Approach for reporting master protocol study designs on ClinicalTrials.gov: qualitative analysis

Key characteristics of master protocol research programs (MPRPs)

Master protocols are being used more frequently across the clinical research enterprise, including in cancer research and in response to the covid-19 pandemic.5678 These types of studies encompass various designs with different names based on specific design features9; however, all of these studies can generally be described as “one overarching protocol designed to answer multiple questions.”10 A master protocol is conducted with “a collection of trials or substudies that share key design components and operational aspects to achieve better coordination than can be achieved in single trials designed and conducted independently.”10 Thus, we use the term MPRP to highlight the use of an overarching research protocol for coordinating an evolving set of studies that constitute a program of research. We use the term substudy to refer to each individual trial within the MPRP to which participants can be assigned. MPRPs have been used to investigate many interventions for a specific disease with an umbrella or platform design, as well as various interventions that target a specific genetic subtype of tumor independent of cancer type with a basket design (box 1).111213 MPRPs can use bayesian or other adaptive techniques for assigning participants after screening to substudies or for determining when a substudy should end.

Generally, MPRPs can allow for greater efficiency than the traditional clinical trial model by coordinating within one shared centralized protocol and infrastructure for the investigation of multiple conditions, experimental interventions, or subgroups across multiple individual clinical trials (sometimes referred to as arms).81415 For example, the RECOVERY MPRP (NCT04381936) efficiently identified eligible participants through a centralized process and assigned participants to substudies evaluating, over time, specific interventions for covid-19, as evidence and knowledge evolved. However, the multistudy nature of MPRPs challenges established reporting practices for trial registration and disclosure of summary results that are aimed at improving clinical research transparency and accountability as well as mitigating biases within the medical evidence base.116 Specifically, MPRPs, which can use a single protocol with multiple substudies, disrupt the current model of one protocol describing one trial in one trial registry record. The practice of referring to an entire MPRP as a single trial (eg, by acronym) might imply that a master protocol should be represented as one study record in a trial registry. Yet, many master protocols include plans for a series of substudies that might start, end, and be analyzed independently of each other (including designs with a shared control arm that is fully or partially used in analyses across substudies). Condensing this information into one study record can obscure details of the substudies, such as the end dates and primary outcome measures, thereby undermining key transparency and accountability goals of the trial reporting system. In particular, international trial registries have uniformly adopted the World Health Organization Trial Registration Data Set, a standardized collection of structured data elements describing key study design components.17 Such structured information allows users to search for and retrieve registered trial records based on specific study characteristics, for needs ranging from patients seeking to enroll in trials to systematic reviewers identifying trials for a specified population, intervention, comparison, or outcome. However, each registry record contains only one set of structured data elements for describing the design characteristics of one trial. Designs for multiple trials or substudies can be reported in a free text field of a single record (eg, different start and end dates in the arms and interventions fields). Yet, such information will not be consistently described in the same place by study sponsors across trials and will not be readily identified by search tools that depend on structured information for high-precision data retrieval.

Overview of trial reporting system and challenges for reporting MPRPs

The trial reporting system was designed to track trials at each stage from registration at study initiation to results reporting after completion. Specifically, registration of key protocol information18 at trial initiation serves as a public record of the trial and provides public access to key information about its research plan, which was prespecified in the protocol. Timely updates to registered information throughout trial conduct help to keep the public informed. Public reporting of information on the summary results in a structured format after trial completion allows for documentation of basic scientific findings in a standardized format.19 This display thereby mitigates publication bias and selective outcome reporting, complements information on the results available from the biomedical literature, and facilitates evidence synthesis via systematic reviewers and other trial landscape analyses. The trial reporting system depends on timely and complete submission of information by study sponsors and investigators and supports important ethical and scientific goals.

On trial initiation, registration on ClinicalTrials.gov involves extracting key information from the protocol for a single study and entering it into a set of structured data elements that support the display, search, and download of trial records.18 This model relies on each clinical trial having defined the study start and end dates based on key prespecified design and analytical features that uniquely identify and define the unit of an individual clinical trial. These features include the following:

• Population: a defined group of participants, as described in the conditions of interest and detailed in the eligibility criteria and target enrollment;

• Interventions: study arms and interventions to which participants are assigned; and

• Comparison and outcomes: primary and secondary outcome measures, including the specification of the arms and interventions being compared within the study to evaluate the effect of an intervention on a health outcome over a defined time frame.

Generally, these key features apply to each substudy in an MPRP but do not apply to the MPRP as a whole; that is, each substudy can have different protocol details. In a basket MPRP, such as NCI-MATCH, each substudy evaluates tumors with a specific genetic change (that is, a unique population) and an intervention intended to target that change (that is, a unique intervention). Unlike a traditional clinical trial, in which all the arms and interventions are prespecified in the protocol from the start, one or more substudies in an MPRP can be prespecified at the outset while other substudies are conceived of and added later. However, MPRPs maintain prespecified plans for the analysis of substudy results, whether pooled across substudies or, more commonly, analyzed independently. In this way, an MPRP can be distinguished from a multiarm study in which results from each arm are compared with other arms in the study.

After trial completion, information on the summary results is reported on ClinicalTrials.gov in a tabular format, by study arm, in four scientific modules (participant flow, baseline characteristics, outcome measures and statistical analyses, and adverse event information); the full study protocol and statistical analysis plan are also provided.19

Patient and public involvement

The approach described in this article for reporting MPRPs to trial registries and results databases such as ClinicalTrials.gov evolved over several years. Our proposal is based on the authors’ direct experience with MPRPs, including reviewing submissions of registration and summary results and working with investigators, study sponsors, and other stakeholders, as well as from monitoring the medical literature. We did not involve patients or the public because the analysis focused on addressing largely scientific challenges in registering and reporting results information for MPRPs with complex designs.

