Both the Society for Clinical Trials Annual Meeting in Phoenix, AZ, and the PSI Annual Conference in Belfast, UK, highlighted the future of Data Monitoring Committees (DMCs).

Across panels and posters, the following were highlighted as areas for positive change:

• The remit of DMCs
• How DMCs are supported by Statistical Data Analysis Centers (SDACs)
• Who are empaneled on DMCs

What stood out was not the importance of these topics to DMCs, but the breadth of ideas within each one.

Extending the scope of DMC activities

Risk-benefit is a baseline expectation, not an expansion of scope

DMCs are expected to review benefit-risk as a core tenant of their obligations in addition to any formal statistical monitoring. While this may feel like a settled topic codified in regulatory guidance and industry best practice, the fact that panels open with this question underscores the lack of adherence to this best practice.

There are sponsor requests for “safety only” DMCs. In these cases, DMCs are provided with safety data and then often use available data — such as reasons for treatment discontinuation — to infer whether there are concerns about the effectiveness of treatments that would make it unethical to continue the trial. It is more efficient and transparent to provide the benefit data. It’s important to note that benefit data are generally non-inferential in nature and may not be the same data that represent the primary endpoint.

DMCs monitoring safety thresholds of interest

DMCs traditionally review unblinded safety and effectiveness data to recommend to the sponsor to continue, modify, or stop the trial based on accumulating data. DMC remits are expanding to more broadly to include monitoring safety thresholds of interest.

“Although DMCs and entities that review accumulating safety data have distinct roles in characterizing safety, it may be possible in some settings to have the DMC conduct these safety evaluations and provide recommendations to sponsors about whether a difference in the occurrence of safety events in the investigational arm compared to the control arm suggests a causal relationship between the investigational product and the adverse event” (DMC FDA Draft guidance).

This topic received much more attention at PSI than SCT, which parallels Cytel’s experience where we see this monitoring extended to DMCs in trials with a larger proportion of subjects enrolled in Europe. In such cases, DMCs are tasked with monitoring a specific threshold and indicate via a checkbox on the DMC recommendation form whether the threshold was met or not.

Adding on-demand visualizations while meeting foundational requirements

Visualizations as an add-on beyond the basics

SCT focused on DMCs being first provided with the basics required to make a recommendation before any additional features are added to DMC summaries. Specifically, DMCs should be provided with a single, bookmarked, PDF document of closed outputs that correctly summarize the appropriate populations in a digestible format for the DMC. Once that threshold is met, there is interest in on-demand visualizations for investigating any items of interest in real time, with this capability most likely led by the independent statistician. While this paradigm does not represent a significant departure from the independent statistician answering real-time queries in the closed session about the data and derivations, visualizations can be a value-add to the DMC’s ability to see the full data story.

PSI focused on how visualization tools could make data more digestible for the DMC without changing current reporting. There was a greater focus on on-demand visualizations becoming a core part of DMC review in the future. Regardless of how on-demand implementation tools are implemented, a full compliant audit trail remains a core capability requirement to ensure that the summaries DMCs are using to develop their recommendation are appropriately documented.

Changing landscape of TLF programming support

DMC TLFs are often supported in one of two operational models:

1. SDAC programs datasets and TLFs
2. Sponsor or sponsor-vendor programs datasets and TLFs in a blinded fashion and SDAC unblinds

In the sponsor-provided programming model, the TLFs for the DMC are often substantially based on TLFs used for other analyses such as the final study reporting. Summaries can be quite long and may not be fit-for-purpose. Using sponsor-provided programming is an increasing trend in DMC support as sponsors look for ways to optimize costs and focus efforts of their biometrics teams. DMC members at SCT expressed concern about this trend, while sponsors noted it allows them to invest biometric team efforts — particularly around data derivation reviews — in a single place. A third option is available that bridges both concerns: sponsor or sponsor-vendor programs datasets in a blinded fashion, while the SDAC programs the TLFs and unblinds the datasets. This allows sponsors to focus their efforts on derivation review to a single set of analysis datasets and ensures that the DMC TLFs developed by the SDAC are fit for purpose.

Expanding the DMC candidate pool

Solving for logical constraints

The number of studies where an independent DMC is recommended continues to increase. With shifts in the academic space, we see fewer people being exposed to DMCs through academia. This creates a dynamic where the experience level of DMC members is high, but the number of DMC members is not keeping pace with demand. Naturally, sponsors want experienced DMC members to serve on their DMCs. There are several ways to ensure appropriate expertise while growing experience:

• In cases where DMC meetings are scheduled with limited notice, such as dose escalation studies, the most experienced DMC member may have less scheduling availability. This is an opportunity to use members with appropriate expertise and more flexible schedules.
• At SCT, a key focus was on having less experienced members shadow experienced members. These members would still be required to complete confidentiality and disclosure information and be compensated for their time. While additional members incur some level of additional burden for the sponsor and SDAC, it increases the pool of members the sponsor can choose from for future studies. This has been used in limited cases.

Final takeaways

The conversations at SCT and PSI made one thing clear: Data Monitoring Committees are evolving alongside the increasing complexity of clinical trials. Expectations are expanding beyond traditional safety reviews to encompass broader risk-benefit assessment, more sophisticated data visualization, fit-for-purpose statistical support, and new approaches to developing the next generation of DMC members. While there was no universal agreement on exactly how these changes should be implemented, there was broad consensus that existing models will need to adapt. For sponsors and SDACs alike, the opportunity is not simply to modernize DMC operations, but to ensure that independent oversight remains effective, efficient, and scientifically rigorous as the clinical research landscape continues to evolve.

Interested in learning more?

Download your copy of our white paper, “DMCs for Oncology Studies”:

Share Via