It's a classic risk/benefit dilemma: Does preventing suicidality--assuming suicidality can predict suicide--justify scaring some doctors away from prescribing antidepressants for young patients?
In September 2004, when the FDA approved the addition of a "black box" warning on selective serotonin reuptake inhibitors (SSRIs) indicating increased risk of suicidality among children and adolescents, the agency was acting on the recommendations of 15 of 23 members of a joint advisory committee. Analysis, reanalysis, and re-reanalysis of a smorgasbord of studies, conducted originally to establish antidepressant efficacy, coupled with testimony from parents, fueled the recommendations and decision.
What the circuitous investigation boils down to is a 3.8% risk of suicidality at the beginning of treatment with an SSRI, compared with a 2.1% risk with placebo. Some of the data on this are available at www.fda.gov/cder/ pediatric/summaryreview.htm. The total test population for the 5 antidepressants described was 1445. Thomas Newman, a member of the FDA advisory committee, said that the probability that the increased risk of suicidality among drug takers would have occurred by chance is 1 in 20,000.
The data the FDA considered in first issuing a warning and then approving a black box label focused on psychiatric indications. "Of the 24 studies we evaluated, 16 were for depression, one for social phobia, and the others for obsessivecompulsive disorder," said Robert Temple, director of the office of medical policy at the FDA's Center for Drug Evaluation and Research.
But these cautions are hardly news to neurologists and primary care physicians. "We've been aware since the beginning of time that when you are treating mood disorders that need antidepressants, you may see the emergence of more apparent suicide ideation as the patient proceeds through treatment," said Eric Caine, MD, chair of the Department of Psychiatry at the University of Rochester Medical Center in Rochester, NY.
Neurologists prescribe SSRIs for depression, sometimes in children. "A lot of parents of children with behavioral disorders go to a child neurologist early on to distinguish between a developmental or psychiatric condition," said Richard Gorman, MD, a pediatrician in private practice in the Baltimore area who represented the American Academy of Pediatrics at the September FDA meeting. The result of such a consult may be an SSRI prescription, although fluoxetine (Prozac, Eli Lilly) is the only one approved to treat depression in children. Neurologists also prescribe SSRIs to help patients cope with other disorders. "SSRIs are used often in patients with migraine, Parkinson's-associated depression, and early Alzheimer disease," said Susan Olson, MD, president of the American Academy of Neurology.
But the elevated suicidality risk might not apply to conditions other than major depressive disorder (MDD). "One of the ideas why treating depression is linked to suicide is that it [involves] people who are immobilized by the disease, and the drug allows them to move. If that's an important mechanism, you might not expect that to matter much in neurological disease. But we don't know--and won't--until studies of reasonable size are done and we can pool them," said Temple.
Although all Selective Serotonin Reuptake Inhibitors (SSRIs) presumably work by increasing availability of serotonin, their pharmacokinetics differ, and therefore they are not created exactly equally. "Receptor actions are potentially different from one drug to another. In clinical practice we use the side-effect profile to decide which drug to prescribe," said Charles Glatt, MD, assistant professor in residence, UCLA Neuropsychiatric Institute.
Because transporters, receptors, and the enzymes that oversee neurotransmitter synthesis and breakdown are proteins, researchers are seeking gene variants that might predict patient responses. This pharmacogenetic approach is 2-tiered: a geneby- gene exploration of known candidates and gene expression profiles that reveal the many protein voices that contribute to mood and its disturbance. It is a little like isolating the contributions of individual musical instruments, compared with considering the overall effect of an orchestra.
The gene-by-gene approach has yielded some very preliminary hits and misses (Table). However, the situation is complex. Variants of the same gene can have opposite effects on drug response, and the same gene variant can have different meanings against the differing genetic backgrounds among human populations.
The gene expression approach is like seeking any biomarker, noting which genes show hiked expression (protein production) in responders compared with nonresponders. "You take medicated depressed patients, and see which people respond, and correlate that to a biomarker. It is a simple idea, but hard to do. This type of study takes a really large sample size to see an effect-- hundreds of people," said Glatt. "Using gene expression profiling to understand SSRI response is in its infancy," added Ma-Li Wong, MD, a professor of psychiatry and biobehavioral sciences at the Neuropsychiatric Institute at UCLA who studies tricyclic antidepressants and St. John's wort.
