FDA Unveils Its Own Medical Queries—A Standardized Approach for Grouping MedDRA Preferred Terms that Will Impact NDA/BLA Safety Analyses and Drug Labeling

On September 14, 2022, FDA/CDER/Office of New Drugs, in collaboration with the Duke-Margolis Center for Health Policy, hosted a virtual meeting on advancing premarket safety analytics, including sessions on new FDA Medical Queries and standardized presentations of safety data. For a number of years, FDA has been including groups of related preferred terms in tables in the Adverse Reactions Section of drug labeling (Section 6), generally describing such groupings with the use of footnotes. For example, Table 20 in Section 6 of the current Latuda labeling includes the term “somnolence” with the footnote “Somnolence includes adverse event terms: hypersomnia, hypersomnolence, sedation, and somnolence.” Such groupings have generally seemed to appear in labeling on an ad hoc basis, without standardization. At the September 14 virtual meeting, FDA demystified these groupings by announcing the FDA Medical Queries (FMQs), followed by an open discussion. The FMQs include some 100 standardized groups of related MedDRA preferred terms to be used in the identification and labeling of adverse drug reactions.

Typically, clinical trial subjects are questioned regarding adverse events, and investigators record them in their own words. Such descriptions are called ‘verbatim terms,’ and may include medical shorthand. For example, ‘Hip Fx after fall’ might be recorded for a patient who fell and sustained a hip fracture. A verbatim term cannot be analyzed, however, until it is translated into its corresponding standard ‘preferred term,’ or in this case, two preferred terms: ‘Fall’ and ‘Hip Fracture.’ There are over 24,000 preferred terms, each serving essentially as a standard safety outcome that can be tabulated for a clinical trial(s). For example, one could calculate the percentages of patients in the drug and control groups with a hip fracture. The problem is that adverse reactions are generally broader than a single preferred term. In this case, for example, it seems likely that if a drug predisposes patients to hip fractures, it would predispose to other fractures. Thus, one would like to quantify all fractures—not only hip fractures. Thus, as illustrated by this example, the objective of a query is to combine similar terms to create meaningful analyses of adverse drug reactions, e.g., fractures, pneumonias, seizures.

FDA shared the following example to show how the use of queries can lead to a more accurate characterization of adverse drug reactions. In this example, only adverse events with a frequency >2% in the drug group were to be included as adverse drug reactions. When assessing anxiety as a single preferred term, the frequency in the drug group was slightly less than 2%; therefore, anxiety was not classified as an adverse drug reaction. When related preferred terms were included in an anxiety query, e.g., ‘nervousness,’ ‘general anxiety disorder,’ the frequency exceeded 2% and was included as an adverse reaction in the drug labeling. This example represents a situation where the existence of the adverse drug reaction depends on whether it is based on a single preferred term (‘anxiety’), or a grouping of anxiety-related preferred terms.

MedDRA, the Medical Dictionary for Regulatory Activities, maintains the list of preferred terms used internationally, and provides a large number of Standard MedDRA Queries (SMQs), which are used routinely by many companies. As I pointed out at the meeting however, as noted on the MedDRA web site, “SMQs are tools developed to facilitate retrieval of MedDRA-coded data as a first step in investigating drug safety issues in pharmacovigilance and clinical development.” Conversely, the FMQs have been developed specifically for use in assessing the safety of new drugs and biologics in clinical development. Dozens of FDA medical experts contributed to the development of the FMQs, all with longstanding interest in drug safety, which makes them fit-for-purpose. FDA is also beginning the development of “algorithmic” FMQs that combine preferred terms with laboratory data and temporal information.

The discussion on FMQs was followed by an introduction to FDA’s new “Standard Safety Tables and Figures: Integrated Guide.” FDA provided presentations on standard safety tables and figures, tabulation of adverse events, statistical considerations in the analyses of adverse events, discussion of relative risk vs. risk differences, and advice on pooling trials (including Simpson’s Paradox), ascertainment windows, standard laboratory analyses, and drug-induced liver injury.

It seems likely that FDA will encourage use of FMQs and these analytical methods and presentations for premarket safety data, but will not mandate these activities at this time. It also seems likely that FDA will be running these FMQs on many NDAs and BLAs under their review, and will be incorporating the results in labeling. Running FMQs on clinical datasets prior to, or during, NDA review could shed important light on FDA’s safety concerns and may become an industry best practice. Because we believe such information is highly important, HPM has developed the capacity to run the FMQs for our clients with a rapid turn-around time.