TL;DR
Researchers have developed PMBC, a manually curated database of 1070 prognostic markers for breast cancer, encompassing mRNAs, lncRNAs, miRNAs, and circRNAs. This resource aims to consolidate scattered information, facilitating the identification and prioritization of biomarkers for precision treatment and accelerating translational research.
What Problem Does This Paper Address?
Breast cancer, despite advances, remains a significant cause of mortality due to its heterogeneity and the development of therapeutic resistance. While numerous biomarkers have been identified to predict patient outcomes, this information is fragmented across numerous publications. Existing tools for prognostic marker identification often rely on computational predictions rather than manual curation, and there's a lack of a centralized, comprehensive database specifically for breast cancer prognostic markers that includes various RNA types. This fragmentation hinders efficient research and the translation of findings into clinical practice.
Methods
The authors employed a systematic approach to build the PMBC database. They performed a literature search on PubMed using specific keywords and date ranges, focusing on publications with an impact factor greater than 5.0 to ensure a certain level of scientific rigor. The identified prognostic markers were then curated manually.
To understand the relationships between these markers, they constructed a competing endogenous RNA (ceRNA) network using data from starBase v2.0. This network analysis involved calculating topological features such as degree, closeness centrality, and betweenness for each RNA. Statistical comparisons were made between the topological features of prognostic RNAs and randomly selected RNAs to identify significant differences. Visualization of the ceRNA network was done using Cytoscape.
The database itself was built using a standard web development stack: Tomcat as the web server, Asynchronous JavaScript and XML (AJAX) for front-end/back-end interactions, and MySQL for data storage and management.
Datasets
The primary dataset for PMBC was derived from manually curated publications found in PubMed. The authors analyzed 714 publications, from which they identified 1070 prognostic markers. For the ceRNA network analysis, data was sourced from starBase v2.0.
Key Insights and Ideas
- Centralization is Key: The fragmentation of prognostic marker information is a significant bottleneck in breast cancer research. PMBC addresses this by creating a single, curated repository.
- Beyond mRNAs: The database acknowledges the growing importance of non-coding RNAs (lncRNAs, miRNAs, circRNAs) as prognostic markers, moving beyond traditional mRNA-focused analyses.
- ceRNA Network Insights: The analysis of the ceRNA network reveals that known prognostic RNAs exhibit higher closeness centrality than random RNAs, suggesting they are more interconnected and potentially play crucial roles in regulatory pathways.
- lncRNAs as Key Regulators: Prognostic lncRNAs show higher degrees and normalized closeness in the ceRNA network, indicating their significant role in maintaining interactions within the network, which could be leveraged for prioritizing them.
- User-Friendly Access: The database is designed with a user-friendly interface, allowing for browsing, searching, and even submission of new markers, fostering community contribution.
Key Results
- PMBC contains 1070 manually curated prognostic markers for breast cancer, sourced from 714 publications.
- The markers are distributed as follows: 81.21% mRNA, 6.54% lncRNA, 9.06% miRNA, 2.99% circRNA, and 0.19% pseudogene.
- The most frequently reported prognostic marker is HIF1A, followed by AR, EGFR, and MIK67.
- The journal 'Cancers (Basel)' published the highest number of prognostic markers.
- Functional enrichment analysis revealed that prognostic markers are significantly enriched in pathways related to epithelial cell proliferation, MAPK cascade, and gland development.
- The ceRNA network analysis identified 631 prognostic markers with interactions, including 602 mRNAs and 28 lncRNAs.
- Prognostic RNAs, particularly lncRNAs and mRNAs, demonstrated significantly higher normalized closeness compared to random RNAs (P < 1.00E-05).
Limitations and Future Work
- CircRNA Interactions: The current ceRNA network analysis does not include interactions for circRNAs, which is a limitation. The authors plan to reconstruct the ceRNA network to incorporate more RNA types, including circRNAs, in future work.
- Continuous Updates: The database will be updated every six months with newly reported prognostic indicators and user-submitted materials.
- Broader Scope: Future extensions aim to expand the database with more comprehensive ceRNA interactions and a wider range of prognostic markers to improve its utility and coverage. The ultimate goal is to provide a global and precise understanding of prognostic markers for breast cancer.
Summary updated