Data Monitoring Committees (DMC) are a key component of the clinical trial process to help ensure that ongoing studies are safe and ethical to continue, especially when a study is randomized and the sponsor does not have access to by-arm results. In this situation, the DMC has access to by-arm results to make recommendations about the trial, ensuring participant safety, while the sponsor stays blinded to help reduce bias. A Statistical Data Analysis Center (SDAC) is called upon to help facilitate DMC review of study results. This typically involves the DMC reviewing output generated by the SDAC at specified intervals (e.g., every six months, after every 40 patients enrolled, or after 25 deaths occur).

This blog describes other activities where the SDAC — by the nature of them having access to the clinical and randomization data and having strong biometrics knowledge and processes for handling sensitive data — can play a supportive role for both the DMC and the sponsor, beyond just the typical “data review meeting” situation.

Randomization reviews: SDAC ensures randomization integrity in complex studies

The SDAC may assist the sponsor with randomization reviews early in the study, particularly if there is a complex randomization schema. The SDAC can verify that patients are randomized properly according to the master randomization schedule. The specific checks performed, as well as the level of information communicated to the sponsor, should be predefined based on the randomization method, enrollment rate, and randomization parameters.

Examples of randomization checks performed by the SDAC may include:

• Reviewing the initial list prior to any randomization to ensure it meets expectations
• Verifying that all participants randomized to date have been allocated to the correct blocks and that blocks are completed as planned
• Confirming that the kits or assignments received by the participants (from a central group assignment list or the sites entered into a CRF) match the treatment assigned by the system

The DMC is not necessarily involved in, or informed of, this process. However, if a true process error is discovered during the randomization audit, the DMC should be informed. If issues are observed, the sponsor might be informed at a high level, depending on the situation, while detailed discussion of the randomization or treatment of specific patients would be done directly between the SDAC and the vendor providing randomization and/or treatment allocation.

These reviews can be conducted using information provided solely by the randomization vendor or in conjunction with merged CRF data entered at the site. Examples of issues identified include:

• A randomized open-label extension was not properly implemented by the randomization vendor in a complicated protocol
• A “user acceptance test” randomization list was not properly transitioned to the actual randomization list at study go-live

• A randomization number entered into a CRF did not match the number assigned

• Inconsistencies between treatment kit numbers recorded in the exposure CRF and those assigned by the electronic system, which could reflect a data entry error or that the participant received the incorrect kit and (possibly) the incorrect treatment

• The randomization ratio was incorrectly specified
• A large group of subjects was missing from a treatment arm due to errors in the export format of the data

• Subjects were incorrectly randomized to an arm because the study design changed and the randomization specification was not revised

These examples vary in severity; however, some — particularly those involving CRF data discrepancies — need to be investigated and cleaned prior to final database lock. Since sponsors typically only obtain randomization data after database lock, the SDAC is often best positioned to identify these issues.

Event tracking: Supporting time-to-event studies through blinded operational insights

Occasionally the DMC or SDAC may be asked to provide input on event tracking, particularly in time-to-event studies. Event tracking typically relies on confidential information, but not necessarily confidential randomization information.

For example, consider a three-arm study comparing control, monotherapy, and combination therapy. The primary focus may be the two pairwise comparisons of monotherapy vs. control and combination therapy vs. control. The protocol may specify that 300 events are required within each pairwise comparison of active therapy against placebo before an interim or final analysis is conducted. Table 1 below shows three scenarios with 450 events total that yield three different situations for which pairwise comparisons, if any, have met the minimum number of events required.

Table 1: Examples of Pairwise Comparisons in Three-Arm Study

Through ongoing event tracking, the DMC or SDAC can inform the sponsor when the 300-event threshold in each pairwise comparison has occurred, giving the green light for the execution of the interim or final analysis.

