Speaker

Biography

Abstract

Simon Bate

Following a PhD in experimental design at Royal Holloway, University of London, Simon started his pharmaceutical career in the Discovery Statistics group at GlaxoSmithKline. This role involved collaborating with life-scientists, running a 3-day training course on statistics and conducting statistical research.

He has also worked at Huntingdon Life Sciences, supporting toxicologists and pharmacologists on the design and analysis of data from regulatory safety assessment and toxicology studies. During this time he also collaborated with a programmer to develop InVivoStat, a free-to-use statistical software package for life-scientists and other researchers. InVivoStat jointly won the 2018 RSS and PSI prize for Statistical Excellence in the Pharmaceutical Industry.

Recently he has spent the last 12 years at GSK supporting pre-clinical research and medicine.

He has been a committee member of the PSI Toxicology SIG and the NC3Rs (National Centre for the 3Rs) experimental design project group and has given external training courses to professional and academic organizations, including the Karolinska Institute, Paris-Sud University, UCL, NC3Rs, European Summer School in Whole Animal Pharmacology and the British Association for Psychopharmacology. His is co-author of the textbook “The Design and Statistical Analysis of Animal Experiments”.

Experimental design plays a fundamental role in improving both the reliability and reproducibility of all experimentation. In pre-clinical research it can be argued it plays an even more important role due to limited resources, the need to apply the 3Rs principals and the underlying complexity of the biological subject matter.

In what is planned to be an interactive session, Simon Bate will present his thoughts and experiences on over 20 years working with life-scientists and will discuss the strategies he has developed over that time. He will offer his ideas on how we can work alongside life scientists, guiding them through the intricacies of designing experiments. By putting experimental design at the front and centre of the research process, he will argue that not only will we get more reliable results, while reducing animal use, but we can avoid some of the common pitfalls but navigate contentious issues surrounding research in this area such as the role of p-values, pseudo-replication and multiple comparison tests.

The session will also include a session on how to collaborate and communicate with scientists, where he will explore several tools available that can aid the scientist in their understanding and application of experimental design in practice.

David Lovell

David Lovell is Emeritus Reader in Medical Statistics at St George’s Medical School, University of London. Previously he was Reader in Medical Statistics at the Postgraduate Medical School, University of Surrey and an Associate Director and Head of Biostatistics support to Clinical Pharmacogenomics at Pfizer Global Research and Development (PGRD) in Sandwich, Kent providing data management and statistical support to pharmacogenetics and genomics. His PhD was from the Department of Human Genetics and Biometry at University College London in 1980. Before joining Pfizer, David was the Head of the Science Division at BIBRA International, Carshalton, which included Molecular Biology, Genetic Toxicology, Biostatistics and Computer Services. At BIBRA David managed the statistical and computing group providing specialised statistical support to BIBRA’s Clinical Unit and contract research work. He conducted and managed research programmes on genetics, statistics and quantitative risk assessment for the EU and UK Government Departments. His research interests at BIBRA were in the use of mathematical and statistical methods together with genetic models in the understanding of toxicological mechanisms and risk assessment problems. David had previously been a Senior Research Officer with the MRC Experimental Embryology and Teratology Unit, a visiting Postdoctoral Fellow at the NIEHS in North Carolina, USA, a Geneticist at the MRC Laboratories, Carshalton and a Research Assistant in Cytogenetics at Birmingham University. He was Vice Chair of the Scientific Committee of EFSA (the European Food Safety Authority) from 2009-12 and a member of the Independent Scientific Advisory Committee (ISAC) for MHRA database research from 2006-12. He was Chair of the UK Government’s Advisory Committee on Mutagenicity of Chemicals in Food, Consumer Products and the Environment (COM) and a member of the Committees on Carcinogenicity (COC) (until April 2021) and is currently a member of the Committee on Toxicity (COT). He was a member of the Board of the NC3R's (2017 - 2022) and a member of its Grant Assessment Panel (2013-2017). In 2019, he receved from UKEMS the Jim Parry Award to a senior scientist who has made substantial contributions to the field of environmental mutagenesis.

