VanderSpek_edit.qxp 4/6/07 4:59 pm Page 19
Scaling Up Clinical Research by Expansion of Electronic Infrastructure
25,000 individuals. Further progress now requires the introduction of Our aim is to develop an ensemble approach to optimise classification
even more advanced IT that can cope with the enormous quantities of within a patient cohort using existing methods such as linear
data involved. Alongside delivering services and technology to all discriminates, support vector machines, random forest and neural
medical disciplines in Erasmus MC, the department of bioinformatics nets. Within this framework we plan to explore the relationship
runs a research programme of its own, providing the biomedical and between sample size, outcome, methods, validation and parameters
technological basis for such activities. It concentrates on the way the for optimal patient stratification. These methods are part of our
whole genome contributes to the evolution, development, structure and biobank and biomarker activities in collaboration with the Erasmus MC
function of the brain. It involves analysis of gene expression in brain department of haematology to develop novel superior molecular
cells, combining genomics, proteomics, imaging and cytogenetic data to diagnostic and prognostic solutions in the field of acute myeloid
identify genes associated with neurological disorders (see Figure 2). leukaemia (AML). Novel and existing methods will be used to define
whether the gene expression pattern (e.g. known pathways, specific
Progressive Research areas of the genome or clusters from a cluster analysis) over the whole
The research data warehouse helps scientists with their biomarker group of genes is related to a clinical outcome.
discovery activities and enables them to easily access content from their
desktop and provide an integrated knowledge base for drug and disease In parallel, we explore the topological properties of transcriptional
biomarkers. Erasmus MC currently implements an ‘-omics’ analytical network and apply different approaches to inferring causal associations
processing platform in which scientists are given support for statistical among genes by integrating genotypic and expression data – a
methods and experimental design. To ensure that data are of the highest necessary first step in reconstructing pathways associated with complex
quality possible, we are developing a standard operating procedure for traits. Our aim is to demonstrate core functional modules making up
quality assessment. This will guarantee the best experimental accuracy the transcriptional networks that are readily identified in these data and
and reproducibility of expression microarrays, DNA SNP chip analyses and that are coherent for several core biological processes associated with
proteomics and metabolomics data, whether the data are from Erasmus disease traits.
MC or from our external collaborators.
Skyline Diagnostics has developed an AML gene signature
1
that will be
For example, there are many methods that can differentiate between used to diagnose AML subtypes. Crosslinks, an Erasmus MC spin-out
diseased and/or treated patients. However, to ensure the analytical company, will develop a robust software platform that is compliant with
validity of such a selection process requires proper estimation of the FDA’s 21 CFR part 11 guidelines that regulate the security and
performance. A test that measures many analytes but with too few reliability of electronic data for the diagnosis of AML. This will provide
patients can easily derive biomarkers that are not based on biology, but secure access to the diagnostic application, maintain an audit trail and
that arise because of random variations. Many published studies ensure high availability of the data from the diagnostic site to the data-
inadequately estimate future biomarker performance and fail to test analysis and management centre.
statistical significance using the class permutation test.
Bird’s-eye View of Erasmus MC
Conclusion
With the flood of data across all biomedical disciplines in large multi-
disciplinary environments such as Erasmus MC, information visualisation
is emerging as a critical component of discovery, diagnostic care and
clinical decision-making processes.
Evidence-based, personalised treatment strategies require a new class of
visualisations that are capable of displaying various data types collected
from multidisciplinary knowledge and interactions between clinicians and
supporting staff. The ability to use visualisations to cross domains and
data types provides the ability to integrate analyses and support fast,
effective clinical decisions.
With the fast-growing amounts of raw data, information and
knowledge, the need for infrastructure grows exponentially. Clinical data
need to be coupled with electronic medical records, internal biobanks
and external public domain databases. This requires a secure, scalable
research data warehouse infrastructure that needs to be developed and
maintained. Creating awareness with government, healthcare authorities
and insurance companies is important since the biomedical informatics
field progresses rapidly; also, the average budget that hospitals have to
spend on ICT is around 3% of its total budget, in contrast to the banking
The city of Rotterdam selects and supports activities such as those described above through
its strategic Economic Vision 2020. industry, which invests an impressive 11%. ■
1. Valk PJ, Verhaak RG, Beijen MA, et al., Prognostically useful gene-expression profiles in acute myeloid leukemia, N Engl J Med, 2004;350(16):1617–28.
DATA MANAGEMENT IN PHARMACEUTICAL RESEARCH & DEVELOPMENT 19
Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26