MICRORNA EXPRESSION BASED OUTCOME PREDICTION IN AML ? CROSS-PLATFORM INTEGRATIVE ANALYSES LEAD THE WAY TO NOVEL INSIGHTS
(Abstract release date: 05/21/15)
EHA Library. Shivarov V. 06/13/15; 103124; S455
Disclosure(s): Sofiamed University HospitalLaboratory of Clinical Immunology
Dr. Velizar Shivarov
Contributions
Contributions
Abstract
Abstract: S455
Type: Oral Presentation + travel grant
Presentation during EHA20: From 13.06.2015 12:30 to 13.06.2015 12:45
Location: Room Lehar 1 + 2
Background
Recent advances in omics technologies allowed for affordable generation of large scale molecular data including microRNA expression profiling. These data are expected to serve as an invaluable source for molecular based classification and prognostication of genetically heterogeneous diseases such as acute myeloid leukemia (AML). Besides, the integration of various layers of omics data may serve for the generation of novel testable hypotheses regarding disease pathogenesis.
Aims
Here, we aimed to develop a novel microRNA expression based prognostic score using data from two different platforms – microarrays and RNA-Seq data from adult patients with de novo AML and to investigate the underlying biological pathways.
Methods
We used microRNA microarray expression data from 91 adult patients with de novo AML enrolled into the AMLSG treatment protocol AML HD98A (ClinicalTrials.gov Identifier: NCT00146120) (training dataset). Another group of 177 AML patients with microRNA expression profiling data (RNA-Seq) from the Cancer Genome Atlas (TCGA) served as a validation set. A scoring system was built using the Robust Likelihood-Based Survival Modeling with Microarray Data method. The selected model included 7 microRNAs (miR-100, miR-132, miR-185, miR-186, miR-302a, miR-330, and miR-422a). A total continuous score was calculated for each patient sample using the Cox regression coefficients obtained for the training dataset multiplied by the expression levels in each sample and summed afterwards. Discrete scores (low vs high) were defined using Receiver Operating Characteristics (ROC) curve analysis. In addition, bioinformatic analysis of the GEP, RNA-Seq and DNA methylation data from the TCGA was performed.
Results
For both the training and validation dataset the discrete scores were significant prognostic factors in univariate analysis (p=0.001 and p=0.002, respectively). The continuous score, which performed almost identically, was also a significant factor in multivariate analysis (p<0.001 and p=0.022, respectively). In an analysis restricted to cytogenetically normal (CN) AML the discrete score was a significant adverse prognostic factor in the training dataset based on the log-rank test (p=0.045), and it appeared as independent prognostic factor in the validation subset of younger CN-AML patients in a multivariate model including age, gender, FLT3-ITD and NPM1 mutational status (p=0.001). Next, we performed a network analysis of the 7 microRNAs and their known and putative targets and found an enrichment with nucleic acids binding proteins. Using the TCGA GEP data we identified 850 probe sets that were differentially expressed between Low and High Score patients at the level of p<0.01. GO analysis showed that “General transcription regulation” was the most significantly overrepresented pathway. Similarly, when Gene Set Enrichment Analysis (GSEA) was used for the subset of CN-AML <61 years the 7 top scoring gene signatures were associated with RNA metabolism and processing. The top scoring gene set was the “RNA SPLICING” signature. The differential exon usage (DEU) analysis between Low and High Score patients showed a total of 7500 differentially expressed tags at a level of significance of 0.05. Finally, we obtained the TCGA DNA methylation data and found a total of 1218 CpG sites differentially methylated between the Low and High Score patients and hierarchical clustering showed a very good correlation with the miRNA score.
Summary
We demonstrated the feasibility to integrate microRNA expression data from different platforms (microarray and RNA-Seq data) for building prognostic scores in AML. Moreover, the integrated analysis of omics data from microRNA expression score-defined subgroups provided further evidence for the potential biological relevance of these subgroups and a role of the RNA splicing machinery deregulation in the pathogenesis of AML.
Keyword(s): AML, Microarray analysis, Prognosis
Session topic: Molecular markers in AML
Type: Oral Presentation + travel grant
Presentation during EHA20: From 13.06.2015 12:30 to 13.06.2015 12:45
Location: Room Lehar 1 + 2
Background
Recent advances in omics technologies allowed for affordable generation of large scale molecular data including microRNA expression profiling. These data are expected to serve as an invaluable source for molecular based classification and prognostication of genetically heterogeneous diseases such as acute myeloid leukemia (AML). Besides, the integration of various layers of omics data may serve for the generation of novel testable hypotheses regarding disease pathogenesis.
Aims
Here, we aimed to develop a novel microRNA expression based prognostic score using data from two different platforms – microarrays and RNA-Seq data from adult patients with de novo AML and to investigate the underlying biological pathways.
