After a two-and-a-half-hour flight which left Brisbane on a sunny Saturday morning mid-October, I was smiling ear to ear being able to roam freely around the streets of the vibrant city of Melbourne and looking forward to attending the Australian Trials Methodology (AusTriM) conference with my colleagues from the University of Queensland, including our team leader Professor Dan Siskind, my post-doc fellow Dr Rebecca Soole, and our amazing clinical trial manager, Andrea Baker. After having spent over 25 hours on the plane alone with a 3- and a 6-year-old just a week prior, returning to Australia from my home country Slovenia, this short trip was a dream flight and I almost felt lost at the Melbourne airport, without needing to keep check if both kids are next to me. The adventurous and nerdy child in me was no doubt doing flips out of excitement as I booked in my hotel with views over the Yarra River and grey Melbourne skies.
Our team had a presentation on Day 2 of the conference. This was on the Adaptive Platform Trial of semaglutide versus metformin versus combination (semagltuide/metformin) versus placebo among people living with schizophrenia. But before I get to the details of our SWiMS trial (Schizophrenia, Weight, Metformin and Semaglutide), the term Adaptive Platform Trial (APT) deserves some explanation.
Platform trial design presents a transformative methodologic approach that involves studying one or more treatments in one or more diseases or disease subgroups using the same overarching trial design. In other words, APT streamlines evidence generation by addressing multiple questions simultaneously and efficiently, typically with a common control group. By using one standardised control arm, against which multiple treatment arms can be compared, platform trials offer a highly efficient approach by saving on patient resources, plus recruited participants having a greater chance of being allocated to more promising novel and more effective treatments. This optimised use of control data reduces the overall sample size required, minimising recruitment time and resource expenditure compared to traditional individual randomised control trials (RCTs). Platform trials use sophisticated modelling and Response Adaptive Randomisation (RAR) techniques to evaluate multiple treatments across different areas of patient management. Moreover, by integrating multiple interventions simultaneously, the platform trial enables direct comparisons and enhances the comparability of observed treatment effects.
Sounds rather abstract, right? Wait until I delve into the statistical requirements to run an adaptive platform trial. To make a trial adaptive, which essentially means, that at any interim analysis throughout the trial, you can drop poorly performing treatment arm(s) and/or change the randomisation ratio to favour better performing treatment, you need a good statistician. Without a statistician who understands the APT design and the RAR technique, you will be left like a fish without water in the APT ocean. This is because the platform trials employ a Bayesian statistical framework, a powerful form of analysis that incorporates prior knowledge together with collected data to produce a combined estimate. This prior knowledge is gathered from an understanding of previous clinical trials, possible endpoints and an understanding about disease progression, to form a comprehensive and accurate assessment of treatment efficacy, superior to traditional analytical approaches. As such, extensive statistical simulations are conducted to determine a range of decision rules including early stopping of treatment arms if meeting pre-defined criteria for superiority or inferiority, leading to greater trial efficiencies and a minimum statistical error. Simulations leverage prior knowledge of participant characteristics and hypothesised treatment trajectories to yield highly accurate predictions of future outcomes. These measures not only enhance efficiency but also have ethical benefits by minimising patient assignment to ineffective treatment arms and enhancing equitable access to better clinical care.
Such innovative trial designs have to date being applied in areas of breast cancer treatment (I-SPY2), severe community-acquired pneumonia (REMAP-CAP), and management of glioblastoma (GBM-AGILE), and are also being rapidly developed for other diseases. I strongly encourage interested readers to satisfy your curiosity and appetite for more information about this transformative clinical trial design by publications on this topic by Professor Steve Webb, the “godfather” of REMAP-CAP, the biggest global platform trial which generated, and is still generating at fast speed, mountains of empirical evidence on potential treatments to prevent mortality from COVID-19, Berry Statisticians, and other clinical academics who are working with APTs. Nevertheless, I will below attempt to provide a summary of the workshop on Platform Trials preceding the AusTriM Conference, led by esteemed biostatisticians, Professor Thomas Jaki and Professor James Wason, and the dense scientific program of the conference program.