While tracking genes to predict drug response for mood disorders is a technology left to the future, other biomarkers may be useful today. Glatt considers presence of akathisia to be a potential predictor of suicidality associated with antidepressant use. This is a state of extreme restlessness and agitation that is a common response to SSRIs. "A lot of people who take psychedelic [drugs] experience it as a 'body buzz.' In depression, it is a new, uncomfortable experience that could exacerbate the condition in the initial period of treatment," he said. Akathisia is a primary reason why people give up on their SSRI medication, he added. "There is an increased risk of suicidal behavior on an SSRI. Why it's related to akathisia as a side effect could be a focus of pharmacogenetics."
SUICIDE AND SUICIDALITY AMONG THE YOUNG
A generation ago, depression wasn't a label typically applied to the young. It became an official risk factor for suicide among teenagers in 1988.1 Teen suicide rates have plummeted in recent years, paralleling the rise in use of SSRIs, although causation hasn't been established. Because older antidepressants, such as the tricyclics, cause severe side effects in some young people, they have not been routinely prescribed for this age group.
The recent warnings about SSRIs, however, refer to suicidality, not suicide. "Suicidality is the sum of thinking about suicide a lot and making gestures or preparations, such as lining up a rope. Suicidality is an adverse reaction, but it doesn't kill you," explained Temple. No suicide deaths were reported in the studies that the FDA reviewed. Caine concurred that the distinction is significant. "A huge number of people have suicidal ideas. A smaller number attempt suicide, and a much smaller number do it." Today in the United States, fewer than 1% of people aged 15 to 19 commit suicide, although 19% experience suicidal ideation, and 9% make a suicide attempt, according to the American College of Neuropsychopharmacology's Task Force on SSRIs and Suicidal Behavior in Youth.
It is difficult to study a phenomenon that is plagued by fuzzy definitions and that may represent a continuum, rather than the 2 distinct entities of suicidality and suicide. "People with real suicidality make some gesture; they do something about it. That is not incredibly common in major depressive disorder. Those who make severe suicide attempts may have a different biological form of depression," said Charles Glatt, MD, an assistant professor in residence at the UCLA Neuropsychiatric Institute. "Another problem in studying suicidality is its rarity. To see a reliable result takes a huge number of people," he added.
COUNTDOWN TO THE BLACK BOX
Establishing an association between antidepressant use and suicidality is fraught with methodological challenges. The consulted
|Venlafaxine||Effexor||Wyeth Pharmaceuticals||Serotonin, noradrenaline|
|Nefazodone||Serzone||Bristol-Myers Squibb||Serotonin, noradrenaline|
|Bupropion||Wellbutrin||GlaxoSmithKline||Serotonin, noradrenaline, dopamine|
|Duloxetine||Cymbalta||Eli Lilly||Serotonin, noradrenaline|
studies evaluated different drugs in different populations, sometimes with different designs and criteria, for a problem considered a symptom, not an adverse event resulting from treatment. Layer onto that the fact that proprietary data from pharmaceutical companies were not publicly available, and a very complex picture emerges. Tracing the story as it unfolded reveals conservative conclusions based on confusing evidence.
The antidepressant-suicidality question emerged with the birth of the SSRIs in the late 1980s. "There's been a debate since the early days of Prozac, with case reports of committing suicide. It was a hot issue then, and it went away. It was thought to be an artifact because it was rare," said Glatt.
The FDA approved fluoxetine as the first SSRI for treatment of MDD in 1987. In 1990, Martin Teischler, MD, a psychiatrist from Harvard Medical School, headed a study of 6 fluoxetine users who had committed suicide. Following media coverage, other cases were reported to the FDA. After Teischler explained possible contributing factors to suicidality at an FDA meeting in September 1991, the agency concluded that the drug could not be linked to increased risk. In 1993, a task force from the American College of Neuropsychopharmacology reevaluated case reports and also denied a link.
Fast forward a decade. In June 2003, reviewers of new drug applications in the Division of Neuropharmacological Drug Products at the FDA noted a few studies of paroxetine (Paxil, GlaxoSmith- Kline) that reported the adverse effect of "emotional lability" in children. Uncertain whether this characteristic contributes to suicidality, the agency asked Glaxo- SmithKline to redefine the term, which led the company to reevaluate study results. GlaxoSmithKline investigators searched reports on the 6 trials for key words--such as "cut," "overdose," "hung" and, of course, "suicide"--describing events occurring within 30 days of the last dose. When the searches were unblinded, evaluators discerned a "signal of increased risk of suicidality for Paxil," according to the company's report to the FDA.
