The Impact of Antiepileptic Drugs on Suicide Risk: An Expert Interview With Andres Kanner, MD

Anne Roc, PhD


January 26, 2009

Editor's Note:

On December 16, 2008, the US Food and Drug Administration (FDA) issued an alert that manufacturers of antiepileptic drugs (AEDs) used to treat epilepsy, psychiatric disorders, and other conditions include a warning in their labeling informing patients about the risk for suicidal thoughts or behaviors (suicidality). The alert was based on the agency's review of 11 AEDs across 199 clinical trials, which showed that patients receiving AEDs had nearly twice the risk for suicidality (0.43%) compared with patients receiving placebo (0.24%). On December 5, 2008, at the 62nd Annual Meeting of the American Epilepsy Society (AES), a panel of epilepsy experts questioned the methodology of the FDA findings and discussed the potential effect on epilepsy management.

Anne Roc, PhD, Scientific Director for Medscape Neurology & Neurosurgery, recently interviewed Andres M. Kanner, MD, to get his comments about the FDA alert. Dr. Kanner is Professor of Neurological Sciences at Rush Medical College in Chicago. He is chairperson of the Epilepsy Section of the American Academy of Neurology (AAN) and the AAN committee on Clinical Guidelines and was the chair of the task force on Suicidality and Epilepsy at AES. Before Dr Kanner addressed the issues specifically related to the FDA findings and the methodologic issues raised by the AES, he provided Medscape with an introduction to the complex relationship between epilepsy and suicidality.

Dr Kanner: In essence, we know that patients with epilepsy have an increased risk for mood disorders and anxiety disorders compared with the general population, and we know that people with epilepsy have a higher risk for suicidality. The concept of suicidality is a term that encompasses completed suicide, suicide attempt, and suicide ideation. Population-based studies, including a study from Denmark published in 2007,[1] have shown that patients with epilepsy, when compared with the general population and after controlling for any psychiatric and psychosocial risk factors, have a twofold greater risk of committing suicide. If you factor in comorbid psychiatric history and, particularly, a history of mood disorders such as major depressive disorder, the risk of committing suicide increases 32-fold. This risk increases 12-fold with a history of anxiety disorders and 11-fold with schizophrenia. That gives you an idea of the relative risk of the general population.

Studies have looked at the percentage of deaths by suicide and have found, depending on the study, a range from 0% to 67%. One review article published a few years ago looked at 21 published studies and showed that the mean percentage of deaths attributed to suicide was 11.5%.[2]

It is important to understand that the relationship between suicidality and epilepsy is bidirectional. Not only can patients with epilepsy have an increased risk for suicidality, but patients with a history of suicidality also have an increased risk for the development of epilepsy. One population-based study in Iceland showed that people with a history of suicidality have a fivefold greater risk for epilepsy.[3] There is also evidence that people with a history of depression have a fourfold to sevenfold greater risk for epilepsy developing.

What that tells you is that there are common pathogenic mechanisms in epilepsy, mood disorders, anxiety disorders, and suicide. One of those pathogenic mechanisms is disturbance in the neurotransmitter serotonin. Multiple studies have documented decreased serotonin activity involving the frontal lobe structures of the brains of patients with major depressive disorders who committed suicide. Studies from animal models of epilepsy have shown the pivotal role of serotonin as a pathogenic factor for epilepsy. Studies of patients with epilepsy using positron emission tomography (PET) imaging have also shown decreased binding of one of the types of serotonin receptor in the temporal lobe, the frontal lobes, and the brainstem, particularly in the raphe nuclei, which lodges the neurons that synthesize serotonin. I am just talking about one common pathogenic mechanism; obviously, the story is more complicated than this.

Having comorbid psychiatric history, particularly a history of major depression or anxiety, is an important risk factor. Recent studies have shown that in patients without epilepsy but with a history of both depression and anxiety, the risk for suicidality developing increases. We were part of a study a few years ago that showed that the occurrence of depression and anxiety was associated with increased suicidal ideation.[2]

The biggest red flag for suicidal risk is a previous suicide attempt; people who have had previous suicide attempts need to be followed very closely because they have a significantly higher risk of attempting suicide again. In a study that was done in people with epilepsy, those who had a previous suicide attempt were at 38.4% greater risk for a completed suicide.[4] These data were replicated in a Swedish study.[5] This is also something that has been seen in people without epilepsy.