MPRP overall screening record

An MPRP screening record can provide information that is common to all substudies in the MPRP, allowing each substudy record to focus on that study’s unique details. The record can include the procedures for assessing all eligible participants (and the eligibility criteria common to all substudies), the methods for allocating participants to substudies or a shared control group, and an overarching description of the research program. The screening record is updated over the conduct of the MPRP and should reflect the date on which the overall MPRP was opened to screen participants for eligibility (start date) and the date on which the eligibility screening ended for all substudies (completion date). Further attention is needed to optimize the reporting of other registration data elements. Particularly, whether the contents of the screening record should be limited to a description of the screening procedures (eg, listing only those interventions necessary for screening) or should the record also include specific key substudy information that provides an overview of the MPRP (eg, listing all interventions being evaluated as potential treatments in each MPRP substudy). Similarly, a screening record could focus on outcome measures related to the screening process, such as the number of participants whose tumor was sequenced, and the number of participants assigned to each substudy based on molecular sequencing.

After screening for all substudies ends, information on results specific to the centralized screening process can be added to the screening record. For example, in the participant flow module, the number started could include the number of participants screened, and the number completed could indicate the number of participants assigned to a substudy. In the adverse event information module, reporting would be limited to events collected during screening and would not include adverse events collected after participants were assigned to a substudy; these events would be reported in the relevant substudy record.

MPRP substudy records

When separate study records are used to represent each substudy in an MPRP, registration and results reporting on ClinicalTrials.gov generally follow the established processes for traditional clinical trials. Each study registration record describes a substudy’s design (eg, single arm or parallel design randomized multiarm study), specific eligibility criteria, prespecified outcome measures, arms and interventions, recruitment status, and facility locations. The substudy record should also include information connecting it to the overall MPRP.

Summary results reported to ClinicalTrials.gov for each substudy describe findings related to data collected from the participants allocated to that particular substudy. In some cases, the prespecified analysis plan for summary results will include a comparison to a control, such as data collected from participants in a shared control group, whether as standard-of-care or other control (fig 1). Although the term “shared control group” is used to refer to one group to which participants are assigned, in practice, the actual set of control participants analyzed can differ across substudies because the participants and interventions change over time, especially if the control is standard of care. If detailed plans for applying data from the control group are prespecified in the MPRP protocol,25 representing information about the specific participants analyzed in the control group in each substudy registration record (that is, as the analysis population for the control arm) would be more informative than as a separate shared control registration record. This information allows the relevant analysis groups in the trial (eg, experimental and control) to be represented and the details of the shared control specific to that substudy (eg, specific interventions and attributes of the participants) to be reported. When reporting results in a substudy record, the relevant set (or subset) of shared control group data should be included. Results from participants (or subsets of participants) in the shared control group can be reported in the control data for more than one substudy record.

We exemplify such reporting of a shared control arm using the trial STAMPEDE (Systemic Therapy in Advancing or Metastatic Prostate cancer: Evaluation of Drug Efficacy MPRP; NCT00268476), an MPRP that is evaluating treatment strategies for prostate cancer.26 STAMPEDE was initiated with one standard-of-care shared control arm and five experimental intervention arms (NCT00268476v.1). Note, an NCT number followed by “v.X” (when X is a number) denotes archived version X of the study record; to view archived versions of a record on ClinicalTrials.gov, click on the “History of Changes” link displayed on the study record. Results of STAMPEDE were then published in two articles, one presenting findings from three of the arms (zoledrenic acid, docetaxel, and zoledronic acid plus docetaxel)27 compared with the shared control group, and the other presenting findings for the two other arms (celecoxib and celecoxib plus zoledronic acid)28 also compared with the shared control group. More experimental intervention arms have been added and analyzed independently of the others, including abiraterone with prednisolone29 and radiotherapy,30 each compared with data collected from patients assigned to the standard-of-care shared control group contemporaneously to the experimental arms (NCT00268476v.50). This pattern of adding new arms and reporting their results independently indicates that an MPRP study design is well suited for the proposed multirecord approach, with each registration record identifying the prespecified interventions and analytical comparisons for each substudy. By contrast, a less common MPRP design plan that prespecifies the pooling of data across substudies for analysis might not be appropriate for multirecord registration of an MPRP.31 For example, a biomarker strategy design, which was implemented in the TCA Ovarian Cancer Trial, compared pooled results from assay-directed assignment of participants to one of 12 biomarker-targeted chemotherapy interventions (biomarker-strategy arm) to participants assigned by physician’s choice (control arm).32Table 2 displays key considerations for reporting MPRPs to ClinicalTrials.gov using multiple substudy records.

Additional factors to consider

Although the use of multiple study records greatly facilitates registration and results reporting for an MPRP, some challenges remain.

Harmonizing trial registries

We have described an approach to registering a single MPRP using multiple study records in the context of the ClinicalTrials.gov registration and results reporting model. Study sponsors might also have an obligation to report MPRPs to other trial registries; therefore, consideration of coordinating MPRP reporting approaches across registries is important. A harmonized approach would entail efficient and reliable registration and results reporting processes that support sponsors and investigators and minimize confusion among end users when information about the same trial is found in multiple places. Because registries support trial reporting policies and legal requirements, broad global collaboration and harmonization across requirements to support this approach for reporting MPRPs would be welcomed.