Search

Objectives: In genomics, variable selection and prediction accounting for the complex interrelationships between explanatory variables represent major challenges. Tree-based methods are powerful alternatives to classical regression models. We have recently proposed the generalized, partially linear, tree-based regression (GPLTR) procedure that integrates the advantages of generalized linear regression (allowing the incorporation of confounding variables) and of tree-based models. In this work, we use bagging to address a classical concern of tree-based methods: their instability., Methods: We present a bagged GPLTR procedure and three scores for variable importance. The prediction accuracy and the performance of the scores are assessed by simulation. The use of this procedure is exemplified by the analysis of a lung cancer data set. The aim is to predict the epidermal growth factor receptor (EGFR) mutation based on gene expression measurements, taking into account the ethnicity (confounder variable) and perform variable selection., Results: The procedure performs well in terms of prediction accuracy. The scores differentiate predictive variables from noise variables. Based on a lung adenocarcinoma data set, the procedure achieves good predictive performance for EGFR mutation and selects relevant genes., Conclusion: The proposed bagged GPLTR procedure performs well for prediction and variable selection., (2015 S. Karger AG, Basel.)

Background: For clinical genomic studies with high-dimensional datasets, tree-based ensemble methods offer a powerful solution for variable selection and prediction taking into account the complex interrelationships between explanatory variables. One of the key component of the tree-building process is the splitting criterion. For survival data, the classical splitting criterion is the Logrank statistic. However, the presence of a fraction of nonsusceptible patients in the studied population advocates for considering a criterion tailored to this peculiar situation. Results: We propose a bagging survival tree procedure for variable selection and prediction where the survival tree-building process relies on a splitting criterion that explicitly focuses on time-to-event survival distribution among susceptible patients. A simulation study shows that our method achieves good performance for the variable selection and prediction. Different criteria for evaluating the importance of the explanatory variables and the prediction performance are reported. Our procedure is illustrated on a genomic dataset with gene expression measurements from early breast cancer patients. Conclusions: In the presence of nonsusceptible patients among the studied population, our procedure represents an efficient way to select event-related explanatory covariates with potential higher-order interaction and identify homogeneous groups of susceptible patients. [ABSTRACT FROM AUTHOR]

Abstract: To identify genomic markers with consistent effect on tumor dynamics across multiple cancer series, discrimination indices based on proportional hazards models can be used since they do not depend heavily on the sample size. However, the underlying assumption of proportionality of the hazards does not always hold, especially when the studied population is a mixture of cured and uncured patients, like in early-stage cancers. We propose a novel index that quantifies the capability of a genomic marker to separate uncured patients, according to their time-to-event outcomes. It allows to identify genomic markers characterizing tumor growth dynamic across multiple studies. Simulation results show that our index performs better than classical indices based on the Cox model. It is neither affected by the sample size nor the cure rate fraction. In a cross-study of early-stage breast cancers, the index allows to select genomic markers with a potential consistent effect on tumor growth dynamics. [Copyright &y& Elsevier]

Background: In genomic medical studies, one of the major objectives is to identify genomic factors with a prognostic impact on time-to-event outcomes so as to provide new insights into the disease process. Selection usually relies on statistical univariate indices based on the Cox model. Such model assumes proportional hazards (PH) which is unlikely to hold for each genomic marker. Methods: In this paper, we introduce a novel pseudo-R² measure derived from a crossing hazards model and designed for the selection of markers with crossing effects. The proposed index is related to the score statistic and quantifies the extent of a genomic factor to separate patients according to their survival times and marker measurements. We also show the importance of considering genomic markers with crossing effects as they potentially reflect the complex interplay between markers belonging to the same pathway. Results: Simulations show that our index is not affected by the censoring and the sample size of the study. It also performs better than classical indices under the crossing hazards assumption. The practical use of our index is illustrated in a lung cancer study. The use of the proposed pseudo-R² allows the identification of cell-cycle dependent genes not identified when relying on the PH assumption. Conclusions: The proposed index is a novel and promising tool for selecting markers with crossing hazards effects. [ABSTRACT FROM AUTHOR]