Should RIMONABANT Get a Second Chance?
One top researcher who advised FDA thinks so—and now all obesity drug trials are using her method for measuring suicide risk
BY RON FEEMSTER
Kelly Posner is an FDA troubleshooter for neuropsychiatric data in clinical trials. A child psychologist and research scientist
at Columbia University, she gets called when the regulatory agency sees signs of psychiatric risk—particularly suicidality—and
needs to make sense out of jumbled trial data. To develop interpretable data from inconsistent reports of adverse events in
pediatric anti-depressant studies, she and her colleagues built the Columbia Classification Algorithm of Suicide Assessment
(C-CASA). Her team worked through the data sets, classifying each incident as suicidal ideation, suicidal behavior, a bona
fide attempt, or a false positive. In the end, they counted half as many actual suicide attempts as the sponsors' own doctors.
FDA called again on the director of the Center for Suicide Risk Assessment when it saw potential risks of suicidal ideation
with rimonabant. Posner's group considered a total of 1,201 "patient narratives" from seven rimonabant trials. Using C-CASA,
they classified 91 cases as either "possibly" or "definitely" suicidal, but eliminated some because they occurred in study
arms without placebo control. The final tally of suicidality cases was 74, with 20 on placebo, 8 on rimonabant 5 mg, and 46
on rimonabant 20 mg; the overall drug-to-placebo ratio was 1.8 to 1. Suicidality was nearly twice as likely in the drug group
based on data collected through self-reported adverse events.
Such retrospective studies based on adverse-event reporting could be misleading, Posner contends, because patients in the
drug group typically spend more time talking to doctors and have more opportunity to self-report suicidal events than placebo
patients. These retrospective studies would be unnecessary if FDA requested prospective monitoring of placebo-controlled trials
using standardized language and a universal scale, she argues.
She may have made her case. In June 2006, when Posner presented her findings to the rimonabant advisory committee, she suggested
using the Columbia-Suicide Severity Rating Scale, a data-collection tool she used in a National Institute of Mental Health
study. "Assessing suicide systematically with a scale is pretty much all upside," Posner says. "Why would you want to set
yourself up for a false signal or a false risk or a misleading result by relying on adverse-event reporting?"
Starting in late 2007, FDA began requiring that drugmakers testing specific compounds in a wide range of diseases use the
scale in their trials. The instrument has since been translated into 90 languages. "C-SSRS is requested for all obesity trials
now," Posner said. "Companies need to understand that it's protective to their medications and their data. When they systematically
monitor, they haven't see any risk [of suicide]."
By implementing a prospective and systematic measurement of suicidality, companies can help clarify the causal relationship,
if any, between suicidal adverse events and medications. "Association doesn't mean causality," Posner says. "Decreasing false
positives and debunking false notions of risk are as important as knowing what risks exist. That gives a fair chance to very
Posner acknowledges that CNS-based obesity drugs may have unexpected psychiatric side effects because they affect more than
one pathway. But the public-health upside of intervening in obesity is so great that she recommends proceeding with drug trials
and carefully monitoring outcomes. Even if an effective obesity medication were found to have a risk of suicidal ideation,
Posner says, depression and suicidality are treatable conditions. "Everything is about balancing risks and benefits," she
Posner, who says that she has never taken a dime in compensation from a drug company or FDA, has worked closely with the agency
to come up with simple, low-burden methods to assess neuropsychiatric risks and benefits. "If you had listened to the discussion
about rimonabant, you might have heard about 50 neuropsychiatric terms thrown around," Posner says. "What do we actually need
to assess? In obesity drugs, it's boiled down at this point to the C-SSRS and the PHQ-9." Patient Health Questionnaire 9 is
a nine-question self-report scale for monitoring signs of depression. Most patients can fill it out in the waiting room and
score it in 10 seconds. C-SSRS can also be taken as a self-reported telephone interview known as IVR (interactive voice response).
Things might have been different for rimonabant if the Sanofi studies had used Posner's prospective method of measuring suicidality.
"The sponsor pulled the drug, not FDA," Posner says. "Rimonabant and the other endocannabinoids are an area with tremendous
potential that, because of a lot of different factors, probably hasn't had its fair chance. I definitely think it is not a
space that should be given up on."