Medical Statistics and Bioinformatics 1

Development and application of statistical models for medical research

Principal investigators

Prof. Dr R. Brand, Dr S. Böhringer, Dr S. le Cessie, Dr J.J. Goeman, Prof. Dr J.J. Houwing-Duistermaat, Dr B. Mertens, Prof. Dr H. Putter, Prof. Dr T. Stijnen, Dr E. van Zwet

Aim and focus

This project bundles all research activities of the section Medical Statistics. The goal of Medical Statistics is the application and development of statistical models and designs in medical research. This covers a very broad spectrum of research. Traditionally, the main involvements are design, analysis and reporting of clinical trials and modelling of observational patient data aimed at understanding the biological processes and obtaining prognostic models relevant for patient care and therapy. Research is focused on prognostic models, survival analysis including multi-state modelling, meta-analysis, statistical genetics, statistical bioinformatics, epidemiological methods and good research data management.
In the last decade we have considerably expanded our activity in the field of statistical genetics and bioinformatics. In bioinformatics the focus is on statistical models for classification and prognosis using expression data, on methods for multiple testing and on computational aspects of statistical models in high-performance computing. In statistical genetics the focus is on gene finding as well as building statistical models for complex diseases and phenotypes. The statistical genetics group has broadened its scope towards the analysis of all kinds of -omics data, including genomics, transcriptomics, epigenomics, proteomics, metabolomics, and glycomics as well as other high-dimensional data such as facial data.

Position in international context

The section of Medical Statistics is one of the main biostatistical groups in Europe. The research in prognostic modelling, survival analysis, genetic statistics, meta-analysis, high dimensional data-analysis and multiple testing is internationally well recognized. Papers from the group are highly cited, members of the group are frequently invited and there are many (informal) collaborations mainly in Europe. The national and international presence in the field of statistical –omics data analysis is presently well established, stimulated by the VIDI grant and appointment to full professor of Dr Jeanine Houwing-Duistermaat.

Content/highlights/achievements

• Tutorial on meta-analysis in 2002  Reprinted in collection of tutorials. Still much reaction by email of medical researchers applying the methods. Number of yearly citations is still growing.
• Development of global test for micro-array and other high dimensional data types. Very frequently used.
• Tutorial on competing risks and multi-state models (ref. 1). Highly cited, frequent invitations to teach international courses on this subject.
• ZonMW TOP-program grants for meta-analysis (2005-2010) and multi-state modelling (2007-2012).
• VIDI grant for prof. dr. Jeanine Houwing-Duistermaat (2006-2010) focusing on further development of linkage and/or association methodology for gene finding.
• Competition on Clinical Mass Spectrometry Based Proteomic Diagnosis, Leiden 2007, organized by dr. Bart Mertens resulting in a special issue of Statistical Applications in Genetics and Molecular Biology (2008) devoted to the competition (ref. 3)
• Book “Dynamic Prediction in Clinical Survival Analysis" (Chapman & Hall) by prof. dr. J.C. van Houwelingen en prof. dr. H. Putter. Appeared in 2011.
• The R-packages developed in recent years, mstate, globaltest, penalized and cherry, are very frequently used.
• Article with discussion in Statistical Science “Multiple testing for exploratory research” by J. J. Goeman and A. Solari. An article with discussion in this journal is very prestigious in biostatistics.
• Special issue of Biometrical Journal in honour of prof. dr. Hans van Houwelingen (2010), guest edited by prof. dr. H. Putter, dr. S. le Cessie and prof. dr. T. Stijnen.
• Participation as second biggest partner in the Marie Curie International Training Network “Novel Statistical Methodology for Diagnostic/Prognostic and Therapeutic Studies and Systematic Reviews” (2011)
• Long term research support of Genetical Statistics group by Reumafonds: collaboration with departments Immunohematology&Bloodtransfusion and Reumatology
• Pioneering paper on genetic association of facial traits (Boehringer et al. 2011) supported by two grants (academic grant Germany, DFG BO 1955/2-1, corporate grant Identitas)

Future themes

• New data types such as proteomics, epigenomics, and glycomics data will add an extra complexity to the statistical modelling.
• Methodology for gene mapping using genome wide association (GWA) and genome wide linkage (GWL) data, with special attention to longevity, and the combination of such information within and between studies.
• Combining data from different sources and different nature (GO-term pathway information and sibpair linkage data, for example)  in an inferential setting.
• Statistical methods for quality indicators.
• Methods of incorporating heterogeneity in survival analysis and in family data.
• Causal inference in observational data. 

Cohesion within LUMC

Traditionally, there is a close cooperation with most of the clinical groups. The large number of co-authorships (about 80 per year) is witness to that. One of our staff members has a joint appointment at the departments of Clinical Epidemiology and of Medical Statistics and Bioinformatics, to stimulate collaborations between these departments. The cooperation with the molecular (genetic) oriented departments in the field of bioinformatics and statistical genetics, which has started around 10 years ago, is presently well established and still growing.