As I mentioned, another very important risk factor for suicidality is a family history of psychiatric disorders. There is evidence in the psychiatric literature that if you have a family history of mood disorders you have a significantly greater risk of having a mood disorder compared with the general population, and if there is a family history of suicidality, you are also at increased risk for suicidal behavior.

In patients with epilepsy, you have what are known as postictal symptoms, or symptoms that occur during a period of time following the seizure. The postictal period can last from a few hours to up to 5 days. Psychiatric symptoms characteristically tend not to occur on the same day as the seizure, but typically they occur 24 to 72 hours after the seizure. When it comes to suicidal attempts during the postictal period, there haven't been many studies. One Japanese study reported more suicide attempts in patients in whom postictal psychotic episodes developed; 7% of patients with postictal psychotic episodes attempted suicide compared with 2% of patients with psychotic episodes unrelated to their seizures.[6] We did a study a few years ago on 100 consecutive patients with fully controlled epilepsy and found that 13% of patients habitually experienced suicide ideation after more than 50% of their seizures.[7] The median duration of the suicidal ideation was 24 hours, which is a prolonged period of time. When we looked at other parameters in patients with postictal suicidal ideation, we found that they had histories of bipolar illness or major depressive disorders and had to be hospitalized in psychiatric units for their mood disorder. So postictal suicidal ideation is a red flag for having had a serious mood disorder.

Of the epileptic syndromes that have been associated with increased suicidal ideation, temporal lobe epilepsy carries a higher risk for suicidality; however, this may be biased because patients with temporal lobe epilepsy are also at increased risk for depression. The association between temporal lobe epilepsy and suicidality has been suggested by several authors, but it has not been confirmed by all authors. One study suggested that the risk for suicidality could be seen in any type of epilepsy.[8]

Medscape: How do AEDs fit in this overall picture?

Dr. Kanner: Unfortunately, the literature that addresses the suicidal risk with AEDs is not that great. There is only 1 uncontrolled study that systematically looked at suicidal ideation and suicidal attempts with exposure to AEDs; that study showed that 0.7% of 517 patients treated with levetiracetam had to be discontinued because of suicidality.[9]

However, there is evidence that some AEDs can cause depression, anxiety, impulsive behavior, and at times, psychotic symptoms. In particular, symptoms of depression and anxiety are found with several AEDs and in particular in the older-generation AEDs, barbiturates, phenobarbital, and primidone. Among the newer-generation AEDs, vigabatrin, topiramate, tiagabine, and levetiracetam have been associated with an increased incidence of psychiatric adverse events. When you look at the risk factors of patients with psychiatric adverse events, however, a significant proportion of those patients had psychiatric histories prior to the exposure of AED, or they had a family psychiatric history, or they had both a personal and a family history. So what that tells you is that the patients who exhibit psychiatric symptoms when exposed to certain AEDs already have a psychiatric disorder or are at risk for a psychiatric disorder. It is possible that AEDs bring psychiatric vulnerability to the surface, which in people with no risk factors would not have as much impact.

The other thing you have to consider is that in many of these patients, mood disorders and anxiety disorders are cyclical. They come and go and it may happen that psychiatric symptomatology may be developing irrespective of exposure to the AED. This is difficult to determine in the absence of controlled studies. That's what we know about AEDs and psychiatric adverse events.

Medscape: Why did the FDA investigate suicidality in AEDs?

Dr. Kanner: The FDA started to look at this because of a report to the FDA that gabapentin could cause an increased incidence of suicidal ideation in people exposed to the drug. One of the manufacturers saw a signal there and the FDA sought to determine whether that was also occurring with other AEDs. They therefore looked at AEDs that had been studied in multicenter, randomized, placebo-controlled trials. The AEDs included carbamazepine, divalproex sodium, felbamate, lamotrigine, gabapentin, levetiracetam, oxcarbazepine, pregabalin, tiagabine, topiramate, and zonisamide. The FDA wanted to ascertain how often patients who were exposed to the drug had suicidal ideation, suicide attempts, or committed suicide, and how did these patients compare with patients exposed to placebo.

