A clinical trial usually seeks to evaluate the effects of a candidate drug in a carefully pre-specified patient population. Every detail of the trial must be outlined in the Clinical Study Protocol (CSP), including the exact inclusion and exclusion criteria for patients, the exact variables to be measured, and the statistical hypotheses to be tested.
Platform trials and basket trials, however, are innovative study designs that allow researchers to explore multiple treatments or target multiple patient populations simultaneously under a single overarching CSP, called a master protocol. Such approaches are elegant in that sponsors may start a new study arm to investigate an additional indication, dose, or inclusion criterion in parallel with the ongoing clinical trial, without needing to write a new CSP for each new study arm (which would also need to be applied for and approved by authorities). On the other hand, the planning and writing of the CSP for platform and basket trials up front requires a lot more effort than that of a traditional study.
Here, I outline benefits and challenges of these groundbreaking methodologies.
Platform trials make it possible to add study arms
Based on a particular disease, a platform trial investigates different treatments, doses, or subgroups of patients, all in different study arms. In particular, it is possible to add study arms that were not predefined in the study protocol: It is possible to start a trial, keep it running over years, and introduce new potential treatments as they appear and after evaluation of the older treatment arms. Platform trials are adaptive, as new parts of the trial may be chosen based on knowledge gained from previously evaluated parts of the trial. Allocation rates between ongoing treatment arms may also be adapted to optimize patient recruitment.
However, platform trials may come with administrative challenges
Platform trials can be notoriously difficult to administer. The CSP (i.e., the master protocol) needs to consider precise instructions for how future decisions will be made regarding the number of interventions active at the same time, the allocation of new patients between interventions and control groups, the frequency of interim evaluations, and the rules for stopping and starting interventions at interim evaluations.
Yet platform trials are helpful in collaborative projects
Despite the administrative challenges, a platform trial may be very beneficial in, for example, collaborative projects between multiple clinics or academic groups worldwide. Multiple groups of researchers may contribute to the larger project, enabling the comparison of different treatment strategies through the streamlined study arms detailed by the master protocol. Research groups may be able to share control groups and quickly adapt to new or evolving therapeutic landscapes. The STAMPEDE prostate cancer study is an example in which 12,000 patients were enrolled between 2005 and 2023.1 Another example is the I-SPY platform trial, in which 28 active interventions against breast cancer have been tested so far since the start of recruitment in 2010.2
Basket trials allow for multiple indications
Unlike platform designs, basket designs do not permit adding new treatments during the trial. Instead, while the trial targets a specific therapy, it allows sponsors to test multiple indications. Think of each basket coming with a new set of patients, with their own inclusion and exclusion criteria, to a trial. Each basket will be randomized to its own study arms (usually active and control treatment arms), but the outcome of the study may be a combination of the results from all the different study arms. This way, a proof of concept may be approached early and jointly between, say, different cancer indications that may be candidates for the same drug. The assessment of each indication may be derived given the results of the other indications, for example, using a Bayesian method.
Common criticism of basket trial designs
Basket designs do get criticized for enabling a positive study outcome even in situations where no indication shows sufficient efficacy on its own. This is a justified comment. As you go into a follow-up study to recruit a larger number of patients with a single indication, your amount of evidence from a positive basket trial may be very light for the specific indication. This means the follow-up study has a larger element of gambling than it would have had were the first efficacy study based on that same single indication. We cannot be sure that there really is a treatment effect in one particular indication.
When are basket designs useful?
For the reason mentioned, a basket design makes the most sense when there is clinical reason that all the indications can be improved by the same molecular drug mechanism. Perhaps because the indications were all caused by the same mechanism. In such cases, coherent results in different patient populations do strengthen each other. The BRAF V600 Vemurafenib is an example basket trial in which patients had the same mutation (BRAF V600) but different diagnoses.3 It included 122 patients from 5 indications (NSCLC, Colorectal, Cholangiocarcinoma, ECD or LCH, and Thyroid) plus one “other” basket.
Interested in learning more? Download our complimentary ebook, Adaptive Trial Design, which outlines common adaptive trial designs, benefits of adaptive trials, how to optimize your adaptive trial, and a ten-point framework to determine if your trial should be adaptive.
