When inspecting clinical trials that incorporate real-world data, the UK medicines regulator says that its good clinical practice (GCP) inspectors expects sponsors to be able to demonstrate that the RWD database they used was suitable for the study and of sufficient quality.
As inspectors, “it’s certainly not our intention to approve registries” or other RWD sources used in a trial, said Jason Wakelin-Smith, lead senior GCP inspector at the Medicines and Healthcare products Regulatory Agency.
But ultimately, following an inspection, the aim is to provide feedback to the MHRA’s assessors on “whether they can have confidence in the processes sitting behind the presented data, and the integrity of the data provided,” he said. Wakelin-Smith was speaking at the MHRA’s 2022 GCP symposium, which was held virtually, from 7 to 9 March, in partnership with US and Canadian regulators.
While RWD is extensively used for monitoring the performance of approved drugs and devices, it is used less frequently for demonstrating the safety and efficacy of an intervention. The use of such data in clinical studies has the potential to increase the speed, and to reduce the cost, of development programs.
The inspection of trials that use RWD would depend on the type of database/resource that is selected, its use in a particular trial and whether there was “prior experience” with that resource as “there are likely to be [RWD] datasets which are commonly used” and these may have been inspected before, said Wakelin-Smith.
The inspection would follow the “data lifecycle” – from acquisition through to reporting – from the point of view of the sponsor or the contract research organization, he explained. The focus would be on aspects, such as, how the RWD resource was selected, how it was deemed suitable and used during the trial, and how the data was extracted. This would then feed into the data management analysis processes of the organization being inspected, he added.
As RWD can “come in all shapes and sizes” and vary in complexity – from disease registries, national health care datasets to patient-reported outcomes – Wakelin-Smith said that sponsors should “carefully consider the provenance of each data set, as well as the processes that underpin the collection and the management of the data in order to support its use in the clinical trial arena.”
Not all RWD sources are equal. Some disease registries “might be backed up by legislation, which in turn is likely to improve the reliability of the information they contain,” he explained. However, as sponsors have no control over how RWD are generated, they must ensure that “they are happy with the dataset and that it’s suitable for their needs, and ultimately contains the required data.”
Wakelin-Smith said the clinical trial protocol was “king” and remained the “prime method of control” over the use of RWD. It should describe “what data will be obtained, what it will be used for and how it will be used in the context of the trial.”
To help sponsors evaluate the quality of an RWD source for its intended use, the MHRA recently issued guidance on this topic. (Also see “UK Explains How To Plan For Randomized Trials Using Mainly RWD Sources” – Pink Sheet, 31 Dec, 2021.)
The senior GCP inspector further elaborated on the guideline and outlined factors that sponsors should consider when deciding whether RWD is suitable for use in a clinical trial. These include:
- Having clarity on the intended purpose of the RWD in a trial – Potential uses can range from helping to identify potential trial participants, supplementing trial data, as a control arm or to identify adverse events. Wakelin-Smith emphasized the importance of choosing the right dataset for the intended job, as well as ensuring that the selected source was of an appropriate quality.
- Understanding the contents of the source database – To ensure that the selected data source will be able to meet the demands of the clinical trial endpoints, Wakelin-Smith said that sponsors should decide whether the database contains an appropriate population in terms of size, coverage and representativeness in order to enable a clinically important treatment effect to be detected. Statistical power calculations can be used for this purpose and “it may be prudent to seek regulatory advice as part of this process,” he added. Other factors to consider on this front include interoperability issues and deciding whether the database can capture the interventions, outcomes and other study variables in sufficient detail and consistently, without bias and with the level of frequency and completeness that is required for the study.
- Understanding the processes of the source database – Clarity is needed on how an RWD source is managed, for example, how often it is refreshed, how queries are managed, what checks are conducted by the data owner, and how data are retained long term and whether these remain accessible for inspection. Also, the “overall quality of the data needs to be considered” to address issues around the “noise of error” in the data, said Wakelin-Smith.
- Data management – Clarity is also needed on whether a case report form (CRF) would be used alongside the RWD extract and for what purpose, eg to collect linkage information or collect other data. In some cases, “it is possible that in reality the CRF is just a specification for the extraction of certain data points” from the RWD dataset, the GCP inspector noted.
- Addressing missing data – Once data is extracted, GCP requirements kick-in and the “real world datasets should be treated in the same manner as any other clinical trial data,” said Wakelin-Smith. He acknowledged, however, that there were “limitations associated with the ability to query inconsistencies between-real world data and the collected trial data” and challenges around “being able to follow up missing data points.” While it is possible to link several RWD sources to gain a more complete data set, this would involve ensuring compliance with data protection requirements, maintaining informed consent as well as ensuring that accurate, reliable linkage is possible. “Ideally, this should be validated prior to use,” he added.
- Investigator oversight – Sponsors also need to consider how the use of RWD would tie in with the central tenet of GCP regarding the trial investigator’s responsibilities for the care of the patient and of the associated data gathered relating to the patient. Consideration is needed on how the collection of RWD will feed back to the investigator, “particularly where there may be the opportunity or ultimately the need for decision-making based on that additional data,” said Wakelin-Smith.
Also at the symposium, the US Food and Drug Administration’s Cheryl Grandinetti explained the agency’s approach to inspections for drug applications incorporating RWD. The FDA expects sponsors to submit patient-level data to allow the agency to replicate the study analysis using the same dataset and analytic approach.
Source: https://pink.pharmaintelligence.informa.com/PS145840/UK-Regulator-Explains-Quality-Expectations-For-Use-Of-RWD-In-Clinical-Trials