.tmb-propic-md.jpg?Culture=en&sfvrsn=c25d2b2f_9 "Professor Louis Chesler (Profile pic)")
Researchers have developed a new statistical approach that could help improve how early-stage cancer trials identify the safest and most effective doses of advanced therapies, such as CAR-T therapy.
The method, called DOSET (Dose Optimization with Simultaneous Efficacy and Toxicity), was developed through a collaboration between King’s College Hospital, King’s College London and The Institute of Cancer Research, London. It offers a more efficient and flexible way to design early-phase clinical trials, particularly in settings where only small numbers of patients can be recruited.
The findings, published in the journal ESMO Open, mark the first publication of a novel statistical trial design developed by ICR Clinical Trials and Statistics Unit (ICR-CTSU) Early Phase and Adaptive Trials Group for one of its clinical trials. This work was supported by grants from Cancer Research UK.
Rethinking how early-phase trials work
Early-phase clinical trials are designed to identify safe doses of new treatments. Traditionally, these studies have focused on finding the maximum tolerated dose, based on the assumption that higher doses are more effective.
But this assumption does not always hold true for newer and advanced therapies, such as CAR-T cell therapy, for which dose-response relationships can be more complex. CAR-T therapy is a highly personalised form of immunotherapy used to treat blood cancers, such as leukaemia and lymphoma, that have not responded to treatments or have relapsed.
In CAR-T trials, additional challenges arise because studies typically enrol relatively small numbers of patients. This is due to limited eligible populations, along with factors such as the complexity of the treatment and the need for personalised manufacturing.
These constraints make it especially important to make the best possible use of data as they accumulate during the trial.
A more adaptive approach
DOSET is a novel dose-optimisation design developed specifically for a first-in-human CAR-T therapy trial of a next generation CAR-T cell product (pCAR19) at King’s College Hospital. It combines several advanced statistical methods into a single adaptive framework, allowing researchers to assess both toxicity and preliminary efficacy at the same time, rather than focusing on safety alone.
As a result, the approach can identify doses that strike a better balance between safety and effectiveness, while reducing the number of patients exposed to doses that are either too low to work or too toxic.
The design also allows trials to stop early if a treatment looks unlikely to be effective or appears unsafe, helping to protect patients and improve efficiency.
Better performance in small trials
To evaluate the method, the researchers compared DOSET with an existing dose-optimisation approach used in early-phase CAR-T trials.
Across a range of realistic clinical scenarios, DOSET demonstrated consistently stronger performances. It showed a higher likelihood of correctly identifying doses that were both safe and effective, while maintaining similar levels of patient safety. It was also better at advising researchers when to stop trials early when no suitable dose could be identified, reducing unnecessary exposure to ineffective treatments.
Importantly, these advantages were seen even in small-sample settings, which are common in CAR-T studies.
First author Dr Xinjie Hu, who was a senior trial statistician in the Early Phase and Adaptive Trials Group in ICR-CTSU, said: “Many modern therapies, such as immunotherapies and targeted therapies, do not follow traditional assumptions about how dose relates to effectiveness. In these settings, more flexible trial designs like DOSET could help improve how doses are selected, ultimately supporting better decision-making in early-stage drug development.”
Future implications
The research team hopes the work will contribute to a broader shift towards evidence-based dose optimisation in early-phase clinical trials, beyond CAR-T, increasing the chances that new treatments are identified more quickly.
Senior author Professor Christina Yap, Professor of Clinical Trials Biostatistics and Group Leader of the ICR-CTSU Early Phase and Adaptive Trials Group, said: “Developing DOSET in close collaboration with clinical teams at King’s College Hospital and King’s College London allowed us to address a real challenge in early-phase trials – how to make the best possible decisions with limited patient data.
“It’s particularly rewarding to see a methodological innovation progress from concept through to implementation in a real clinical trial and publication. We hope this approach will support more informed, patient-centred decision-making and contribute to improving outcomes in early-phase cancer research.”
Chief Investigator Dr Reuben Benjamin, Haematologist and Clinical Senior Lecturer at King’s College London, said: “For advanced therapies such as CAR-T, finding the right dose is not always as straightforward as administering higher doses and expecting a better outcome. Our aim is to identify the precise doses needed to maximise benefit for patients while minimising unnecessary risks. DOSET gives us a more sophisticated way to analyse clinical trial data, and we hope this approach will not only benefit this CAR-T trial but also help improve the development of future cell and gene therapies for patients with blood cancer.”
Banner image: Clinical trials pharmacy (Jan Chlebik for the ICR, 2014)