The FDA used a method developed at Columbia University known as C-CASA or Classification Algorithm of Suicide Assessment. They went into the data source of the studies to extract the reports of suicidal ideation, attempts, and completed suicide retroactively. Of 43,892 patients, 0.4% of people exposed to the drug had some kind of suicidality vs 0.2% exposed to placebo, which led to the conclusion that if you are exposed to AEDs you have an increased risk. However, there is a problem with this. First, a lot of patients who have suicidal ideation won't necessarily bring it to the attention of physicians in the clinic. For example, many patients may have had suicidal ideation for a long time; they've learned to live with it and they don't see it as an issue. Patients may choose not to bring it to the attention of their physician because they don't want to be labeled as being crazy or they don't want to be sent to a psychiatrist.

Spontaneous reports of suicidality by patients do not necessarily reflect what is happening in the cohort of patients. In fact, the prevalence of suicidality in published drug studies is much greater. For example, in a study that was done in 5 epilepsy centers (and we were one of the centers), it was found that 12% of 139 patients had suicidal ideation and 20% had suicide attempts and lifetime suicide attempts.[2] Whereas the FDA performed a retrospective analysis of AED studies, we were looking systematically at the presence of suicidal ideation and behavior in epilepsy centers and found a high frequency of suicidality among samples of patients. The FDA is missing a lot if they found such a low rate with the methodology used.

There is another well-known phenomenon called report bias, which the FDA acknowledged in their statement. When a patient is enrolling in a study, if the patient is placed on the active drug as opposed to placebo, he or she is more likely to experience an adverse event by chance because of exposure to the drug. Patients recruited in drug studies are asked to contact the investigator, coordinator, or nurse to report side effects. If there is increased contact with patients reporting adverse events of any kind, then patients are more likely to report suicidality by chance, which as shown by the data I cited above is a relatively common phenomenon in patients with epilepsy, particularly suicidal ideation. In fact, 67% of the suicidality events identified consisted of suicidal ideation.

Medscape: The safety of AEDs was discussed at the December 2008 meeting of the AES. Please describe the meeting discussions on AED risk for suicidality and explain why some members dispute the FDA findings.

Dr. Kanner: The AES organized a symposium, which I chaired. Panelists included an adult psychiatrist, John Barry; a child psychiatrist, Rochelle Caplan, who had been on the advisory board of the FDA; 2 epileptologists, Jackie French, who is very involved in drug studies, and myself; and 2 neuroepidemiologists who have a lot of experience in methodology of epilepsy research, Dale Hesdorffer from Columbia and Anne Berg from Northern Illinois University.

Dale Hesdorffer and Anne Berg went into the source document and looked at how the data were analyzed by the FDA. The first thing they noted was that when you work with data of this nature retroactively, a lot of data are going be missed because not all patients were asked about having suicidal ideation or behavior from the beginning and throughout the study, and as I stated above, the meta-analysis was based on spontaneous reports only. When a patient is entered into a drug study, we don't always investigate whether he or she has had suicidal ideation in the past. You do a rather general assessment of the psychiatric profile and exclude patients with serious psychiatric history. If a patient wants to be in the study, he or she is not going to volunteer suicidal ideation or prior suicidal attempts unless you ask. That's a very important methodologic problem.

The experience with selective serotonin reuptake inhibitors (SSRIs) is an example of how data collected in a retroactive manner can give the wrong impression. A black box warning was mandated by the FDA for all SSRIs. This also resulted from a concern that SSRIs increase suicidal ideation and behavior in children exposed to the drugs. When the FDA looked at that, they came to the conclusion that there was a 1.954 statistically significant increase in risk for suicidality in patients taking SSRIs. However, when suicidality data were later systematically reviewed using rating scales completed by patients during their participation in the SSRIs trials, the relative risk was found not to be 1.95, but 0.92, which is no different from placebo. This illustrates that reliability on 1 method can provide erroneous results that differ from those of a systematic and prospective investigation in both patients exposed to drug and patients exposed to placebo.