Methods
We used microRNA microarray expression data from 91 adult patients with de novo AML enrolled into the AMLSG treatment protocol AML HD98A (ClinicalTrials.gov Identifier: NCT00146120) (training dataset). Another group of 177 AML patients with microRNA expression profiling data (RNA-Seq) from the Cancer Genome Atlas (TCGA) served as a validation set. A scoring system was built using the Robust Likelihood-Based Survival Modeling with Microarray Data method. The selected model included 7 microRNAs (miR-100, miR-132, miR-185, miR-186, miR-302a, miR-330, and miR-422a). A total continuous score was calculated for each patient sample using the Cox regression coefficients obtained for the training dataset multiplied by the expression levels in each sample and summed afterwards. Discrete scores (low vs high) were defined using Receiver Operating Characteristics (ROC) curve analysis. In addition, bioinformatic analysis of the GEP, RNA-Seq and DNA methylation data from the TCGA was performed.
Results
For both the training and validation dataset the discrete scores were significant prognostic factors in univariate analysis (p=0.001 and p=0.002, respectively). The continuous score, which performed almost identically, was also a significant factor in multivariate analysis (p<0.001 and p=0.022, respectively). In an analysis restricted to cytogenetically normal (CN) AML the discrete score was a significant adverse prognostic factor in the training dataset based on the log-rank test (p=0.045), and it appeared as independent prognostic factor in the validation subset of younger CN-AML patients in a multivariate model including age, gender, FLT3-ITD and NPM1 mutational status (p=0.001). Next, we performed a network analysis of the 7 microRNAs and their known and putative targets and found an enrichment with nucleic acids binding proteins. Using the TCGA GEP data we identified 850 probe sets that were differentially expressed between Low and High Score patients at the level of p<0.01. GO analysis showed that “General transcription regulation” was the most significantly overrepresented pathway. Similarly, when Gene Set Enrichment Analysis (GSEA) was used for the subset of CN-AML <61 years the 7 top scoring gene signatures were associated with RNA metabolism and processing. The top scoring gene set was the “RNA SPLICING” signature. The differential exon usage (DEU) analysis between Low and High Score patients showed a total of 7500 differentially expressed tags at a level of significance of 0.05. Finally, we obtained the TCGA DNA methylation data and found a total of 1218 CpG sites differentially methylated between the Low and High Score patients and hierarchical clustering showed a very good correlation with the miRNA score.
Summary
We demonstrated the feasibility to integrate microRNA expression data from different platforms (microarray and RNA-Seq data) for building prognostic scores in AML. Moreover, the integrated analysis of omics data from microRNA expression score-defined subgroups provided further evidence for the potential biological relevance of these subgroups and a role of the RNA splicing machinery deregulation in the pathogenesis of AML.
Keyword(s): AML, Microarray analysis, Prognosis
Session topic: Molecular markers in AML
Abstract: S455
Type: Oral Presentation + travel grant
Presentation during EHA20: From 13.06.2015 12:30 to 13.06.2015 12:45
Location: Room Lehar 1 + 2
Background
Recent advances in omics technologies allowed for affordable generation of large scale molecular data including microRNA expression profiling. These data are expected to serve as an invaluable source for molecular based classification and prognostication of genetically heterogeneous diseases such as acute myeloid leukemia (AML). Besides, the integration of various layers of omics data may serve for the generation of novel testable hypotheses regarding disease pathogenesis.
Aims
Here, we aimed to develop a novel microRNA expression based prognostic score using data from two different platforms – microarrays and RNA-Seq data from adult patients with de novo AML and to investigate the underlying biological pathways.
Methods
We used microRNA microarray expression data from 91 adult patients with de novo AML enrolled into the AMLSG treatment protocol AML HD98A (ClinicalTrials.gov Identifier: NCT00146120) (training dataset). Another group of 177 AML patients with microRNA expression profiling data (RNA-Seq) from the Cancer Genome Atlas (TCGA) served as a validation set. A scoring system was built using the Robust Likelihood-Based Survival Modeling with Microarray Data method. The selected model included 7 microRNAs (miR-100, miR-132, miR-185, miR-186, miR-302a, miR-330, and miR-422a). A total continuous score was calculated for each patient sample using the Cox regression coefficients obtained for the training dataset multiplied by the expression levels in each sample and summed afterwards. Discrete scores (low vs high) were defined using Receiver Operating Characteristics (ROC) curve analysis. In addition, bioinformatic analysis of the GEP, RNA-Seq and DNA methylation data from the TCGA was performed.
