FROM AN FDA ADVISORY PANEL MEETING
The majority of attendees at a joint meeting of two Food and Drug Administration advisory committees voted Sept. 14 to eliminate the black box warning about neuropsychiatric risks for the smoking cessation drug varenicline.
Panelists weighed the need for greater adoption of effective smoking cessation interventions against the possibility that varenicline (Chantix) might increase the risk of such serious adverse events (AEs) as aggression and suicidal behavior. In assessing findings of a clinical trial designed to sort out neuropsychiatric risk, they also had to wade through differing interpretations of the clinical trial data presented by the drug’s sponsor and the FDA.
Members of the Psychopharmacologic Drugs Advisory and the Drug Safety and Risk Management committees heard Chantix’s sponsor, Pfizer, present results of the postmarketing EAGLES trial. The global multisite trial compared varenicline with bupropion, also approved for smoking cessation, as well as with nicotine replacement therapy and placebo.
Of the 19 voting advisory committee members, 10 voted in favor of eliminating the boxed warning, while 4 members voted for a modification of the existing boxed warning, and 5 voted to retain the warning. The vote represents a reversal for the committees. In 2014, the joint panels recommended retaining the black box warning on the drug. The FDA usually follows the recommendations of its advisory panels, but the recommendations are not binding.
The boxed warning, which advises prescribers of the potential for “serious neuropsychiatric events,” arose from individual postmarketing reports of agitation, hostility, depressed mood, and suicidal ideation and behavior. Though postmarketing observational studies had not shown a consistent signal for increased risk of neuropsychiatric events, a plausible psychopharmacologic mechanism existed for the reports, and a clinical trial examining neuropsychiatric safety was designed to settle the question.
The primary objectives of EAGLES were to assess the risk of “clinically significant” AEs for individuals using varenicline, bupropion, nicotine replacement, or placebo, and to determine whether a past history of psychiatric disorders increased the risk of neuropsychiatric AEs when either prescription medication was used for smoking cessation.
A total of 8,000 patients were randomized 1:1:1:1 to receive varenicline, bupropion, nicotine replacement via patch, or placebo. The double blind, triple dummy design meant that all patients took pills and applied a patch daily. Each study arm was evenly divided so that half of the participants had a prior psychiatric diagnosis of a mood, anxiety, psychotic, or personality disorder. Overall, 70.7% of the psychiatric cohort had a mood disorder, and just under half were receiving a psychotropic medication at baseline.
The other half of patients in each study arm had no current or past psychiatric diagnoses. Approximately 80% of participants completed the trial overall.
The FDA and Pfizer worked together to develop a composite neuropsychiatric safety endpoint that broadly included the postmarketing AEs that had been reported for varenicline. The composite was designed to increase sensitivity, but only “moderate” or “severe” events were included in the analysis to increase specificity by minimizing inclusion of symptoms that might be attributable to nicotine withdrawal.
Results were similar across treatment arms for the primary composite endpoint, with higher incidence of events for the psychiatric cohort regardless of treatment, according to Pfizer’s analysis of the data. Since the study was not designed with a specific hypothesis, the results were not expressed in terms of statistical significance, but rather in risk differences with accompanying confidence intervals (CIs). The confidence interval crossed zero for all study drugs (including nicotine), except when varenicline was compared with placebo in the nonpsychiatric cohort (risk difference, 1.28; 95% CI, –2.40 to –0.15).
Overall, Pfizer found that about 2% of smokers without mental illness experienced neuropsychiatric AEs, compared with 5%-7% in the psychiatric cohort, regardless of treatment arm.
The FDA also found a small decrease in AEs for varenicline when compared with placebo in those without mental illness, and a higher incidence of neuropsychiatric events for those patients with mental illness who took varenicline and bupropion than those on placebo. In the FDA’s analysis, 6.5% of those on varenicline and 6.7% of those on bupropion met the primary composite AE endpoint. However, the FDA found that severe neuropsychiatric events and scoring on the Columbia Suicide Severity Rating Scale ( C-SSRS ), though lower for participants without mental illness, were similar across treatment arms in both cohorts.
Reporting problems also complicated the panel’s task. As the FDA conceived it, said Celia Winchell, MD, each AE was to be accompanied by a narrative that captured not just test scores and coded clinician assessments, but also the verbatim patient report and any other relevant data, such as the police report of a traffic accident, or family members’ appraisals of the event. Instead, she said, many of the reports appeared to have been automatically generated and were lacking in detail. Compared with other studies she’s reviewed, “The quality of the narratives that were submitted were unusually uninformative,” said Dr. Winchell, clinical team leader in the FDA’s Division of Anesthesia, Analgesia, and Addiction Products ( DAAAP ).
Clinical trials were run at 139 sites in 16 countries, treating a total of 8,058 patients. Despite Pfizer’s extensive efforts to standardize adverse event categorization and reporting, the FDA and some panelists remained concerned that language difficulty and cultural variation in interpreting neuropsychiatric symptoms may have resulted in miscategorization or underreporting of such events. In addition, Pfizer’s own surveillance revealed data reporting problems at two overseas sites that resulted in those sites’ data being eliminated from analysis.
All of the trial data, together with a review of several observational studies, had to be weighed against the individual and public health risks of smoking. Individuals with serious mental illness are significantly more likely to smoke heavily and to relapse after quitting compared with the general population. Yet, those individuals are less likely to be prescribed medication to help stop smoking, Anne Eden Evins, MD , director of the Center for Addiction Medicine at Massachusetts General Hospital, Boston, said in her testimony on Pfizer’s behalf.
“The implication of EAGLES is that we can offer treatment to all smokers, including those with stable mental illness,” she said. “It is imperative we find ways to increase use of the most effective smoking cessation treatment for our patients who try time and again to quit smoking.”
Many committee members agreed, providing anecdotal evidence that the current boxed warning deters both prescribers and patients from considering varenicline for smoking cessation in patients with mental illness.
For David Pickar, MD , who voted to remove the warning, the decision was easy. “I have no ambivalence, whatsoever,” he said. “The risk-benefit ratio to me is as clear as anything I have seen. I have never, ever, ever seen a schizophrenia patient on this drug,” said Dr. Pickar, a psychiatrist affiliated with the Uniformed Services University of the Health Sciences, Bethesda, Md.
However, many panelists were disturbed by the variability seen between sites during the clinical trials and by the lack of depth in Pfizer’s adverse events reporting. Said Rajesh Narendran, MD , professor of radiology and psychiatry at the University of Pittsburgh: “I don’t think the trial was conducted with the same cleanliness and elegance” of a trial for a new drug application. “I wasn’t fully reassured, and I think that removing the black box does send the wrong message that this is now a safe drug,” he said.
Although the joint committees’ chair, Ruth Murphey Parker, MD , voted for modification of the boxed warning, she agreed. “I also felt that removal of the black box would be a signal for safety,” she said.
Others who voted for modification felt that prescribers should talk to their patients about risks and benefits. Said Christianne Roumie, MD , professor of internal medicine and pediatrics at Vanderbilt University, Nashville, Tenn., and a staff physician at Veterans Affairs Tennessee Valley Healthcare System, also in Nashville: “As a clinician, I always have the discussion. This should be a conversation and not … carte blanche.”
On Twitter @karioakes
