Background: The proposed link between fluoxetine and suicidal ideation is explained by fluoxetine-induced akathisia and other dysphoric extrapyramidal reactions.

Method: The following literature is reviewed: (1) the subjective response of schizophrenics to akathisia, including evidence that akathisia gives rise to suicidal ideation; (2) the subjective reports of patients taking fluoxetine; and (3) preclinical studies describing the role of serotonin in the extrapyramidal system and suggesting a mechanism whereby fluoxetine can induce extrapyramidal side effects.

Results: The literature suggests that fluoxetine-induced extrapyramidal reactions may be a mediator of de novo suicidal ideation.

Conclusion: We propose a syndrome which we name Extrapyramidal-Induced Dysphoric Reactions, one extreme manifestation of which is the emergence of suicidal ideation. We further propose a heuristic "Four Neuron Model of the Extrapyramidal Motor System" in which increased serotonin activity, by inhibiting the nigrostriatal dopamine tract, is capable of inducing extrapyramidal side effects.

(J Clin Psychiatry 1992;53:401-406)

Received Dec. 30, 1991: accepted April 23, 1992. From the Department of Psychiatry, College of Physicians and Surgeons of Columbia University, New York, and the Department of Psychiatry, Presbyterian Hospital. New York (Drs. Hamilton and Opler), and the New York State Psychiatric Insti tute. New York (Dr. Hamilton).

The authors thank Mr. David Lane for technical assistance. Reprint requests to: Lewis A. Opler, M.D., Ph.D., Neurological Institute (Room 617), 710 West 168th Street, New York, NY 10032.

ince the publication of the article by Teicher et al.' in February 1990 suggesting a causal relationship between the initiation of fluoxetine therapy and the emergence of de novo suicidal ideation, the psychiatric literature as well as the lay press has put forth numerous opinions, case reports, and articles, some consistent with these ideas while others claim to discount the proposed link between fluoxetine and suicidal ideation. In reviewing the cases of Teicher et al. as well as subsequent case reports, we have been impressed that many of the cases 'described had in common a change in psychomotor status, specifically either the emergence of purposeless motor restlessness or the induction of psychomotor retardation in

those fluoxetine-treated patients who were also experiencing suicidal ideation. We hypothesize that this is due to fluoxetine's ability to affect the extrapyramidal motor system with either a de novo induction or a worsening of extrapyramidal symptoms (EPS) in general, and of akathisia in particular.

Our hypothesis is built upon two separate literatures derived from searching MEDLINE (1966 to present) and Excerpta Medica Psychiatry (1980 to present) data bases (1) a clinical literature describing mainly schizophrenic patients in whom the induction of EPS (akathisia in particular) led to profound changes in ideation and behavior, including both suicidal ideation and homicidal thinking. Thus, we argue, "adverse reactions" to fluoxetine may be in many instances the result of fluoxetine-induced extrapyramidal side effects (akathisia or akinesia in particular) with resultant suicidal ideation and behavior. Therefore the suicidal ideation reported with fluoxetine may represent an extreme behavioral response to the unpleasant seasation of EPS. This may be an extreme example of the syndrome of "behavioral toxicity" as previously described by Van Puten et al., which might more aptly be described as Extrapyramidal-Induced Dysphoric Reaction, an extreme manifestation of which is suicidal ideation; and (2) a preclinical literature documenting anatomical and pharmacologic mechanisms whereby serotonergic agents can interact with the dopaminergic nigrostriatal tract, thereby causing extrapyramidal effects (this will be elaborated below both in our literature review and in our "Four Neuron Model of the Extrapyramidal Motor System"). In addition, as an example, we wish to report a similar case of a patient in whom this syndrome developed during treatment with fluoxetine.

CLINICAL EVIDENCE OF A LINK BETWEEN EPS AND SUICIDAL IDEATION

Several reports already exist in the literature documenting the development of EPS in association with fluoxetine, but without necessarily linking this to an increased incidence in suicidal ideation. Specifically, Lipinski et al.' first reported the occurrence of akathisia in five patients treated with fluoxetine. Bouchard et al.' reported that EPS developed in several of their patients while they were being treated with fluoxetine and in other patients the baseline levels of EPS worsened during fluoxetine treatment. Symptoms noted included bradykinesia, cogwheel rigidity, and akathisia. Tate' reported that a patient who had previously tolerated haloperidol alone had an increase of EPS (including parkinsonism and akathisia) when fluoxetine was added. Stein' reported a case of tardive dyskinesia that developed when a low dose

235

WA of haloperidol was added to fluoxetine. In the cases reported by Teicher et al.,' four of the six patients described complained of an inner restlessness which Opler' has previously argued could reflect that they were experiencing akathisia. Wishing et al. recently reported that five patients treated with fluoxetine experienced "agitation, restless motor movement, dysphoria, pacing, an internal sense of desperation, and suicidal ideation," and they too suggest "that fluoxetine-induced akathisia can lead to suicidal ruminations."

