Date

Venue

Topic

Led by

Materials

Webinar

Vaccines, between hope and doubt: statistical insights

Abstract:

The lack of information regarding the existing COVID-19 vaccines tends to reinforce defiance against vaccines and raises concerns amongst the general public. When compared to the long years spent in general to make a vaccine before its licensing, the extreme rapidity by which the COVID-19 vaccines have become available raises questions. At what price have the vaccine trials been accelerated? In this talk, statistical insights on vaccines and vaccine safety will be presented. Examples of studies on previous vaccines, their safety and results will be discussed in the light of the current COVID-19 vaccines. Vaccine safety evaluation will be demonstrated using the well- known self-controlled case series method (reference method). Vaccine safety for all licenced vaccines is proven when vaccine benefit to risk ratio is extremely high. Safety checks continue even after vaccine licensing with a view to managing a potential crisis and to identifying previously unreported adverse reactions. This is the role of the area of pharmacovigilance. The global characteristics of the COVID-19 pandemic favoured a transparent and reactive management of the vaccines’ safety during the trials and would allow a quicker response to the safety-related questions that the vaccine trials did not clarify.

Trials, tribulations, machine learning and serendipitous discovery in cancer genomics

Abstract:

The REMoDL-B trial used a machine learning approach with transcriptomic data to assign patient tumours to different cell-of-origin groups, with the biologically motivated hypothesis that that these groups would respond differently to the proteosome inhibitor Bortezomib. This was the first trial in the aggressive lymphoma field to use real time transcriptomic or genomic data generation in patient randomization. Ultimately it did not demonstrate any advantage of the new treatment, but I will tell the story of the trial, the many other things we learned from it. Discoveries are been taken forward in new similar trials and raise issues of efficient trial design and how trials might be better designed in the future.

A comparison of dual biomarker threshold identification procedures within a confirmatory clinical trial

Abstract:

There is increasing evidence to suggest that multiple biomarkers are needed to sufficiently identify sensitive patients for some drugs or drug combinations. In this work, a variety of dual biomarker threshold identification procedures are applied in a phase III trial setting and their performance contrasted. It is of interest to identify thresholds for two continuous biomarkers simultaneously, which dichotomise the respective biomarkers into sensitive and non-sensitive patients, thus defining a two-dimensional patient subgroup i.e. patients who are defined as sensitive for both biomarkers. Four methods were implemented within Freidlin and Simon’s Adaptive Signature Design (ASD) framework, these being: a grid search, a modelling-based method, recursive partitioning and prognostic peeling. Methods were contrasted by their ability to accurately identify biomarker threshold locations and by the proportion of trials which achieved significant efficacy, both overall and subgroup specific. This work was carried out using a simulation study. In the simulation study, recursive partitioning methods showed the best overall performance, with respect to both the threshold identification accuracy and trial operating characteristics. All methods suffered when the expected proportion of sensitive patients was low and when the magnitude of treatment effect was modest. In conclusion, dual biomarker threshold identification can be successfully incorporated into a confirmatory phase III setting, without jeopardising the ability to detect an overall treatment effect. In such low dimensional settings, recursive partitioning methods should be taken into consideration.

Brief update on COVID – A Gastroenterologists perspective

Go/No Go Framework: Our Criteria at AstraZeneca in Early Phase Decision Making

Abstract:

Throughout development, AstraZeneca uses a standard approach to set study decision criteria. This is key in Early Clinical Development where it is essential to make robust decisions quickly to enable compounds to progress or to be stopped. The approach uses a three outcome design and is based on the confidence of observing a result better than a Lower Reference Value (LRV) for a Go decision or a result worse than a Target Value (TV) for a No Go decision. If neither of these criteria are met, the outcome is in a “consider” zone, where teams need to use other information to move to a clear decision. When setting the criteria teams objectively set the LRV and TV based on the available evidence and agree the required confidence levels. They also assess the operating characteristics of the decision framework such that the probability of being in the “consider” zone is acceptable. Using this approach has enabled decision criteria in Early Clinical Development to be based on the same framework with the use of standard displays in agreeing these criteria with senior stakeholders.

