Data Monitoring Committees (DMCs) review data from ongoing clinical trials and make recommendations based on risk-benefit and other criteria — they are an essential component to help protect the safety of current and potential patients. However, while the ethical benefits of including a DMC are well-discussed, less so are the financial considerations of DMCs. These financial considerations consider both the overall cost of the DMC and the potential savings of early stopping or other information learned earlier to help improve the likelihood of success of the study, or success of other studies within the clinical program.

Implementing a DMC is low-cost relative to the overall cost of the study and yields significant value. There is even more value when stopping guidelines are thoughtfully put in place. Additional value is found with forethought to the DMC and the groups supporting the DMC and the processes in place for DMC review. So, while acknowledging there is a cost to having a DMC, the DMC is certainly no place to skimp — it is well worth the cost, and the components below will show the areas to ensure that the cost of the DMC is value-added to the clinical program and doesn’t go to waste.

For some studies, the DMC is like an insurance policy. If results are not as expected, the DMC protects against excess patients being put at risk, or money being spent on continuing a study where the answer is already clear. In both cases, we want a good insurance plan — a high-quality DMC and high-quality support of the DMC so that the assessments can be performed to serve as that insurance.

Below are ten key components of the cost-effective DMC:

1. Stopping guidelines for overwhelming efficacy

More frequent use of stopping guidelines can allow for earlier stopping of the study, resulting in cost savings. Even if there is a decision to continue the study in order to allow more information to be gathered (perhaps with subjects crossing over to the investigational arm), a result of overwhelming efficacy mid-study could allow for extra months or even years of acceleration to the marketplace.

2. Stopping guidelines for statistical futility

More frequent use of stopping guidelines for statistical futility would greatly improve the efficiency of clinical research. In the context of many studies, if there is no observed benefit at, say, three-quarters of the completion of the study, there is limited value in the expense of continuing the study to completion. Too many studies don’t have stopping guidelines for futility in place. Stopping a study for statistical futility saves time, money and potentially enrollment if enrollment is still ongoing at the time of the analysis. Appropriately recommending stopping for statistical futility not only has a benefit on savings within the study or other studies being conducted in parallel within the clinical program but also allows for reinvestment of resources to other studies — perhaps conducted more precisely given the information gained from the stopped study.

3. Stopping guidelines for safety and informal futility

Even if there is no guideline for statistical futility, there is still value in providing the high-level context of the study to the DMC and using them as a partner in clinical development. There are many situations where — with relatively mature data — there is a slightly higher rate of toxicity in the investigational arm, and it is fairly apparent to the DMC that there is no or only minimal efficacy. Each aspect alone might not be compelling enough to recommend stopping, but in combination with a comprehensive review of the data, there may be sufficient evidence that there is no future for a regulatory pathway to approval.

Providing high-level guidance to the DMC and supporting them to use the totality of data (perhaps aided by additional outputs to the DMC such as conditional power or non-binding futility bounds) can result in appropriate recommendations to stop early in such situations (to be reviewed and likely — but not necessarily — endorsed by upper management at the sponsor), resulting in cost savings. Some sponsors have criticized the DMC at the end of a completed study for not making a recommendation for early stopping when the answer was clear, but the sponsor also did not proactively give guidance in the charter or at earlier DMC meetings about the context of the study in terms of regulatory and other corporate concerns.

4. DMC membership

Early on the DMC process, the decision will need to be made on how many DMC members are empaneled. The smaller the group, the cheaper the payments to the DMC. But there will need to be sufficient expertise within the DMC to fully evaluate the data that they are reviewing. Having DMC members with extensive past DMC experience could increase the cost, but that experience is likely well worth it. That being said, having a range of DMC experience within the DMC can work well. The DMC can also — especially prior to the first DMC data review — act as another group of key opinion leaders to help give insights into the study design and protocol review. That early insight could help the eventual results be more obtainable and compelling to the clinical community.

5. DMC payment structure

DMC members are paid for their efforts using fair-market rates, which are determined at an individual level based on role and experience. The decision will also need to be made whether members are paid hourly or per-meeting. In addition, payment for atypical reviews — such as offline reviews or in-person meetings — may also need to be considered.

