Soon after the quit-smoking pill Chantix debuted in 2006, the U.S. Food and Drug Administration began receiving reports of severe psychiatric disturbances in people taking it.
Experts who analyzed the FDA's database of "serious adverse events" found Chantix was suspected in more cases of violence than any other prescription drug. It also generated more reports of suicide, self-injury, and depression than any other smoking-cessation therapy.
Despite this troubling link - which has led the FDA to require prominent warnings on the Chantix package insert - researchers at the University of Pennsylvania and other medical centers are now testing it in people who may be especially vulnerable to such problems. These include alcohol, cocaine, and methamphetamine addicts.
Testing is warranted, experts say, because Chantix is generally safe; millions worldwide have used it with no big complaints. And some addictions are so devastating that the benefits of a medication that promotes abstinence would far outweigh the risks of possible side effects.
No drugs have been approved to treat cocaine or meth addiction, despite years of research.
At the National Institute on Alcohol Abuse and Alcoholism, which is also testing Chantix, principal investigator Joanne Fertig said, "Alcoholism is a life-threatening disease. I think it's imperative for us to keep looking and screening medications that are out there."
Still, scientists conducting these pilot trials - mostly funded with federal grants - realize that using Chantix in high-risk patients is ethically and medically dicey.
As a precaution, the studies exclude addicts with suicidal tendencies or a mental illness such as bipolar disorder. However, many smokers who reportedly developed psychiatric problems while on Chantix had no history of it.
At the University of California, Los Angeles, where Chantix is being tried in meth users, physician Keith Heinzerling said, "It's a delicate balance between the ethics of exposing volunteers and finding answers to questions we need to answer."
Researchers are encouraged that in about a dozen studies under way or finished, patients have had no major psychiatric events - so far, anyway.
Chantix, also known as varenicline, curbs the compulsion to smoke by binding to certain nicotinic receptors in the brain, blocking nicotine inhaled from cigarettes. Chantix also partly activates the receptors, stimulating release of dopamine, the "pleasure chemical." The dopamine surge is less than with nicotine, but enough to ease withdrawal and cravings.
Alcohol and other addictive substances also increase dopamine, partly through nicotinic receptors, so Chantix may be able to weaken non-nicotine addictions. Studies in animals support this theory, although it's too soon to tell in humans.
Taken for 12 weeks, or sometimes 24, Chantix is no magic bullet. In a major study by the manufacturer, Pfizer Inc., Chantix worked better than the quit-smoking medicine bupropion (Zyban), an antidepressant - but only through 24 weeks. At a year, 28 percent of Chantix users were smoke-free compared with 23 percent with bupropion - a difference that could have been by chance. Among placebo-takers, 14 percent were abstinent at one year.
In the six trials that got Chantix approved in May 2006, no major psychiatric events were reported. Common complaints were nausea, headache, and vivid dreams.
But those studies did not mimic the real world. Not only were the 3,600 volunteers given intensive support, they were unusually healthy. Medical or mental illnesses that might complicate interpretation of the results were excluded. While that's standard in drug development, it meant Chantix was untested in people with seizures, depression, panic disorder, or drug or alcohol dependence, much less psychosis or bipolar disorder.
After approval, Chantix became a blockbuster, with $700 million in sales in 2007.
Soon, the FDA began receiving reports from consumers, doctors, and Pfizer itself of serious adverse events.
Pfizer points out that such reports are anecdotal, maybe unreliable, and don't prove Chantix - taken by over seven million Americans - was the cause.
Still, such anecdotes - typically a fraction of the actual number - are often the only way to detect side effects that are too rare to show up until a drug is widely used.
In 2008, Thomas J. Moore, a researcher with the Institute for Safe Medication Practices in Horsham, decided to see which drugs had the most adverse-event reports in the last quarter of 2007. He was trying new software, not looking for what he found - a smoking gun.
"Chantix was off the chart," Moore recalled.
Moore and Curt Furberg, a Wake Forest University researcher, then collaborated on a series of closer looks. They found Chantix was associated with seizures, visual disturbances, heart failure, swelling, and life-threatening skin conditions, as well as psychiatric emergencies.
They examined 26 out of 400-plus cases of Chantix-linked violence looking for patterns. The aggression was typically unprovoked, unleashed on anyone nearby. A woman, 24, beat her boyfriend in bed because he "looked so peaceful." A father, 47, "came out of a room," hit his daughters, then fatally shot himself.
The fallout from these analyses was considerable.
By 2009, Chantix's package insert carried highlighted warnings about "serious neuropsychiatric events." The Department of Veterans Affairs restricted the drug to veterans who have failed with other quit-smoking aids. Federal authorities banned it for pilots, air controllers, and missile crews.
Prescription data from abroad bolstered concerns - but there were also reviews affirming Chantix's safety.
Last month, Moore and Furberg published a paper focused on suicidal behavior, self-injury, and depression for all smoking-cessation therapies over the last 13 years.
Chantix was wildly disproportionate, accounting for 90 percent of the 3,249 reports, even though it was on the market for only about four of those 13 years. Bupropion had 229 reports, and nicotine replacement had 95.
Included were 272 suicides linked to Chantix - well over one a week on average. Bupropion had 19 suicides; nicotine replacement, four.
Although smoking is the top cause of preventable death in the United States, the authors concluded that Chantix should not be a first-choice therapy because it often fails, safer options exist, and the worst risks - however remote - are grave: "Violent and suicidal behavior can have immediate, catastrophic and irreversible effects."
What about using Chantix in other addictions?
That, Furberg said, could change the balance of risks and benefits.
"If we have no good treatments for other types of addictions, let's look at Chantix in an unbiased way," he said. "In the field of cancer, we take almost any side effects and say, 'That's OK, let's use the therapy because the benefit is so critical.' "
At the FDA's behest, Pfizer is now doing a large Chantix trial that includes smokers with psychiatric disorders. The results, expected in 2017, should at last quantify risks.
The company would not say whether it would like to get Chantix approved for non-nicotine addictions. However, Pfizer is sponsoring a German study of Chantix in alcoholics, and it has provided its product for some medical center trials.
"Our goal is to turn the data [from early trials] over to Pfizer," said Fertig, the U.S. government researcher. "The company could decide whether they wanted to go forward with a larger" definitive trial.
So far, the pilot studies - in which some or all of the patients also smoked - have produced mixed signals.
Among cocaine users, for example, a Yale University study found Chantix was no better than a placebo, while Penn found it was better.
For alcohol abuse, early results were also conflicting.
The Mayo Clinic is now testing Chantix in heavy-drinking smokers, hoping it curbs both habits.
At UCLA, Chantix reduced meth use in 20 carefully selected volunteers.
"To see something in even that small a number is impressive," Heinzerling said. "And there were no serious adverse events."
If there were, sorting cause from effect could be tricky. Meth addiction can cause hallucinations, aggression, suicidal behavior, paranoia - all also linked to Chantix.
In any addiction, psychiatric symptoms such as depression may be triggered or worsened by substance abuse - or by quitting the substance, with or without therapy.
"These populations are more complicated to begin with," Yale psychiatrist Stephanie O'Malley said. "And the studies are too small to rule out a rare serious adverse event. Does that mean we shouldn't do a study? I don't think so."
If Chantix does show clear promise, even that could create problems. Addicts desperate for therapy - and doctors eager to provide it - may rush to use Chantix for an "off-label," meaning unapproved, purpose.
"The first step is seeing if [Chantix] is safe and effective in a highly selected group," Heinzerling said. "I try to make this clear to doctors thinking of using it based on early clinical trial results."