Survival prediction and gene identification with penalized global AUC maximization

A six-gene prognostic signature for both adult and pediatric acute myeloid leukemia identified with machine learning

BACKGROUND

Although it is well-known that adult and pediatric acute myeloid leukemias (AMLs) are genetically distinct diseases, they still share certain gene expression profiles. The age-related genetic heterogeneities of AMLs have been well-studied, but the common prognostic signatures and molecular mechanisms of adult and pediatric AMLs are less investigated.

AIM

To identify genes and pathways that are associated with both pediatric and adult AMLs and discover a gene signature for overall survival (OS) prediction.

METHODS

Through mining the transcriptome profiles of The Cancer Genome Atlas (TCGA) data sets of adult cancers and The Therapeutically Applicable Research to Generate Effective Treatments (TARGET) data of pediatric cancers, we identified genes that are commonly dysregulated in both pediatric and adult AMLs, further discovered a common gene signature, and built two risk score models for TCGA and TARGET cohorts, respectively with L ₀ regularized global AUC (area under the receiver operating characteristic curve) summary maximization.

RESULTS

We identified 57 genes that are differentially expressed and prognostically significant in both adult and childhood AMLs. The top 4 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways enriched with those 57 genes include transcriptional misregulation, focal adhesion, PI3K-Akt signaling pathway, and signaling pathways regulating pluripotency of stem cells. We further identified a 6-gene signature including genes of ADAMTS3, DNMT3B, NYNRIN, SORT1, ZFHX3, and ZG16B for risk prediction. We constructed a risk score model with one dataset (either TCGA or TARGET) and evaluated its performance with the other. The test AUCs for the risk prediction of TCGA data with a 2-year and 5-year OS cutoffs are 0.762 (P = 2.33e-13, 95% CI: 0.69-0.83) and 0.759 (P = 7.26e-08, 95% CI: 0.66-0.85), respectively, while the test AUCs of TARGET data with the same cutoffs are 0.71 (P = 3.3e-07, 95% CI: 0.62-0.79) and 0.72 (P= 5.25e-09, 95% CI: 0.65-0.80), respectively. We further stratified patients into 3 equal sized prognostic subtypes with the 6-gene risk scores. The P-values of the tertile partitions are 1.74e-07 and 3.28e-08 for the TARGET and TCGA cohorts, respectively, which are significantly better than the standard cytogenetic risk stratification of both cohorts (TARGET: P = 1.64e-06; TCGA: P = 1.79e-05). When validated with two other independent cohorts, the 6-gene risk score models remain a significant predictor for OS. Investigating the common gene expression program is significant in that we may extrapolate the findings from adults to children and avoid unnecessary pediatric clinical trials.

Extending information retrieval methods to personalized genomic-based studies of disease

Genomic-based studies of disease now involve diverse types of data collected on large groups of patients. A major challenge facing statistical scientists is how best to combine the data, extract important features, and comprehensively characterize the ways in which they affect an individual’s disease course and likelihood of response to treatment. We have developed a survival-supervised latent Dirichlet allocation (survLDA) modeling framework to address these challenges. Latent Dirichlet allocation (LDA) models have proven extremely effective at identifying themes common across large collections of text, but applications to genomics have been limited. Our framework extends LDA to the genome by considering each patient as a “document” with “text” detailing his/her clinical events and genomic state. We then further extend the framework to allow for supervision by a time-to-event response. The model enables the efficient identification of collections of clinical and genomic features that co-occur within patient subgroups, and then characterizes each patient by those features. An application of survLDA to The Cancer Genome Atlas ovarian project identifies informative patient subgroups showing differential response to treatment, and validation in an independent cohort demonstrates the potential for patient-specific inference.

Efficient support vector machine method for survival prediction with SEER data

Support vector machine (SVM) is a popular method for classification, but there are few methods that utilize SVM for survival analysis in the literature because of the computational complexity. In this paper, we develop a novel [Formula: see text] penalized SVM method for mining right-censored survival data ([Formula: see text] SVMSURV). Our proposed method can simultaneously identify survival-associated prognostic factors and predict survival outcomes. It is easy to understand and efficient to use especially when applied to large datasets. Our method has been examined through both simulation and real data, and its performance is very good with limited experiments.

Kernel based methods for accelerated failure time model with ultra-high dimensional data

Background

Most genomic data have ultra-high dimensions with more than 10,000 genes (probes). Regularization methods with L₁ and L_p penalty have been extensively studied in survival analysis with high-dimensional genomic data. However, when the sample size n ≪ m (the number of genes), directly identifying a small subset of genes from ultra-high (m > 10, 000) dimensional data is time-consuming and not computationally efficient. In current microarray analysis, what people really do is select a couple of thousands (or hundreds) of genes using univariate analysis or statistical tests, and then apply the LASSO-type penalty to further reduce the number of disease associated genes. This two-step procedure may introduce bias and inaccuracy and lead us to miss biologically important genes.

Results

The accelerated failure time (AFT) model is a linear regression model and a useful alternative to the Cox model for survival analysis. In this paper, we propose a nonlinear kernel based AFT model and an efficient variable selection method with adaptive kernel ridge regression. Our proposed variable selection method is based on the kernel matrix and dual problem with a much smaller n × n matrix. It is very efficient when the number of unknown variables (genes) is much larger than the number of samples. Moreover, the primal variables are explicitly updated and the sparsity in the solution is exploited.

Conclusions

Our proposed methods can simultaneously identify survival associated prognostic factors and predict survival outcomes with ultra-high dimensional genomic data. We have demonstrated the performance of our methods with both simulation and real data. The proposed method performs superbly with limited computational studies.

Survival associated pathway identification with group Lp penalized global AUC maximization

It has been demonstrated that genes in a cell do not act independently. They interact with one another to complete certain biological processes or to implement certain molecular functions. How to incorporate biological pathways or functional groups into the model and identify survival associated gene pathways is still a challenging problem. In this paper, we propose a novel iterative gradient based method for survival analysis with group L_p penalized global AUC summary maximization. Unlike LASSO, L_p (p < 1) (with its special implementation entitled adaptive LASSO) is asymptotic unbiased and has oracle properties [1]. We first extend L_p for individual gene identification to group L_p penalty for pathway selection, and then develop a novel iterative gradient algorithm for penalized global AUC summary maximization (IGGAUCS). This method incorporates the genetic pathways into global AUC summary maximization and identifies survival associated pathways instead of individual genes. The tuning parameters are determined using 10-fold cross validation with training data only. The prediction performance is evaluated using test data. We apply the proposed method to survival outcome analysis with gene expression profile and identify multiple pathways simultaneously. Experimental results with simulation and gene expression data demonstrate that the proposed procedures can be used for identifying important biological pathways that are related to survival phenotype and for building a parsimonious model for predicting the survival times.