In an effort to have a black box warning removed from the label for its smoking-cessation product varenicline (Chantrix), Pfizer recently conducted postmarketing studies of the drug’s psychiatric risks, including suicidal thoughts, hostility, and agitation. Now, in a blow to the company, FDA staff members have questioned the collection and interpretation of the data––two days before the agency’s Pharmacologic Drugs Advisory Committee and Risk Management Advisory Committee will hold a joint meeting to discuss removing the psychiatric warning.
In March 2015, the FDA left the warning on the varenicline label while awaiting the outcome of the postmarketing investigations, which involved more than 8,000 adult smokers.
Pfizer will lose patent exclusivity for varenecline in four years.
In a briefing document prepared for the upcoming joint committee meeting, the FDA staff expressed concerns with the ascertainment of the primary neuropsychiatric adverse event outcomes.
“In brief,” the staff wrote, “the interview tools that were developed to optimally capture the neuropsychiatric adverse events of concern were not utilized to the extent intended to get a nuanced understanding of such adverse events. In some cases, lack of information about the circumstances of events limited [the] FDA’s ability to determine whether they were primary outcome events. Additionally, there was substantial variability among investigators in how the adverse event severity was coded and [in] how the adverse event terms were applied, leading to variability in the number of neuropsychiatric adverse events included in the primary outcome. Finally, there was inconsistency in how suicidality cases were handled, with the Columbia Suicide Severity Rating Scale (C-SSRS) results not being reconciled with the adverse event reporting. All of these factors likely served to lower the overall number of primary outcome events, which may result in biasing towards a null finding.”
In addition, the staff noted that each of the reviewed studies had key design limitations. The most important limitations were: 1) the use of outcome measures with suboptimal sensitivity and specificity; 2) residual confounding; 3) the use of bupropion (another smoking-cessation drug with a neuropsychiatric risk) as the reference group to examine the neuropsychiatric risk of varenicline; and 4) the inability to assess the influence of pre-existing psychiatric illness on the association between smoking-cessation treatments and neuropsychiatric outcomes.