Event tracking may also include key blinded biomarker information. For example, the study might require a certain number of events in the overall intent-to-treat () group while also monitoring the number of events from participants whose baseline results exceeded a key blinded biomarker. In such cases, the SDAC can review both endpoint and biomarker data and provide periodic updates to the sponsor without disclosing patient-level biomarker data.

Event updates may be communicated in several ways, including:

• Simple “yes/no” notifications indicating whether enough events have occurred
• Semi-blinded binning (e.g., “270–279” in one pairwise comparison and “260–269” in the other)

Semi-blinded binning can help reduce the risk of unblinding the sponsor to emerging treatment effects during the period between these analyses. However, even with semi-blinded binning, event tracking information can still be informative to treatment effect and a carefully defined communication plan is strongly recommended.

In some cases, event tracking may also include non-inferential statistical summaries of events to date, such as the number of events that have accrued per month.

Event projection and forecasting study milestones

While event tracking focuses on the number of events at a specific time, event projection estimates the date at which a target number of events is expected to occur. Event projection is typically requested in event-driven studies where the primary endpoint is time-to-death or time to PFS (death or disease progression).

Event projection models can incorporate enrollment rates, discontinuation rates, and hypothesized treatment effects. When these calculations can be conducted without randomization knowledge, they may be conducted internally by the sponsor. However, a sponsor may seek support from the SDAC, given their biometrics resources and access to specialized software.

In some cases, more precise event projections can be obtained by factoring in the current observed treatment effect. Such analyses require access to unblinded data and are therefore more appropriately conducted by the SDAC.

As with any analysis involving randomization knowledge, the dissemination of results should be done using a carefully pre-defined communication plan to avoid any appearance of unblinding and introducing bias to the study.

Collaboration with external vendors: Supporting PK/PD analyses with controlled unblinding

In some cases, the SDAC is asked to support internal sponsor groups or other vendors, such as for pharmacokinetics/pharmacodynamics (PK/PD) analyses. Those analyzing PK/PD data may only want to focus on patients who were treated with the sponsor’s investigational product, and have no interest in patients treated with placebo or active control. Through careful planning, the SDAC could pass along the needed randomization data and other relevant CRF data.

For example, a sponsor may want to conduct PK/PD analysis on all participants on active and 10% of the participants on placebo. The SDAC would be responsible for supplying this specific data.

Leveraging DMC outputs for DSUR and CSR preparation

The outputs provided to the DMC are designed to fit the needs of the DMC in their role of reviewing key results from an ongoing study and are not intended for other audiences. Nonetheless, these outputs can also be leveraged for other uses as an efficient use of funds. For example, a subset of outputs from the DMC package could be used for the Development Safety Update Report (DSUR) on an annual basis. Key results from the DMC outputs could be routinely cross-checked against outputs independently created by the sponsor or the group that will create the Clinical Study Report (CSR) to ensure programming standards match. In some situations, the DMC outputs are run one final time after database lock to provide the sponsor with first interpretable results, since the sponsor or another vendor may take longer to generate a more comprehensive set of outputs that utilize a fully-compliant CDISC process.

The SDAC may also be called upon to help with requests from regulatory agencies. For example, for a study that has met statistical criteria for progression-free survival (PFS), but long-term overall survival (OS) data continues to accrue, the sponsor might apply for accelerated approval based on PFS but remain blinded to ongoing OS results. A regulatory agency might request current OS results as part of their review and the SDAC can support that request. With prior approval from the sponsor, the SDAC and the regulatory agency could communicate directly, or a small group within the sponsor’s regulatory department could facilitate the transfer of requests and responses between the SDAC and regulatory agencies.

Final takeaways

SDACs support far more than periodic DMC reviews. From randomization oversight and event tracking to vendor collaboration and regulatory support, SDACs play an important role in maintaining trial integrity while preserving study blinding.

In Part 2 of this blog, we will examine the SDAC’s expanding role in safety surveillance, adaptive trial design, and cumulative risk assessment. Stay tuned!

Interested in learning more?

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

Share Via