Genotoxicity testing is an important component in the pre-clinical testing of drugs and other chemicals. It is 40 years since the first OECD guideline on bacterial mutagenicity, the Ames test. The International Workshops on Genotoxicity Testing (IWGT) were set up at about the same time to provide a forum for industry, academic and government experts to meet and review regulatory and other topics related to genotoxicity. The 8th Workshop in August 2022 in Ottawa, Canada covered a range of topics where statistical issues arose including statistical approaches and data interpretation, predictivity of in vitro genotoxicity testing and analysis of genotoxicity dose-responses. The relationship between statistical significance and biological importance often arose during the group discussions. This presentation will outline some of the conclusions reached and indicate some of the challenges facing genotoxicity testing as new methodologies become available and how statistical input can help.

Odile Coudert

Odile Coudert Berthion holds a MSc in biostatistics from ISUP (Statistics Institute of Paris Universities) that she validated by a research master thesis within the department of statistics of the University of Glasgow. She has been working for more than 13 years for the Non Clinical Efficacy &Safety (NCES) department of Sanofi as a contractor, first at Keyrus Life Science then at IT&M stats. She had the opportunity to analyze a wide range of studies from various therapeutic areas, she also contributed to improve the statistical methods applied in NCES. She has given trainings on mixed models. In November 2023, she has joined Ipsen to provide statistical leadership and support to the Research, External Innovation and Early Development (REED).

In preclinical and research field, departures from the hypotheses underlying parametric analyses, normal distribution and variances homogeneity, are frequently encountered; this is all the more concerning since the sample size is mall and the data may be repeated.

Anova-Type Statistics (ATS) is a non-parametric method used for analyzing longitudinal data from factorial designs without making any assumption on the data distribution and the variances homogeneity. Particularly adapted to small sample sizes, it remains valid even when there is no variability within some factors levels. The ATS method is based on data distributions instead of positional parameters. The distributions and the relative factors effects are estimated using the ranks amongst all observations. This method can easily be implemented using the Mixed procedure of SAS software.

Bernard Francq

Bernard G Francq is Lead Statistician with GSK Biologicals, driving statistical innovation for CMC projects worldwide. He holds a PhD in Statistics (UCLouvain, 2013).

The world's foremost expert on errors-in-variables (EIV) regressions in method comparison studies, his work has been awarded Best MSc Thesis Biostatistics (Quételet 2008, Belgium), Best Chemometrician Prize (Chimiométrie 2009, Paris), and Best Young Researcher (Agrostat 2012, Paris). His communication skills have been recognized with the Greenfield Challenge Award (ENBIS 2012, Ljubljana), and Best Communication Award (ENBIS 2012, Ljubljana). Recent work on tolerance intervals in bridging studies was awarded Best GSK Statistical Paper (2020).

Bernard's research interests include DOE's, CMC statistics, clinical trials, non-linear mixed models, tolerance intervals, and advanced data visualization techniques. He lectures at UCLouvain and regularly offers trainings to statisticians in the (bio)pharmaceutical industry.

In the past, screening (which process parameters are impactful) and optimisation were 2 distinct phases performed by 2 different designs of experiments (DoE). Then, the definitive screening designs (DSDs) published approximately 10 years ago attracted a lot of attention from both statisticians and non-statisticians, espcially in the pharma industry. The idea is to combine screening and optimisation in a single DoE. This allows to reduce the total number of experiments and the research development time with a substantial gain in the budget. Furthermore, it accelerates the time-to-market, at least by 2, of future drugs and vacccines.

Recently, a new type of DoE called OMARS for orthogonal minimally aliased response surface has been published. These OMARS DoEs outperform DSDs in many criteria. Firstly, the orthogonality criteria where the independence between main effects is fulfilled, and also between main effects and interaction (+ quadratic) terms (also in presence of categorical factors). Secondly, the projection property where OMARS designs are more likely able to estimate a response surface model from the selected significant parameters.

In this presentation, we will assess the impact and discuss the deployment of OMARS DoEs in the pharma industry. The first case study will illustrate fermentations on Ambr system for vaccine development (with continous process parameters). The second case study will involve categorical factor for analytical development. We will then discuss the opportunities and deployment of OMARS DoEs in non-clinical and pre-clinical phases.