Walk, Don’t Run: Considerations Before Data Monitoring Committee Review Begins
There’s often a rush to form data monitoring committees (DMCs) and get the organizational meeting out of the way as study initiation approaches, but sponsors and sites need to take time to make sure that plans for the group are solid before their work begins.
The initial DMC organizational meeting should be held, ideally, prior to the start of participant recruitment and is still best done in-person, even with the remote measures that have been adopted and successfully used during the pandemic, said Matt Downs, a statistical scientist at WCG Statistics Collaborative. That’s because the committee will, in many cases, be working together for several years.
Because it involves a review of the trial’s study design, the organizational meeting marks a critical last chance for the committee to give feedback. It is also the point where discussions on potential trial obstacles can be held while the DMC is still naïve to results.
While some opt to hold the organizational meeting after recruitment begins, that’s not recommended, according to Downs, who offered his “hit list” of critical items that should be addressed at the organizational meeting at a WCG webinar last week.
First, make sure the DMC’s charter — the “manual of operations” for the committee that lays out its process for the trial — is crystal clear and explains what exactly the committee is responsible for. While trial safety and conduct are almost always part of the DMC’s responsibilities, it may not always be asked to formally evaluate whether to stop the trial early for benefit or for futility. If stopping for benefit or futility is not something intended to be part of the committee’s charge, that should be explained at the organizational meeting.
“Let’s say you’re in an oncology setting and you’ve been told by the FDA that basically you should allow this trial to go to its completion because we want to ensure that there’s a sufficient safety database, and we’re basically not going to approve the drug if you don’t have sufficient safety data. That would be … something to tell the DMC at the organizational meeting,” Downs said. “Therefore, a recommendation to stop early for benefit might not be appreciated. It’s good news in a way but it often puts the sponsor in sort of a difficult bind.”
Similarly, a recommendation to stop for futility may not be desired if it’s already well-known from related indications that the drug is working and the trial is being done to expand the label.
Another key point is whether the DMC should have access to efficacy data, “one of the stickiest issues that comes up,” Downs said, warning that if the subject isn’t brought up at the organizational meeting, it’s going to come up later and will likely be an issue.
“The point here is that a lot of DMCs feel they need to have access to the efficacy data, even when reviewing safety, because they’re being asked to assess risk vs. benefit, so if you’re seeing a potential safety signal, if there’s not an offsetting benefit, that would make that safety signal acceptable. They might be inclined to stop the study, but if they’re not seeing that potential benefit, they only see the risk, they’re not really able to make that calculus,” he said.
Sponsors may have a number of reservations about giving the committee access to efficacy data. For example, they may be concerned that regulators would require a statistical penalty for each safety look when efficacy data are also presented.
The currentness of data is also an important item to discuss at the organizational meeting, and it’s a topic that is often not brought up. Sponsors and DMCs should discuss how current the data the committee receives will be, Downs said, noting that in some cases, committees have been aggravated at the first safety review because the data were not as recent as they’d liked it to be.
He also advised that the charter allow the DMC to conduct unscheduled meetings without notifying the sponsor. For instance, the committee may want to hold those meetings if they’re seeing a potential safety signal and don’t want to wait on the review cycle to assess the data.
During the organizational meeting, DMCs should also be consulted on what level of masking they want in their closed report. Most commonly, it’s decided that the committee should know the identity that corresponds to the labels. By contrast, some committee members will feel that the DMC should only know the identity by request and if needed.
“Very often, the report template that’s reviewed at the organizational meeting, that’s a starting point,” Downs said. “The DMC report, it actually often will change quite a bit over the life of the study. If there’s additional analyses that are needed because the DMC is considering stopping for futility, they might want to look at subgroups to see if there’s some subgroup that’s showing any sign of hope, and there might be new safety signals that require more follow-up.”