The AusTriM workshop and presentations delved deep into master protocols, a class of efficient clinical trial designs aiming to answer multiple research questions within a single trial protocol, as well as the different platform trials that are currently being developed or trialled across Australia. The conference also provided valuable knowledge about the intricate framework required for designing multi-arm multi-stage platform trials which are capable of adapting based on accumulating data, a methodology integral to the adaptive platform trial presented, information borrowing methods for basket and umbrella trials; methodologies for sharing information across different arms of the trial, optimizing resources, and enhancing overall efficiency, a concept seamlessly aligning with the APT’s goals, software tools available for implementing master protocols which can empower researchers to navigate the complexities of these trial designs and support them in the practical application of these theoretical concepts, and last, but not least, the different APTs designs. These designs include Group-sequential design where arms can be stopped at different stages for efficacy and/or futility, Drop-the-losers design where you take forward a (pre-specified) fixed number of experimental treatments at each stage based on interventions performance, and the Adaptive randomisation design which is useful for trial with low number of likely effective treatments.
This low number of likely effective treatments brings us back to our SWiMS trial, which I presented on Day 2 of the conference (while Dan Siskind, who, as a lead Chief Investigator on this project, would be more deserving to deliver the presentation, was in his clinic seeing clients). This MRFF-funded trial, which was halted due to the global shortcake of the famous tik-tok drug, semaglutide (branded under the name Ozempic) in 2022, was designed to address the significant cardiometabolic burden faced by individuals experiencing schizophrenia, which is largely driven by the metabolic side effects of antipsychotic medications. This delay in the commencement of SWiMS trial made us pivot our efforts and writing of the clinical trial protocol into writing a Master protocol which would support a schizophrenia-focused platform trial. As such, the arm included in SWiMS (semaglutide, metformin, combination semagltuide/metformin, and placebo) would represent the first comparison embedded within the schizophrenia adaptive platform trial. While we are currently looking to obtain the funding to fund the setup of the platform, it is anticipated that by making SWiMS an adaptive trial, we will reduce the total number of participants needed to reliably assess the efficacy of the intervention arms for weight reduction in individuals with schizophrenia who are also grappling with antipsychotic-induced obesity. This marks a pioneering effort within the field, as it represents an opportunity for all future trial with people living with schizophrenia to generate evidence more efficiently on potential effective treatments and the effective treatments to be translated into clinical care more rapidly.
Overall, the AusTrim Conference 2023 transcended the traditional RCT research world, giving us a glimpse of in a new era of clinical trial design, and importantly, of new possibilities for patients of health services and specifically for schizophrenia research and treatment. No doubt, the idea of running a multi-site adaptive platform trial is an ambitious one. However – following my moto in life, where there is a will, there’s a way - with combined efforts of different stakeholders, including statisticians, psychiatrists, trialists, clinical trial managers, and last, but least important of consumers, these novel designs bring a promise to redefine how we approach and identify the optimal set of treatments to address the physical and mental health challenges of schizophrenia, offering hope and progress to those impacted by schizophrenia. Returning from a refreshing 10 degrees Celsius, with a head slightly overloaded from all the statistical terms and complexities of platform trial designs, to scorching 33 degrees in Brisbane, there was – unfortunately or fortunately - no time to dwell on thinking how are we going to create such a complex structure for trials in schizophrenia as we had a final, end-of-the-year Ethics Committee deadline to catch so we can bring SWiMS as an adaptive platform trial to life in 2024.
Although diving into the APTs world and their underlying statistical processes no doubt comes with a steep learning curve, this innovative trial design undoubtedly provides an opportunity: it fosters a collaboration among clinicians, researchers, and other stakeholders involved in the care of individuals with schizophrenia to streamline and standardises clinical trial processes so that evidence on the efficacy of promising treatments to address pressing issues related to the physical and mental health of this population is produced in a more effective and comprehensive manner, while facilitating the translation to clinical practice more rapidly.
Author
Urska Arnautovska
Senior Research Fellow, Faculty of Medicine, The University of Queensland
Senior Mental Health Clinician, Metro South Addiction and Mental Health Services, Queensland Health