A second issue was that the FDA suggested that 8 of the 11 AEDs showed greater risk for suicidality. When Dale Hesdorffer and Anne Berg looked at the actual data, they found statistically significant differences in only 2 of the AEDs. There were 2 AEDs, carbamazepine and valproic acid, with a small protective effect; that is, patients were significantly less likely to exhibit suicidality. Moreover, only 4 patients of 43,000 died of suicide. Compared with what I told you earlier, this is a very small percentage.

A third issue that was confusing was that the FDA combined all AEDs into 1 class, with the rationale that all AEDs are used to treat epilepsy and therefore they must be doing the same thing. Yet AEDs can have very different mechanisms of action. When the FDA was challenged about this, they responded that they didn't want to single out specific drugs because that would influence the way clinicians prescribe AEDs. The agency reasoned that some clinicians who have read the FDA alert may not touch the 11 drugs analyzed by the FDA for fear of legal ramifications if the patient commits suicide. The FDA acknowledged that they included all AEDs because they didn't want clinicians to single out certain drugs. That's not science, that's more an administrative or political tactic or something else. You cannot make a fact out of concern for what clinicians may or may not do. Scientific data must be presented for what they are.

A fourth problem was that drug trials of people with epilepsy showed a greater incidence of suicidal behavior than drug trials of people with psychiatric disorders and pain. However, the epilepsy and nonepilepsy drug trials were very different in that 92% of drug trials for epilepsy were add-on trials; that is, the study drug was added to an existing regimen of AEDs. The drug studies for mood disorders and pain were monotherapy. Polytherapy regimens have a significantly greater risk for side effects, so it is difficult to determine whether the reported suicidality was the result of the study drug or of polytherapy.

Fifth, suicidal behavior was greater in certain geographic regions. For example, the ratio of suicidality was 1.38 in North American studies and 4.53 in studies done elsewhere, outside of North America. What does this mean? Are people in North America less likely to commit suicide or have suicidal ideations? There is obviously something in the methodology of these studies that makes the data questionable. There was also a difference in incidence of suicidality by the type of study: studies on epilepsy had 3.5 the odds ratio for suicidality vs 1.5 for studies on psychiatric disorder and 1.8 for studies on pain.

When you start adding all of these problems, there is enough concern to question whether the FDA alert data really demonstrate suicidality or whether there is something else going on or whether their findings are a result of the methodology they used. Nobody questions that exposure to AEDs can increase the risk of experiencing symptoms of depression, anxiety, and other psychiatric symptoms. I think that not accepting this fact would be rather foolish, but as I said before, the available data suggest that these symptoms take place in patients who have a higher predisposition for psychiatric symptoms. I think that when we present the problem of suicidality and AEDs, we have to consider the complexity of the relationships between suicidality and seizures, types of epilepsy, psychiatric antecedence of individuals, family history, and the other things.

Medscape: Why is the AES concerned about the FDA alert?

Dr. Kanner: The AAN, AES, the American Neurological Society, and the Child Neurology Society are professional societies that are very worried about the impact that warnings in package inserts may have. First, a lot of patients may, in reading those warnings, get very frightened and not take the drug. The FDA advisory board recommended against using a black box warning, but they agreed to have a warning. However, patients don't differentiate between black box warnings and other warnings. If a warning label is written and patients see it, they will get pretty anxious. Second, there are patients who don't like to take medications, and the patients may say, "Well, I'm not going to take the medication if it's going to cause this," and the consequence of that is that patients can die from stopping their medications. A recent study showed that poor compliance with AED medications increases the risk for death by fourfold to fivefold.[10] This is not something to take lightly. Patients may decide not to take their medications, not to consult with their physician, or to decrease the dosage of their medications on their own.

