The Cytel Blog

Highlights from the PSI 2017 Conference

Written by Cytel | Jun 1, 2017 12:06:00 PM

A number of the Cytel team were in London, 14th – 17th May 2017 for the PSI Annual conference. The PSI Conference is an important forum for statisticians and we found this year's event insightful, informative and well attended. In this blog, we'll summarize some of the particular highlights from the sessions that our team members attended.

Keynote Session Day 1: Richard Stephens - 40 years of PSI - From Patients Staying Ignorant, Through Patients Showing Interest, to Patients Starting Involvement 

The session on day one provided a thought-provoking perspective from the patient’s point of view. Richard Stephens set the scene by talking about his journey through healthcare over the last 40 years, and his experience of being diagnosed with two cancers, undergoing a heart emergency, and suffering various other health challenges. He discussed the state of cancer treatment today, cancer research, participation in clinical trials, and participation with advisory groups. Personally, Richard is involved with many cancer patient advocacy groups including: NCRI’s Consumer Forum, NIHR, NCRAS, MRC CTU, NHS England and Genomics England, Cancer Research UK and Eupati. The patient voice is increasing in importance in clinical development today, and Richard's talk was extremely well-received by all Cytel conference participants.

Keynote Session Day 2: David Spiegelhalter - Engaging without manipulating: the balanced communication of statistics

David Spiegelhalter is Winton Professor for the Public Understanding of Risk at the University of Cambridge and well-known as an engaging speaker. He discussed with the delegates the inherent challenge faced by statisticians as they aim to engage an audience’s interest in the data they are presenting, without misleading. He explained that numbers are often deliberately manipulated – whether to alarm or reassure us, and used as rhetorical devices to influence our emotions. Of course, anyone who consumes mainstream media is regularly on the receiving end of this approach. One of the examples Spiegelhalter discussed was an example of how a newspaper misrepresented the dangers of eating bacon  by presenting its associated relative risk of bowel cancer rather than the absolute risk. As statisticians in biopharma, we liaise daily with a wide range of stakeholders who rely on us to communicate data responsibly. It is critical that we carefully walk this line between engagement and accuracy.

Biosimilar Development Session:  Chair - Byron Jones (Novartis)

For Pantelis Vlachos, Director of Strategic Consulting at Cytel the Biosimilar Development session was a particular highlight as it is an area he has personally worked in and presented on extensively. The session incorporated three interconnected presentations. The first, from Johanna Mielke, of Novartis discussed a methodology of how to deal with transition data in a Phase 3 bioequivalence trial. In biosimilars development, where the objective is to effectively copy the originator or reference product, the clinical development   program typically incorporates 2 trial phases - Phase 1 (pharmacokinetic) and Phase 3 (efficacy). Generally, the Phase 3 biosimilar trial involves a simple head to head comparison design. However, some regulators may ask for additional data on switching from the sponsor, to see whether patients can interchangeably switch from reference to originator or vice-versa. Typically this is done in a descriptive manner data. Mielke presented a way to perform this activity statistically using a mixed model. The second presentation by Heike Woehling shared a case study of how Sandoz has applied this model in one of their Phase 3 trials.

The final presentation in the stream from Koenig provided a review of those clinical trials which have led up to biosimilar drug approvals. The review examines a number of aspects including the number of trials in each stage of development and the type of trial design used. This provides useful insights for sponsors involved in biosimilar development, and the associated systematic review article is available open access in the [1] British Journal of Clinical Pharmacology.

Bayesian Methodology Session: Chair - Rachel Moate (Medimmune)

This session was a highlight for Andrea Hita, a Data Scientist from our Barcelona office. GlaxoSmithKline presented a tool that they have developed for prior functions elicitation when Bayesian approaches are needed for different clinical trial projects. Clinical experts have a wealth of valuable knowledge available, but there is a challenge in converting that knowledge into specific prior functions. The interactive tool developed by GlaxoSmithKline helps to bridge that gap, by both educating the clinical experts on some of the main statistical concepts that are behind these functions, and helping them to translate their insights about a specific project, into a mathematical function that can later be used to provide input into the project. This tool can be extremely helpful when planning a clinical trial, and provide a platform for interaction between the statistician and the clinician.

Other important themes of the conference included Estimands, the latest insights on regulatory hot topics, and innovative methodologies. Cytel's Madhusmita Panda also presented in the Innovative Methodology session on the topic of Measuring Intergroup Agreement and Disagreement.

In an upcoming blog, we will share a summary of her interesting talk, and her slides. To get an update from the Cytel blog  direct to your inbox click below.


1) Mielke, J., Jilma, B., Koenig, F. and Jones, B. (2016). Clinical trials for authorized biosimilars in the European Union: a systematic review. British Journal of Clinical Pharmacology, 82(6), pp.1444-1457.