Quantitative decision-making in the context of early-phase biomarker-adaptive designs with survival endpoints.

Abstract:

In recent years, a number of decision-making frameworks have been proposed as an alternative to the Null Hypothesis Significance Testing (NHST) approach. The three-outcome framework from Frewer et al (2016), based on Lalonde et al (2007) utilises two pre-set values, namely the Target Value and Lower Reference Value, by which to judge the estimated outcome and categorise it as either a go, consider or stop decision.

The work presented here generalises the three-outcome framework by applying it to a novel biomarker-adaptive design with survival outcomes. The biomarker-adaptive design stratifies the population into two subgroups with an option to enrich one of the subgroups at the second stage should the data at the interim support that decision. We evaluate the operating characteristics of such a design by simulation of trial scenarios. The simulations are used to look at several metrics to access performance as the typical error rates and power do not carry over to the framework being used. The biomarker-integrated three-outcome framework presented here has the benefit of using a novel design leading to actionable decision-making for early-phase oncology trials with survival endpoints.

Schedule:

Each of the 3 speakers will have 25 minutes (15 mins Presentation +10 mins Discussion). 

14:00 – 14:05 Welcome and Introduction

14:05 – 14:30 James Griffin Presentation

14:30 – 14:55 Stella Zhan Presentation

14:55 – 15:20 Nigel Stallard Presentation

Titles and Abstracts: 

James Griffin

Title: The RACER studies: Challenges of conducting RCTs investigating robotic interventions

Abstract: This talk will introduce two new NIHR-funded RCTs being conducted by Warwick CTU; RACER-Knee and RACER-Hip. Both studies are investigating whether patients undergoing joint replacement surgery do better using the standard joint replacement or whether using a robotic arm to plan and conduct the procedure improves patient outcomes. I will then briefly discuss the ongoing challenges in designing and conducting the studies in the current climate and present some of the solutions we hope to implement.

Stella Zhan

Title: Comparison of frequentist and Bayesian methods for two-arm borrowing of historical data

Abstract: Incorporating historical data into a current trial is a potential strategy to reduce development costs and patient burden in clinical trials. This so-called extrapolation or borrowing is particularly valuable in rare disease and paediatric trials, when patient recruitment is difficult. The main issue associated with this strategy is the potential for inflation of the type I error rate. Therefore, it is important to choose the right extrapolation method, ensuring that the amount of strength borrowed from the historical study is appropriate and is adjusted to the agreement between the two trials with the aim of increasing the power of the current trial whilst at the same time controlling the type I error rate.

Various frequentist and Bayesian approaches exist for borrowing historical control-arm data. However, there is relatively little research on borrowing information from both the control and treatment arms of a single historical two-arm trial. In this work, we extend static and dynamic borrowing methods proposed for the control-arm borrowing to the setting of two-arm borrowing. These methods are then evaluated in simulation studies investigating a two-arm trial with a binary outcome to find appropriate borrowing parameters whilst optimising the trade-off between type I error and power.

Our simulation studies show that the degree of type I error inflation is mainly affected by the historical rate difference. Dynamic borrowing approaches are shown to offer better control of the type I error inflation over a wide range of scenarios, with the choice of borrowing parameters playing an important role.

Nigel Stallard

Title: Using short term endpoint data in interrupted clinical trials

Abstract: While the SARS-CoV2 pandemic has led to the rapid initiation of an impressive number of clinical trials, it has also had a major impact on the large number of ongoing clinical trials in other disease areas. Interruption in the normal delivery of healthcare and the inability to invite patients for routine follow-up visits has led to many trials being stopped or temporarily paused.

The impact of the pandemic will vary from trial to trial, but some trials in chronic conditions will have a large number of patients already in the follow-up phase with some early outcome data for whom long term primary endpoint data are unavailable. This talk will consider how these early outcome data can be used to improve the estimation of treatment effects on the primary endpoint.