6. Frequency of DMC meetings

More frequent DMC meetings will cost more — both from DMC payments, but also from the CRO’s and sponsor’s time needed to support the DMC review. So, the balance must be made between sufficient cadence to ensure the safety of study participants as the data evolves and incorporate any formal interim analyses of efficacy, while not exhausting resources by having too frequent meetings. A thoughtful approach will give the most value — for example, conducting meetings more frequently early on if enrollment is rapid and there is not much information on the safety profile, and less frequent meetings later in the study after enrollment is complete (and perhaps even treatment is completed) and subjects are in long-term follow-up. Or perhaps based on accrual of endpoints, or accrual of specified patient-years of follow-up.

7. Data used for DMC review

The data used as input for the DMC will not be as clean or complete as expected for final analysis. Derivations used for final analysis (such as visit windowing or date imputations) may not be needed for standard DMC outputs. The use of formally created ADaM datasets also may be overkill for DMC reporting. The DMC is focused on major issues of safety. Aspects of the data that only would change results on the margins may not be worth the effort to implement. (This does not apply to formal interim analyses the DMC might be asked to review.)

Implementing a DMC also can serve to stress-test and accelerate data and programming that would be used for the final analysis. Having a DMC review every four months (as an example) ensures that data management vendors are actively ensuring that extracts can be made at the necessary level of completeness and cleanliness. The sponsor and groups facilitating DMC work can check for gross inconsistencies and outliers that may not have been triggered for queries in the data. The DMC or group facilitating the DMC work may even specifically check randomization and kit lists for issues or inconsistencies that might be invisible to queries by the sponsor internally. Particularly for complicated randomization schemes, an early review by the DMC could be invaluable as the one external group to ensure that is successfully implemented before the study is too advanced the damage is irreversible. In a similar way, the DMC could review other restricted data (e.g., PK data, biomarker data) to ensure that no obvious issues are seen that would impact the validity of the study, but which would only be discovered otherwise at the end of the study.

8. DMC programming model

There are numerous models for how the group facilitating DMC activities (the statistical data analysis center, SDAC) can generate the outputs for the DMC. The SDAC can receive raw data and create programs to generate outputs. The SDAC could receive SDTM or even ADaM data and use that to generate outputs. The SDAC could receive programming from the sponsor or another CRO and use that to generate outputs (after splicing in the randomization assignment). The cost of the SDAC might change depending on the model used, but the overall expense for the study may not be all that different.

9. Outputs for DMC review

Careful consideration should be made of the tables, listings, and figures (TLFs) the DMC receives. There is a natural desire to leverage TLF programming that will eventually be used for the final analyses and use those for the DMC. However, the outputs for the DMC should be carefully designed for their use. Even if the TLFs for final analyses are available (not a certainty since the DMC may meet relatively quickly after enrollment starts), these are likely not truly what the DMC needs. An SDAC with focused experience in DMCs will be able to give guidance on the set of TLFs most appropriate and appreciated by the DMC. Depending on the situation, the DMC may be able to provide value to proposing protocol amendment or other actions to expand the benefit or rescue the value of the clinical trial. That needs to be carefully considered since advice from the DMC can be considered unblinding (for randomized, blinded studies) and therefore introduce bias. But in some situations, the DMC can provide advice mid-study to senior study leadership, which will not introduce bias, or is judged not to be excessive bias.

10. Program-wide DMCs

In many situations, a DMC will oversee multiple studies within a clinical program. This does lead to efficiencies. Generally, the programming effort is easier — assuming that the set of TLFs and the input data is fairly consistent across the studies. There likely will be fewer DMC meetings (although perhaps longer and more complex), assuming the DMC reviews multiple studies at each DMC meeting. Requested updates to TLFs or recommended minor action items will also be more efficient with only one DMC asking for consistent changes across the studies, rather than divergent (possibly inconsistent) requests from multiple DMCs. Having a program-wide DMC may also make it easier for the DMC to observe global trends and act accordingly to generate appropriate recommendations to stop or alter the entire clinical program efficiently.

Axio®, a Cytel Company, has supported clients in the planning and management of more than 1,000 Data Monitoring Committees in all major therapeutic areas — making us one of the foremost companies in this field. Our experts can help ensure that the key aspects of the DMC process run smoothly, while considering efficiencies that can lead to a more cost-effective DMC.

Interested in learning more?

Watch David Kerr’s on-demand webinar, “Maximizing the Value of the DMC: A Cost-Effective Approach to Safeguarding Clinical Development”:

Share Via