A Data Similarity-Based Strategy for Meta-analysis of Transcriptional Profiles in Cancer

Abstract

BackgroundRobust transcriptional signatures in cancer can be identified by data similarity-driven meta-analysis of gene expression profiles. An unbiased data integration and interrogation strategy has not previously been available.Methods and FindingsWe implemented and performed a large meta-analysis of breast cancer gene expression profiles from 223 datasets containing 10,581 human breast cancer samples using a novel data similarity-based approach (iterative EXALT). Cancer gene expression signatures extracted from individual datasets were clustered by data similarity and consolidated into a meta-signature with a recurrent and concordant gene expression pattern. A retrospective survival analysis was performed to evaluate the predictive power of a novel meta-signature deduced from transcriptional profiling studies of human breast cancer. Validation cohorts consisting of 6,011 breast cancer patients from 21 different breast cancer datasets and 1,110 patients with other malignancies (lung and prostate cancer) were used to test the robustness of our findings. During the iterative EXALT analysis, 633 signatures were grouped by their data similarity and formed 121 signature clusters. From the 121 signature clusters, we identified a unique meta-signature (BRmet50) based on a cluster of 11 signatures sharing a phenotype related to highly aggressive breast cancer. In patients with breast cancer, there was a significant association between BRmet50 and disease outcome, and the prognostic power of BRmet50 was independent of common clinical and pathologic covariates. Furthermore, the prognostic value of BRmet50 was not specific to breast cancer, as it also predicted survival in prostate and lung cancers.ConclusionsWe have established and implemented a novel data similarity-driven meta-analysis strategy. Using this approach, we identified a transcriptional meta-signature (BRmet50) in breast cancer, and the prognostic performance of BRmet50 was robust and applicable across a wide range of cancer-patient populations.

SPARCL1 suppresses metastasis in prostate cancer

“…We recently reported an association between a 50‐gene expression signature and CaP metastasis (Qiu et al., 2013). We hypothesized that genes enriched in the 50‐gene signature are mechanistically involved in CaP metastasis.…”

“…The methods used for signature extraction, signature database development, and EXALT analysis were previously reported (Yi et al., 2007; Wu et al., 2009). Iterative EXALT analysis for identification of the 50‐gene expression signature and its association with CaP metastasis has been described elsewhere (Qiu et al., 2013).…”

“…There are a limited number of published CaP gene‐expression studies having clinical survival outcome data for meta‐analysis. We previously used a robust meta‐analysis of gene expression profiles from hundreds of breast cancer datasets (Yi et al., 2007; Wu et al., 2009; Qiu et al., 2013). Using this approach, we discovered a novel and conserved gene expression signature predictive of metastasis risk in multiple cancers (breast, lung, and prostate cancer) (Qiu et al., 2013).…”

“…We previously used a robust meta‐analysis of gene expression profiles from hundreds of breast cancer datasets (Yi et al., 2007; Wu et al., 2009; Qiu et al., 2013). Using this approach, we discovered a novel and conserved gene expression signature predictive of metastasis risk in multiple cancers (breast, lung, and prostate cancer) (Qiu et al., 2013). We hypothesized that this expression signature is enriched for genes that are mechanistically involved with cancer metastasis including CaP.…”

A Murine Model of K-RAS and β-Catenin Induced Renal Tumors Expresses High Levels of E2F1 and Resembles Human Wilms Tumor

“…Each human signature formed a signature cluster with the three human orthologous signatures based on their data similarities such as identical gene symbols, concordance in the directions of gene expression changes, and significant confidence scores 20 . Signature genes within each cluster were combined, and overlapping genes across four signatures were then compiled as a meta-signature by data similarity-driven meta-analysis (EXALT) of transcriptional profiles 21 . In so doing, a human renal cancer signature and three murine derived human orthologous signatures derived from Kras G12D /Catnb Δex3 generated a meta-signature.…”