A separate clinical literature suggests that akathisia can at times lead to emergence of suicidal ideation. Akathisia is defined as an "inner sense of restlessness" and an "inability to sit still." Patients who experience this often give reports such as "I feel like I'm jumping out of my skin." As akathisia is a common side effect of neuroleptic medications, information regarding subjective response to akathisia exists primarily, although not exclusively, in the literature on schizophrenia. In 1974 Van Putten et al. noted that nine schizophrenics treated with high-potency neuroleptics showed "behavioral toxicity" associated with akathisia. Three of these patients developed de novo suicidal ideation. Schulte reported five cases of violent behavior, including completed suicides, as a result of akathisia in patients treated with neuroleptics. Shear et al." reported two cases of completed suicide by jumping in patients who the authors argue were suffering from akathisia. Drake and Ehrlich" also reported two cases of suicidal ideation secondary to akathisia. In one case the patient stated that he did not intend to die but that he would do anything to escape the intolerable feeling of restlessness. Drake and Ehrlich noted that these patients were unable to distinguish the akathisia from the ongoing symptoms of their psychiatric illness. Weidea" reported that the use of prochlorperazine for nausea in a patient receiving chemotherapy led to akathisia which was very distressing to the patient. In 1986 Weddington and Banner" successfully used chlorpromazine and metoclopramide to treat intractable hiccups but found that after 3 days of treatment the patient became restless, felt like he was "going crazy," and began obsessing about suicide. During a crossover study involving haloperidol and BW2344-U (which is characterized by the absence of dopamine receptor affinity), Shaw et al." noted that during haloperidol treatment the patients experienced a clinical decline characterized by severe akathisia and an increase in violent behaviors, as well as suicidal ideation and homicidal thinking. None of the symptoms were present with BW2344-U. In a 1987 review article, Van Putten et al.' cite several studies in which it was noted that akathisia leads to suicidal ideation or homicidal thinking. They called this the "behavioral toxicity" of antipsychotic medication. By 1988 Hermesh et al." began studying the use of propranolol to treat akathisia because of the authors' familiarity with the above literature and their concern that akathisia might lead to suicide attempts.

Case Report

A 32-year-old woman, with a history of major depression and panic attacks, presented complaining of de

pressed mood, decreased sleep, increased appetite, and anhedonia, but notably without suicidal ideation, for 4 months. Three years earlier, she had been successfully treated with desipramine but had found the side effects of this medication highly unpleasant and preferred not to take a tricyclic antidepressant at this time. During the earlier episode, she had experienced mild-to-moderate suicidal ideation but without intent and without any attempts, and she did not require hospitalization. The patient was started on fluoxetine 20 mg/day. Within 10 days she began complaining of panic-like symptoms, anxiety, and palpitations, but at that point was without suicidal ideation; the fluoxetine was reduced to 5 mg/day with resolution of these symptoms. She was maintained on this dose for 3 weeks and her depression began to resolve. At this point, however, the patient started complaining of symptoms she had never experienced before, specifically feeling restless and out of control. "I feel like I need to hold onto my chair or else I'll jump out the window." The patient stated that although her mood was good, she was afraid that she would kill herself because of these restless and out-ofcontrol feelings. She was emphatic that the way she was feeling now was different from and more frightening than the previous time that she had experienced suicidal ideation as part of a depressive episode, but that nonetheless she was experiencing suicidal ideation. As the emergence of suicidal ideation seemed linked to fluoxetine treatment, fluoxetine was immediately discontinued; after several days the feelings of restlessness as well as the suicidal ideation thoughts simultaneously ceased. She subsequently responded well to nortriptyline.

PRECLINICAL EVIDENCE OF A LINK BETWEEN SEROTONERGIC AGENTS AND EXTRAPYRAMIDAL EFFECTS

How might an agent like fluoxetine, believed to act primarily as a potent serotonergic reuptake inhibitor, induce and/or worsen extrapyramidal side effects? In a 1991 paper, Opler" proposed a heuristically useful "Three Neuron Model of the Extrapyramidal Motor System" in which he describes mechanisms whereby dopaminergic. cholinergic, and GABAergic agents can have effects on the extrapyramidal motor system. In this model, the first neuron represents the major input into the basal ganglia, the inhibitory nigrostatal dopamine tract. The second neuron represents the excitatory cholinergic interneurons, and the third neuron represents the major inhibitory GABAergic outflow tract that modulates voluntary motor activity. In this model, the use of neuroleptic agents that block dopamine at the first synapse causes a disinhibition or increased firing of the second neuron and subsequently an increased firing of the third neuron (releasing GABA). As the GABA system inhibits the voluntary motor system, an increase in GABAergic output leads to an increase in tonic inhibition of voluntary motor activity, translating clinically into the bradykinesia observed in both idiopathic and neuroleptic-induced parkinsonism.