Another thing to consider is that physicians who are not epileptologists or neurologists, such as general practitioners or internists, might approach this problem and decide not to risk being sued because a patient they treated with AEDs committed suicide. Those physicians may no longer treat patients with epilepsy. So what happens when you live in a rural area and there are no neurologists? Who is going to treat those people?

These warnings issued by the FDA can have serious consequences in the way patients, families, and physicians react. What patients and families have to realize is that the risk for suicidality relative to taking the medication is very small. One in 500 patients on average will have an increased risk for suicidality. By comparison, the risk for injury from stopping medication or lowering the dose without the supervision of a physician is significantly greater and can result in death. There is a big debate among psychiatrists because of what's been happening since the FDA came out with the SSRI black box warning. There has been an increase in the number of completed suicides by children and adolescents because parents did not allow their children to be treated with SSRIs and physicians did not want to prescribe them.

Medscape: What precautions can clinicians use to monitor patients taking AEDs?

Dr. Kanner: With seizures, unfortunately you have to treat patients with AEDs. Psychotherapy doesn't work for seizures as it may work for psychiatric conditions. What this is going to require -- and this is a good thing -- is for neurologists or any physician treating a patient who has epilepsy to investigate for the presence of mood, anxiety, and other psychiatric symptoms, and to refer those patients to mental health professionals when such symptoms are identified. Comorbid anxiety and depression are relatively frequent in patients with epilepsy and this is an area that physicians are still not identifying and treating.

Medscape: Most of the trials in the analysis did not extend beyond 24 weeks. What about patients with epilepsy who've been taking the same AED(s) beyond 24 weeks?

Dr. Kanner: I think if you haven't had side effects of depression or anxiety during the first 24 weeks, then you are probably not going to have them if you stay on a particular AED. When a patient has been on an AED for a long time, and he or she calls and asks whether there is a chance of committing suicide from this AED, the response is no, because side effects usually occur early in the course of treatment and not later on. This applies to patients with refractory and new-onset epilepsy.

Medscape: Can you comment on any possible differences in increased risk for suicidality between branded vs generic AEDs?

Dr. Kanner: In theory there should be none because a generic drug is supposed to be the same molecular compound as the branded drug. There is a lot of controversy about whether the efficacy of branded and generic medication is really the same. Side effects or worsening of seizures may occur in approximately 10% to 15% of patients switched from branded to generic AEDs, and these patients require a switch back to branded medication. That has been seen with certain AEDs in studies done in Canada. When it comes to suicidality, however, I don't think that generic vs branded comes into play.

Medscape: With safety in mind, what's your opinion about emerging investigational AEDs?

Dr. Kanner: I'm sure that the FDA now is going to require that manufacturers of any drug being developed determine the potential for suicidality with the drug. When data are obtained prospectively and the specific suicidality symptoms are systematically assessed in every patient entered into the study, including both those randomized to active drug and those randomized to placebo, we will get the answer to the issue that the FDA tried to address in the meta-analysis. That's a study that needs to be done. So we will get the answer with the new drugs that are coming down the pipes.

Medscape: Is increased risk associated with dose?

Dr. Kanner: I think that it depends on the drugs. In some drugs, such as phenobarbital, psychiatric adverse events are more likely at higher doses. In other drugs such as levetiracetam, psychiatric adverse events can occur at low and at high doses. It's individual to the drug.

Medscape: In the FDA meta-analysis of AEDs, do you think there are specific patient populations, by age or epilepsy types, who are at greater risk for suicidality?

Dr. Kanner: I think that is probably what happened. There may have been a subgroup of patients with a prior psychiatric history of depression and anxiety and/or a family psychiatric history -- that is, patients already with an increased risk of having psychiatric adverse events when exposed to certain AEDs. I don't think this applies to all AEDs, and I think clinicians will need to investigate prior history of psychiatric illness before starting patients on AEDs. For drugs that are known to cause psychiatric symptoms, clinicians need to be more cautious or maybe leave those drugs on the back burner and try AEDs that are known to have a positive psychiatric profile. Patients should always be advised to call clinicians if symptoms of depression or anxiety appear.

Medscape: This has been a very interesting conversation. Thank you so much Dr. Kanner for your time and insights.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: