Abstract
Simple SummaryThere is a tremendous amount of gene expression information available for prostate cancer, but very few tools exist to combine the disparate datasets generated across sample types and technical platforms. We present a method of integrating different types of expression data from different study cohorts to increase analytic power, and improve our understanding of the molecular changes underlying the development and progression of prostate cancer from normal to advanced disease. Using this approach, we identified nine additional disease stage-specific candidate genes with prognostic significance, which were not identified in any one study alone. We have developed a free online tool summarizing our results, and making the complete combined dataset available for further translational research.Next-generation sequencing of primary tumors is now standard for transcriptomic studies, but microarray-based data still constitute the majority of available information on other clinically valuable samples, including archive material. Using prostate cancer (PC) as a model, we developed a robust analytical framework to integrate data across different technical platforms and disease subtypes to connect distinct disease stages and reveal potentially relevant genes not identifiable from single studies alone. We reconstructed the molecular profile of PC to yield the first comprehensive insight into its development, by tracking changes in mRNA levels from normal prostate to high-grade prostatic intraepithelial neoplasia, and metastatic disease. A total of nine previously unreported stage-specific candidate genes with prognostic significance were also found. Here, we integrate gene expression data from disparate sample types, disease stages and technical platforms into one coherent whole, to give a global view of the expression changes associated with the development and progression of PC from normal tissue through to metastatic disease. Summary and individual data are available online at the Prostate Integrative Expression Database (PIXdb), a user-friendly interface designed for clinicians and laboratory researchers to facilitate translational research.
Highlights
Prostate cancer (PC) is the second most common cancer diagnosis in men worldwide, and the fifth leading cause of cancer-related death [1]
We identified novel gene expression alterations associated with the transition from normal prostate to high-grade prostatic intra-epithelial neoplasia (HGPIN), localized cancer and metastatic disease (Figure 3)
We found substantial overlap (35 of 39) between the pathways identified in our study and those identified in a meta-analysis of 18 array datasets, where Ingenuity Pathway Analysis (IPA) was used to identify canonical pathways involved in the transition from normal prostate to localized and metastatic disease [70]
Summary
Prostate cancer (PC) is the second most common cancer diagnosis in men worldwide, and the fifth leading cause of cancer-related death [1]. It is clinically and pathologically heterogeneous, presenting a challenge to identifying robust molecular biomarkers for patient management. The limited statistical power of individual studies, different experimental and analytical parameters, differences in tumor tissue content, and inter- and intra-tumor heterogeneity contribute to the poor overlap between currently available molecular signatures [3]. A significant challenge to effective translational research more generally, is that the sheer volume of ‘-omics’ data available from individual studies is not matched by appropriate tools to synthesize disparate datasets, from different disease stages and generated on different platforms, into a coherent whole.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have