On June 6, 2003, the Medicines and Healthcare Regulatory Agency (MHRA) in the United Kingdom released a public advisory stating that paroxetine should not be used to treat depression in people younger than 18 years. The United States took a bit longer, recommending on June 19 that paroxetine not be used to treat depression in children and teenagers. Then the FDA requested similar scrutiny of the efficacy trials for 8 other antidepressants and on July 22 added fluoxetine (Prozac, Eli Lilly), sertraline (Zoloft, Pfizer), fluvoxamine (Luvox, Solvay Pharmaceuticals), citalopram (Celexa, Forest Pharmaceuticals), bupropion (Wellbutrin, GlaxoSmithKline), venlafaxine (Effexor, Wyeth Pharmaceuticals), nefazodone (Serzone, Bristol-Myers Squibb), and mirtazapine (Remeron, Organon) to the list of drugs that could increase risk of suicidality in the young. At the same time, the agency rechecked all adverse event reports.
In August 2003, Wyeth Pharmaceuticals changed the label on venlafaxine to include disclosure about increased reports of hostility and suicidality, and in September, the MHRA issued a public advisory. The FDA held back, awaiting further data analysis, and on September 16, 2003, at an internal briefing, decided that a third party was needed to make sense of the various studies. The task fell to a group at Columbia University that already had a method to identify and classify suicidality. "We consulted the Columbia researchers because we were seeking reassurance that these cases looked to experts like they did to us," explained Temple. The Columbia analysis found problems too. In one study, for example, a young girl slapping herself in the face was interpreted as a suicide attempt.
In December 2003, the United Kingdom entered the picture again--this time issuing public advisories against use of all SSRIs and related drugs, except fluoxetine (its longer half-life is thought to account for its greater efficacy), for persons with depression who were younger than 18. By February 2004, the FDA was inching closer to the MHRA's level of caution. At an internal meeting, medical officer Andrew Mosholder argued for the suicidality link, but his statements weren't publicized for fear of alarming patients. By March, the agency relented, issuing a Public Health Advisory listing symptoms that physicians and caregivers should watch for in persons taking the drugs listed above, plus escitalopram (Lexapro, Forest Pharmaceuticals).
The Columbia researchers submitted their findings to the FDA in July 2004. Then, the FDA analyzed those data and posted the "new" conclusions on the agency Web site.
On August 30, 2004, members of the the FDA's Psychopharmacologic Drugs and Pediatric Advisory Committees heard testimony from parents of affected children. On September 8, Janet Woodcock, MD, acting deputy commissioner for operations, explained the law of "pediatric exclusivity," which became part of the Best Pharmaceuticals for Children Act in 2002. It extends a company's exclusivity of a drug for 6 months if the company conducts at least 2 efficacy trials on children, whether or not the drug is eventually approved for them. But if a company has pediatric data that it does not submit, those results do not have to be made public. Although pediatric exclusivity is intended to encourage research on the effects of drugs in children, it may have the effect of hiding negative data from analysis, reanalysis, or meta-analysis.
Despite the shortcomings of the data that the FDA evaluated, the 2 committees voted 15 to 8 on September 16 to recommend the black box warning for all 10 drugs, even those for which pediatric results were not available. However, in recognition of their great value, the drugs are not contraindicated in pediatric patients.
"Many physicians and mental health patient advocates are concerned that the result of the black box warning will be . . . a vast reduction in appropriate use," said Moory Smulevita, a global products communication officer at Eli Lilly. The FDA shares this concern regarding the labeling, a work still in progress. "We are trying to frame the language so we do not scare away people who need these drugs," said Temple. Although drug companies' spokespersons claim that they do not track sales, several who were interviewed reported that sales had not declined since September.
Clinicians debate whether depression among patients with Parkinson disease (PD) is a distinct disorder or part of the pathology. Whatever the designation, the comorbidity is common-- the National Institute of Mental Health estimates that half of people with PD are also depressed.
