A metadata approach for clinical data management in translational genomics studies in breast cancer

Abstract

BackgroundIn molecular profiling studies of cancer patients, experimental and clinical data are combined in order to understand the clinical heterogeneity of the disease: clinical information for each subject needs to be linked to tumour samples, macromolecules extracted, and experimental results. This may involve the integration of clinical data sets from several different sources: these data sets may employ different data definitions and some may be incomplete.MethodsIn this work we employ semantic web techniques developed within the CancerGrid project, in particular the use of metadata elements and logic-based inference to annotate heterogeneous clinical information, integrate and query it.ResultsWe show how this integration can be achieved automatically, following the declaration of appropriate metadata elements for each clinical data set; we demonstrate the practicality of this approach through application to experimental results and clinical data from five hospitals in the UK and Canada, undertaken as part of the METABRIC project (Molecular Taxonomy of Breast Cancer International Consortium).ConclusionWe describe a metadata approach for managing similarities and differences in clinical datasets in a standardized way that uses Common Data Elements (CDEs). We apply and evaluate the approach by integrating the five different clinical datasets of METABRIC.

Single-Source Tumor Documentation - Reusing Oncology Data for Different Purposes

“…Although this problem has been partially addressed (e.g. in the NCI-founded caBIG™) [20]andtheNationalHealthService(NHS)-funded CancerGridinitiative [21],toourknowledgenosingle-source referencearchitecturebasedon1standardizeddatasetexists. Wecouldalsofindnopublicationsonacomprehensivecancer center IT implementation process in which such a singlesourceapproachhasbeenstrictlyfollowedtosupportclinical care, quality assurance, cancer registration and translational research.…”

The CancerGrid experience: Metadata-based model-driven engineering for clinical trials

“…The CONSORT trials metamodel and the forms-based approach to model-driven engineering are described in a couple of conference papers [24,18]. The case studies on prospective data collection [23], retrospective data integration [25], and forms metamodelling [37] have each appeared in short papers.…”

“…One retrospective data analysis application of the CancerGrid approach was undertaken in the context of the Molecular Taxonomy of Breast Cancer Internal Consortium (METABRIC) [25], a collaborative study involving five hospitals in the UK and Canada, using molecular profiling in order to understand the clinical heterogeneity of breast cancers. The hospitals in the consortium all use different data definitions; they often hold incomplete datasets, because of patient movements; and record keeping and clinical procedures are likely to have changed over time during the relatively long treatment periods.…”

BeadDatapackR: A Tool to Facilitate the Sharing of Raw Data from Illumina BeadArray Studies

“…)12 and it has been shown that the analysis of raw Illumina BeadArray data, rather than the default summarized data, can be advantageous. However there is no data repository oriented towards storing these raw Illumina data, and there has been only limited success in storing such data in standard repositories 13,14.…”