and may benefit from adjuvant chemotherapy. Official guidelines give suggestions but no clear recommendation for best risk stratification. Here we describe the development a robust signature that predicts disease relapse and can assist in treatment decisions. Methods: Fresh frozen tumor tissues from 180 patients with stage I, II and III colorectal cancer undergoing surgery were analyzed using high density Agilent 44K oligonucleotide arrays. Median FU was 70.2 months; 85% of patients did not receive adjuvant chemotherapy. Unsupervised hierarchical clustering based on full-genome gene expression measurement indicated the existence of 3 main colon molecular subclasses. Survival analysis of the 3 classes showed that subtype C (n= 27) had a poor outcome and subtype A (n= 48) good outcome. Only the intermediate group B (n=104) was used to develop a signature by using a cross validation procedure to score all genes for their association with 5-yr distant metastasis free survival (DMFS) and subsequently applied to all samples (n=180). The obtained gene signature was further validated on an independent cohort of 178 stage II + III colon samples. Results: A set of 38 prognosis related gene probes showed robust DMFS association in over 50% of all iterations in the Training Set of 180 samples. The gene signature was validated on an independent cohort of 178 samples from stage II + III colon cancer patients. The profile classified 61% of the validation samples as low-risk and 39% as high-risk. The low- and high-risk samples showed a significant difference in DMFS with a HR of 3.19 (P= 8.5e-4). Five-year DMFS rates were 89% (95%CI 83–95) for low-risk and 62% (95%CI 50–77) for high-risk samples. Moreover, the profile showed a significant performance within stage II (P=0.0058) and III (P=0.036) only samples. The performance of the profile was significant for both untreated (P=0.0082) and treated patients (P=0.016) suggesting that its power is independent of treatment benefits. Conclusions: ColoPrint is able to predict the prognosis of stage II and III colon cancer patients and facilitates the identification of patients who would benefit from adjuvant chemotherapy. [Table: see text]