"Many patients assume that it's normal to feel this way. That's not true. If you treat the depression, they'll still have the other symptoms of the disease, but they feel better. It's very treatable," said Irene Richard, MD, a neurologist at the University of Rochester in New York. Depression is not as common with some other disturbing diagnoses, such as rheumatoid arthritis, she added. Richard heads a 15-center clinical trial comparing paroxetine with venlafaxine in alleviating depression in patients with PD.
Matthew Menza, MD, an assistant professor of psychiatry at the Robert Wood Johnson Medical School, New Brunswick, NJ, agreed. "Major depression is diagnosed in patients with PD as one would diagnose it in any other patient, but it is more common than one would expect from the degree of disability. This makes us think that it has something to do with the neurodegeneration of the illness." He heads a clinical trial comparing the tricyclic antidepressant nortriptyline with paroxetine to treat Parkinson-related depression. "Most physicians use SSRIs, but older antidepressants may be just as good or better. For other disorders, such as stroke, tricyclics work better. So that gives us reason to wonder what will work best in PD," he added.
Depression in patients with PD manifests somewhat differently than in major depressive disorder. Anxiety is more pronounced and common, and suicidality is higher, but actual suicide rates are lower. In addition, the characteristic sadness is not associated with guilt. Symptoms and drug response could be affected by the differences in brain chemistry that are at the core of the condition. Although PD primarily disturbs dopamine metabolism, serotonin and noradrenaline levels are affected too.
Richard suggests that physicians treating patients with PD acknowledge the depression. "We've found that if a physician brings up the topic, patients will be honest and will discuss their depression, but oftentimes they won't bring it up themselves."
LIMITATIONS OF THE STUDIES
Pooling apples and oranges provides a different picture than examining 1 piece of fruit in detail. Consider the Treatment of Adolescents With Major Depression Study (TADS)2,3 and other studies. The 13-center TADS study assigned 429 persons aged 12 to 17 with moderate to severe MDD to 4 groups: fluoxetine (blinded), placebo, cognitive-behavioral therapy (CBT; not blinded), and fluoxetine and CBT (not blinded).
TADS produced clear results. Suicidal ideation fell in all groups, but fell to the greatest degree among those treated with both drug therapy and CBT, followed by drug therapy alone, and then CBT alone. A "suicide-related event" occurred in 6% of the total population, and incidence was distributed evenly among all 4 groups, with no events resulting in death. However, the protocol excluded persons who had a recent history of a suicide attempt or lack of a support system to prevent future attempts.
At about the same time as the MHRA and the FDA were evaluating evidence, so was the task force from the American College of Neuropsychopharmacology, although members admit that they were not privy to all industrial data. Their executive summary concluded: "The evidence from case reports linking SSRIs to suicidal behavior is weak. The most likely explanation for cases of suicide or attempted suicide while taking SSRIs is that the underlying depression is responsible, not the SSRIs." However, selective reporting of trial results for the SSRIs altered the results. When Craig Whittington, MD, and his team at the Centre for Outcomes Research and Effectiveness at University College London included unpublished trial results in their analysis, increased suicidality risk emerged for several SSRIs, with the exception of fluoxetine.4 The researchers got the information from the MHRA's Committee on Safety of Medicines.
The very nature of depression also contributes to the inconsistencies seen when comparing clinical trial results. Many investigations briefly follow a condition that typically comes and goes over the course of a lifetime. Further, participation in a trial may elicit a placebo effect. "The warm, nurturing environment of the trial is a little like therapy, and depression is cyclical. By the time a person has been observed for 3 to 4 weeks, maybe he or she is starting to get better," said Temple. Depression is even harder to assess in children, he added, because some may actually have bipolar disorder, and history taking sometimes is not as complete as in adults. ■
1. Brent DA. Antidepressants and pediatric depression-- the risk of doing nothing. N Engl J Med. 2004;351:1598-1601.
2. Glass RM. Treatment of adolescents with major depression. JAMA. 2004;292:861-863.
3. March J, Silva S, Petrycki S, et al. Fluoxetine, cognitive-behavioral therapy, and their combination for adolescents with depression: Treatment for Adolescents With Depression Study (TADS) randomized controlled trial. JAMA. 2004;292:807- 820.
4. Whittington CJ, Kendall T, Fonagy P, et al. Selective serotonin reuptake inhibitors in childhood depression: systematic review of published versus unpublished data. Lancet. 2004;363:1341-1345.