While the "Three Neuron Model" is heuristically useful for understanding the dopaminergic, cholinergic, and

GABAergic interactions, to address the influence of serotonergic agents on the extrapyramidal motor system, we will in this paper develop the rationale for and propose a "Four Neuron Model," in which we add a new first neuron that is serotonergic, depicting the raphe-striatal tract, functioning to inhibit firing of the nigrostriatal tract, and therefore itself capable of inducing extrapyramidal side effects (Figure 1).

What is the basis for proposing such an inhibitory serotonergic input? Much of the evidence is found in the preclinical literature describing both (1) anatomical projections from the raphe to the nigrostriatal tract and (2) the effects of different pharmacologic agents on catalepsy in

rats.

Neuroleptic-induced catalepsy in rats remains the most widely used animal model for neuroleptic-induced parkinsonism in humans. We acknowledge at the outset that there is no adequate animal model for akathisia and that, while EPS and akathisia are related phenomena, adrenergic as well as dopaminergic and serotonergic mechanisms have been suggested as possible mechanisms for akathisia. The well-documented response of akathisia, but not parkinsonism, to propranolol" of course argues for their dissimilarity. For purposes of generating testable hypothe. ses, and in the absence of a more precise animal model, with the above caveat, we will utilize catalepsy in rats as a model for EPS, including akathisia.

ANATOMICAL EVIDENCE FOR A RAPHE-STRIATAL PROJECTION

Azmitia" has summarized evidence for the presence of serotonergic projections from the raphe to the substantia nigra: Histochemical techniques have been used to demonstrate that the midbrain raphe neurons contain serotonin and that these neurons have very long projections into the forebrain that disappear when the raphe cell bodies are destroyed; anatomical studies show that fibers from the dorsal raphe arcuate tract project directly to the substantia nigra; and neurophysiologic studies demonstrate that serotonin-containing fibers inhibit the firing of caudateputamen cells and may also innervate the substantia nigra as well as other forebrain nuclei. Thus the extensive innervation of the extrapyramidal system by serotonergic fibers has been well documented.

PHARMACOLOGIC EVIDENCE FOR EFFECTS OF SEROTONERGIC AGENTS ON THE EXTRAPYRAMIDAL SYSTEM

In 1975 Maj et al." demonstrated that cyproheptadine (a serotonin antagonist) antagonizes the catalepsy induced by neuroleptics and potentiates the anticataleptic activity of levodopa and amantadine. Costall et al." demonstrated that serotonergic mechanisms are important in the mediation of catalepsy. It was thought that the actions of cerebral serotonin and dopamine might be related. Carter and Pycock" showed that treatment with serotonin antagonists leads to a decrease in the cataleptic effect of haloperidol in rodents. They also noted that increased levels of serotonin

transmission enhances catalepsy in rodents only after the blockade of dopamine receptors has been established. Davies and Tonogroach used electrical stimulation in rat brains to elucidate possible pathways. They suggest that serotonin may mediate dorsal raphe nucleus-evoked inhibition of striatal neurons and found that this inhibition is antagonized by methysergide, a serotonin antagonist. This is evidence of a raphe-striatal serotonergic pathway, suggesting a location for the fourth neuron in our model. Westfall and Tittermary studied rat brain striatal tissues and found that serotonin bas inhibitory effects on the depolarization-induced release of dopamine from dopamine terminals in the striatum, again suggesting that serotonin modulates dopaminergic transmission. In 1985 Ceulemans et al. reported a pilot study of setoperone, an antipsychotic medication that antagonizes the action of both dopamine and serotonin. They found that patients treated with this medication experienced less EPS, which supports the hypothesis that serotonin blockade permits a reduction in the amount of dopamine blockade needed for antipsychotic effects. In a recent letter Baldessarini and Marsh" reported that fluoxetine inhibits synthesis of catecholamines in dopamine-rich areas of forebrain and this effect persists in hippocampus and striatum. This finding suggests that serotonin inhibits dopamine neurons and can therefore lead to decreased stimulation of dopamine and thus cause EPS.

SEROTONERGIC-DOPAMINERGIC INTERACTIONS: AN EVOLVING STORY

We are aware that the evidence cited in support of our "Four Neuron Model of the Extrapyramidal Motor System" does not do full justice to the complexity of serotonergic-dopaminergic interactions within the central nervous system, a topic beyond the scope of this paper. Preclinical research regarding serotonergic-dopaminergic interactions is presently an area of intense interest and productivity. This is an exciting development that promises to enrich our understanding of basic brain mechanisms. Additionally, the ability of serotonergic drugs to affect dopaminergic systems opens new vistas as regards their potential use in the treatment of psychotic and substance abuse disorders.

We wish to acquaint the reader with some of this exciting new research. Since we have until now selectively focused on anatomical and pharmacologic evidence supporting our "Four Neuron Model of the Extrapyramidal Motor System," in the interest of balance, we will fo