Please activate JavaScript!
Please install Adobe Flash Player, click here for download

ISCB2014_abstract_book

28 ISCB 2014 Vienna, Austria • Abstracts - Oral PresentationsMonday, 25th August 2014 • 14:00-15:30 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust choosing an appropriate clustering method to be recast as statistical model choice problems. Third, it allows for covariates adjustment simultaneously with the fitting process and the size of pattern to depend on a set of concomitant vari- ables. Additionally, FMM is invariant to linear transformation, for example standardization. We discuss these advantages and illustrate the approach with an analysis of the NESCAV (Nutrition, environment and cardiovascular health) dataset (Alkerwi et al, 2010) and show how identified dietary patterns and their associated uncertainty can be used to predict disease.   C11.5 A statistical model of breast cancer tumour growth with estimation of screening sensitivity as a function of mammographic density L Abrahamsson1 , K Humphreys1 1 Karolinska Institutet, Stockholm, Sweden Understanding screening sensitivity and tumour progression is impor- tant for designing and evaluating screening programs for breast cancer. Relevant variables, such as tumour size, are typically only observable at time of diagnosis. How can one estimate tumour growth when the size of each tumour is only measured once? There exists information in differenc- es between tumours found at screening and tumours found symptomati- cally. Stochastic models of cancer development and detection can there- fore be constructed, which yield the distribution of observable variables at diagnosis. Several approaches for estimating tumour growth rates have been described, some of which simultaneously estimate (mammography) screening sensitivity. None of these approaches have incorporated mam- mographic density, although it is known to have a profound influence on mammographic screening sensitivity. We describe a new approach for estimating breast cancer tumour growth which builds on recently described continuous tumour growth models and estimates mammographic screening sensitivity as a function of tu- mour size and mammographic density.   C12 Vaccine studies and infectious diseases C12.1 Taking into account strains heterogeneity in the estimation of vaccine efficacy against seasonal influenza A Benoit1 , P Lebrun2 , W Dewé3 , C Legrand1 1 ISBA, Université Catholique de Louvain, Louvain-la-Neuve, Belgium, 2 Arlenda, Mont-Saint-Guibert, Belgium, 3 GSK Biologicals, Rixensart, Belgium Influenza is an infectious disease caused by several virus strains whose repartition varies between regions and seasons. Typically, a vaccine con- tains 3 or 4 strains and the antigen content is annually reconsidered based on the WHO recommendation. For the same vaccine formulation, phar- maceutical regulations only require efficacy again clinical disease to be shown for a single season, which is performed through a large phase III trial. Subsequent annual modifications of the strain related portion of the vaccine only have to be validated through immunogenicity trials. Classically, influenza vaccine efficacy (VE) trials take place over one season but over several regions assuming common VE. However, depending on the circulating strains characteristics such as their immunogenicity and their matching levels with the vaccine strains, the vaccinal protection level may vary from one season/region to another. We argue that not taking this into account provides incomplete and un- reliable response as for the benefit of the vaccine in the future. We there- fore propose to run phase III VE trials over several regions and seasons in order to characterize the VE heterogeneity. We consider VE as the sum of a common quantity to all clusters (season and region) and of a random clus- ter-specific part. VE is reported based on a tolerance interval, providing insight on the range of future VE across seasons and regions. Our model parameters and the tolerance interval for the cluster-specific VE are esti- mated using Bayesian statistics. Our work will be illustrated by discussing real data examples and simulation results.   C12.2 Estimating the effects of time-since-exposure using case-control data, motivated by a study of vaccine efficacy over time RH Keogh1 , P Mangtani1 , P Nguipdop Djomo1 , L Rodrigues1 1 London School of Hygiene & Tropical Medicine, London, United Kingdom It is sometimes of interest to study whether the effect of an exposure on an event rate depends on time since exposure. Our motivation is an in- vestigation of whether the effect of the BCG vaccination on occurrence of tuberculosis (TB) wears off as time since vaccination increases. In prospective studies the effects of exposure duration can be estimated using a proportional hazards model, allowing exposure effects to differ across time bands or by modelling the exposure effect as a function of time. For a variety of reasons it is of course common to study exposure effects using a case-control study. We discuss the challenges of estimat- ing the effects of exposure duration from population-based case-control studies, which have not typically been used to incorporate information of exposure duration due to their retrospective nature and sampling which disregards time. In the motivating example, cases were historicalTB cases and controls were sampled from the underlying population using frequency matching on birth cohort. Individuals and their parents were interviewed, and medical records examined, to establish whether or not they had received the BCG vaccination, and when. Methods for making efficient use of controls at multiple times-since-vacci- nation are discussed and different methods of analysis considered. These include fitting a series of logistic regression models within time bands based on time-since-vaccination. Another possibility is to view the case- control sample as a form of case-cohort study and to apply a modified proportional hazards analysis. The appropriateness and efficiency of dif- ferent methods are compared using simulation studies. C12.3 Integrative analysis of high-dimensional data in clinical trials: an example in HIV vaccine development R Thiébaut1 , B Hejblum1 , Y Levy2 1 INSERM U897, Bordeaux, France, 2 Vaccine Research Institute, Creteil, France Background: Because there is no definitive surrogate endpoint, many im- munological markers are measured in HIV vaccine trials. Furthermore, the availability of high throughput assays leads to the constitution of high- dimensional data. We present a modelling strategy in 2 steps to analyse all available information in order to identify the gene signature of the ob- served viro-immunological response. Methods: DALIA is a trial evaluating the response to an ex vivo gener- ated DC loaded with HIV-lipopeptides in 19 HIV patients on antiretroviral therapy (ART). Gene expression in whole blood was measured by microar- rays (Illumina HumanHT-12) at 14 time points. Post vaccination immune responses were evaluated using various assays. In step 1, a Time-course Gene Sets Analysis (TcGSA) was performed using hierarchical models

Pages Overview