As part of Novartis’ new drug application for its recently approved multiple sclerosis drug Mayzent (siponimod), the US Food and Drug Administration (FDA) questioned the drugmaker’s data practices during a clinical inspection.
“During the sponsor inspection, significant data reliability concerns were identified,” FDA said, citing the company with an Official Action Indicated (OAI).
FDA explained how in one study protocol
, blinding “was not adequately maintained as specified in the protocol throughout the course of the trial at 62 (21%) of 294 sites. Study personnel were given inappropriate access to the first dose and main databases affecting 285 (17%) out of 1651 total study subjects.”
The agency found that tracking user activities in two databases, such as which users accessed databases and when, as well as what data were viewed, was limited because of a lack of system access audit trails.
FDA also noted that “Novartis did not have sufficient procedures outlined in their monitoring plan for granting and revoking access as well as detecting inappropriate access to databases that contained study information that could be potentially unblinding. Novartis CRAs [clinical research associates] did not properly grant study personnel (e.g., study coordinators, investigators, and EDSS raters) access to the databases used in the trial. In addition, the CRAs failed to identify cases where inappropriate database access was granted to site personnel.”
But FDA could not say whether this inappropriate access to certain information led to bias.
However, the agency also said it could not rule out the introduction of bias in terms of the reliability of the study data, “specifically for 163 of the 285 subjects who had study assessments performed by main database users and EDSS raters (i.e., blinded study personnel) who were granted inappropriate access to the first dose and/or main databases.”
In an email last December, FDA called on Novartis to conduct a sensitivity analysis, excluding in the per-protocol analysis subjects who may have been impacted, including 62 subjects who were potentially affected by the 11 EDSS raters who had inappropriate access to the first dose or main databases and 101 subjects who were potentially affected by the 32 users of the main database who had inappropriate access to the first dose database.
Novartis did not respond to a request for comment.