Results
For both the training and validation dataset the discrete scores were significant prognostic factors in univariate analysis (p=0.001 and p=0.002, respectively). The continuous score, which performed almost identically, was also a significant factor in multivariate analysis (p<0.001 and p=0.022, respectively). In an analysis restricted to cytogenetically normal (CN) AML the discrete score was a significant adverse prognostic factor in the training dataset based on the log-rank test (p=0.045), and it appeared as independent prognostic factor in the validation subset of younger CN-AML patients in a multivariate model including age, gender, FLT3-ITD and NPM1 mutational status (p=0.001). Next, we performed a network analysis of the 7 microRNAs and their known and putative targets and found an enrichment with nucleic acids binding proteins. Using the TCGA GEP data we identified 850 probe sets that were differentially expressed between Low and High Score patients at the level of p<0.01. GO analysis showed that “General transcription regulation” was the most significantly overrepresented pathway. Similarly, when Gene Set Enrichment Analysis (GSEA) was used for the subset of CN-AML <61 years the 7 top scoring gene signatures were associated with RNA metabolism and processing. The top scoring gene set was the “RNA SPLICING” signature. The differential exon usage (DEU) analysis between Low and High Score patients showed a total of 7500 differentially expressed tags at a level of significance of 0.05. Finally, we obtained the TCGA DNA methylation data and found a total of 1218 CpG sites differentially methylated between the Low and High Score patients and hierarchical clustering showed a very good correlation with the miRNA score.
Summary
We demonstrated the feasibility to integrate microRNA expression data from different platforms (microarray and RNA-Seq data) for building prognostic scores in AML. Moreover, the integrated analysis of omics data from microRNA expression score-defined subgroups provided further evidence for the potential biological relevance of these subgroups and a role of the RNA splicing machinery deregulation in the pathogenesis of AML.
Keyword(s): AML, Microarray analysis, Prognosis
Session topic: Molecular markers in AML
Type: Oral Presentation + travel grant
Presentation during EHA20: From 13.06.2015 12:30 to 13.06.2015 12:45
Location: Room Lehar 1 + 2
Background
Recent advances in omics technologies allowed for affordable generation of large scale molecular data including microRNA expression profiling. These data are expected to serve as an invaluable source for molecular based classification and prognostication of genetically heterogeneous diseases such as acute myeloid leukemia (AML). Besides, the integration of various layers of omics data may serve for the generation of novel testable hypotheses regarding disease pathogenesis.
Aims
Here, we aimed to develop a novel microRNA expression based prognostic score using data from two different platforms – microarrays and RNA-Seq data from adult patients with de novo AML and to investigate the underlying biological pathways.
Methods
We used microRNA microarray expression data from 91 adult patients with de novo AML enrolled into the AMLSG treatment protocol AML HD98A (ClinicalTrials.gov Identifier: NCT00146120) (training dataset). Another group of 177 AML patients with microRNA expression profiling data (RNA-Seq) from the Cancer Genome Atlas (TCGA) served as a validation set. A scoring system was built using the Robust Likelihood-Based Survival Modeling with Microarray Data method. The selected model included 7 microRNAs (miR-100, miR-132, miR-185, miR-186, miR-302a, miR-330, and miR-422a). A total continuous score was calculated for each patient sample using the Cox regression coefficients obtained for the training dataset multiplied by the expression levels in each sample and summed afterwards. Discrete scores (low vs high) were defined using Receiver Operating Characteristics (ROC) curve analysis. In addition, bioinformatic analysis of the GEP, RNA-Seq and DNA methylation data from the TCGA was performed.
Results
For both the training and validation dataset the discrete scores were significant prognostic factors in univariate analysis (p=0.001 and p=0.002, respectively). The continuous score, which performed almost identically, was also a significant factor in multivariate analysis (p<0.001 and p=0.022, respectively). In an analysis restricted to cytogenetically normal (CN) AML the discrete score was a significant adverse prognostic factor in the training dataset based on the log-rank test (p=0.045), and it appeared as independent prognostic factor in the validation subset of younger CN-AML patients in a multivariate model including age, gender, FLT3-ITD and NPM1 mutational status (p=0.001). Next, we performed a network analysis of the 7 microRNAs and their known and putative targets and found an enrichment with nucleic acids binding proteins. Using the TCGA GEP data we identified 850 probe sets that were differentially expressed between Low and High Score patients at the level of p<0.01. GO analysis showed that “General transcription regulation” was the most significantly overrepresented pathway. Similarly, when Gene Set Enrichment Analysis (GSEA) was used for the subset of CN-AML <61 years the 7 top scoring gene signatures were associated with RNA metabolism and processing. The top scoring gene set was the “RNA SPLICING” signature. The differential exon usage (DEU) analysis between Low and High Score patients showed a total of 7500 differentially expressed tags at a level of significance of 0.05. Finally, we obtained the TCGA DNA methylation data and found a total of 1218 CpG sites differentially methylated between the Low and High Score patients and hierarchical clustering showed a very good correlation with the miRNA score.
Summary
We demonstrated the feasibility to integrate microRNA expression data from different platforms (microarray and RNA-Seq data) for building prognostic scores in AML. Moreover, the integrated analysis of omics data from microRNA expression score-defined subgroups provided further evidence for the potential biological relevance of these subgroups and a role of the RNA splicing machinery deregulation in the pathogenesis of AML.
Keyword(s): AML, Microarray analysis, Prognosis
Session topic: Molecular markers in AML
{{ help_message }}
{{filter}}