Dataset Information

Can survival prediction be improved by merging gene expression data sets?

ABSTRACT:

Background

High-throughput gene expression profiling technologies generating a wealth of data, are increasingly used for characterization of tumor biopsies for clinical trials. By applying machine learning algorithms to such clinically documented data sets, one hopes to improve tumor diagnosis, prognosis, as well as prediction of treatment response. However, the limited number of patients enrolled in a single trial study limits the power of machine learning approaches due to over-fitting. One could partially overcome this limitation by merging data from different studies. Nevertheless, such data sets differ from each other with regard to technical biases, patient selection criteria and follow-up treatment. It is therefore not clear at all whether the advantage of increased sample size outweighs the disadvantage of higher heterogeneity of merged data sets. Here, we present a systematic study to answer this question specifically for breast cancer data sets. We use survival prediction based on Cox regression as an assay to measure the added value of merged data sets.

Results

Using time-dependent Receiver Operating Characteristic-Area Under the Curve (ROC-AUC) and hazard ratio as performance measures, we see in overall no significant improvement or deterioration of survival prediction with merged data sets as compared to individual data sets. This apparently was due to the fact that a few genes with strong prognostic power were not available on all microarray platforms and thus were not retained in the merged data sets. Surprisingly, we found that the overall best performance was achieved with a single-gene predictor consisting of CYB5D1.

Conclusions

Merging did not deteriorate performance on average despite (a) The diversity of microarray platforms used. (b) The heterogeneity of patients cohorts. (c) The heterogeneity of breast cancer disease. (d) Substantial variation of time to death or relapse. (e) The reduced number of genes in the merged data sets. Predictors derived from the merged data sets were more robust, consistent and reproducible across microarray platforms. Moreover, merging data sets from different studies helps to better understand the biases of individual studies and can lead to the identification of strong survival factors like CYB5D1 expression.

SUBMITTER: Yasrebi H

PROVIDER: S-EPMC2761544 | biostudies-literature | 2009 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Can survival prediction be improved by merging gene expression data sets?

Yasrebi Haleh H Sperisen Peter P Praz Viviane V Bucher Philipp P

PloS one 20091023 10

<h4>Background</h4>High-throughput gene expression profiling technologies generating a wealth of data, are increasingly used for characterization of tumor biopsies for clinical trials. By applying machine learning algorithms to such clinically documented data sets, one hopes to improve tumor diagnosis, prognosis, as well as prediction of treatment response. However, the limited number of patients enrolled in a single trial study limits the power of machine learning approaches due to over-fitting ...[more]

PMID: 19851466

Similar Datasets

Project description:BackgroundIndependently derived expression profiles of the same biological condition often have few genes in common. In this study, we created populations of expression profiles from publicly available microarray datasets of cancer (breast, lymphoma and renal) samples linked to clinical information with an iterative machine learning algorithm. ROC curves were used to assess the prediction error of each profile for classification. We compared the prediction error of profiles correlated with molecular phenotype against profiles correlated with relapse-free status. Prediction error of profiles identified with supervised univariate feature selection algorithms were compared to profiles selected randomly from a) all genes on the microarray platform and b) a list of known disease-related genes (a priori selection). We also determined the relevance of expression profiles on test arrays from independent datasets, measured on either the same or different microarray platforms.ResultsHighly discriminative expression profiles were produced on both simulated gene expression data and expression data from breast cancer and lymphoma datasets on the basis of ER and BCL-6 expression, respectively. Use of relapse-free status to identify profiles for prognosis prediction resulted in poorly discriminative decision rules. Supervised feature selection resulted in more accurate classifications than random or a priori selection, however, the difference in prediction error decreased as the number of features increased. These results held when decision rules were applied across-datasets to samples profiled on the same microarray platform.ConclusionOur results show that many gene sets predict molecular phenotypes accurately. Given this, expression profiles identified using different training datasets should be expected to show little agreement. In addition, we demonstrate the difficulty in predicting relapse directly from microarray data using supervised machine learning approaches. These findings are relevant to the use of molecular profiling for the identification of candidate biomarker panels.

Project description:BackgroundThirty to forty percent of patients with Diffuse Large B-cell Lymphoma (DLBCL) have an adverse clinical evolution. The increased understanding of DLBCL biology has shed light on the clinical evolution of this pathology, leading to the discovery of prognostic factors based on gene expression data, genomic rearrangements and mutational subgroups. Nevertheless, additional efforts are needed in order to enable survival predictions at the patient level. In this study we investigated new machine learning-based models of survival using transcriptomic and clinical data.MethodsGene expression profiling (GEP) of in 2 different publicly available retrospective DLBCL cohorts were analyzed. Cox regression and unsupervised clustering were performed in order to identify probes associated with overall survival on the largest cohort. Random forests were created to model survival using combinations of GEP data, COO classification and clinical information. Cross-validation was used to compare model results in the training set, and Harrel's concordance index (c-index) was used to assess model's predictability. Results were validated in an independent test set.ResultsTwo hundred thirty-three and sixty-four patients were included in the training and test set, respectively. Initially we derived and validated a 4-gene expression clusterization that was independently associated with lower survival in 20% of patients. This pattern included the following genes: TNFRSF9, BIRC3, BCL2L1 and G3BP2. Thereafter, we applied machine-learning models to predict survival. A set of 102 genes was highly predictive of disease outcome, outperforming available clinical information and COO classification. The final best model integrated clinical information, COO classification, 4-gene-based clusterization and the expression levels of 50 individual genes (training set c-index, 0.8404, test set c-index, 0.7942).ConclusionOur results indicate that DLBCL survival models based on the application of machine learning algorithms to gene expression and clinical data can largely outperform other important prognostic variables such as disease stage and COO. Head-to-head comparisons with other risk stratification models are needed to compare its usefulness.

Dataset Information

Can survival prediction be improved by merging gene expression data sets?

Background

Results

Conclusions

Publications

Can survival prediction be improved by merging gene expression data sets?

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets