Additional related information may be found at:
Neuropsychopharmacology: The Fifth Generation of Progress

Back to Psychopharmacology - The Fourth Generation of Progress

The Treatment of Tardive Dyskinesias

George Gardos and Jonathan O. Cole


The treatment of tardive dyskinesia (TD) poses unusual problems. This iatrogenic condition is at the interface of psychiatry and neurology insofar as psychiatric patients are most likely to develop TD, while TD, being a movement disorder, is in the province of neurology. A second difficulty is that TD is a heterogeneous entity with respect to its clinical features, topography, and pathophysiology. A third anomaly is that most cases of TD also involve an underlying psychiatric disorder with major impact on the development, course, and therapy of TD.


The coexistence of TD and a psychiatric disorder, frequently but not invariably a chronic psychosis, raises complex risk-benefit issues in treatment planning.

The prevalent strategy is to arrest the progression of TD by minimizing neuroleptic (NL) exposure while simultaneously providing appropriate pharmacotherapy. Implicit in this paradigm, although seldom stated explicitly, is the notion that the psychiatric disorder (most often schizophrenia) is the more disabling condition, while the movement disorder is usually less severe and does not cause functional impairment.

Neuroleptic drug discontinuation is the method which appears straightforward and conforms to the principle that removal of the causative agent may result in cure. In this instance, clinical thinking is partly supported by research data. Jeste and Wyatt (52) reviewed 23 studies in which NLs were withdrawn for the purpose of treating TD for periods varying from 1–2 weeks to 3 years. Of the 631 patients, 37% had remission of symptoms (defined as a 50% or greater decrease in rating scale scores). A more recent review by the same group (50) cited three additional NL withdrawal studies where the averaged remission rate of TD was 55%. The clinical feasibility of neuroleptic withdrawal is severely limited by the high risk of psychotic relapse in chronic psychotic patients.

Long-term studies have shown that TD can be successfully managed within the context of appropriate NL treatment of coexistent severe mental illness (22). Follow-up studies of TD for 5 years or longer have tended to confirm that TD typically stays stable or improves even with continued neuroleptic treatment (37), although neuroleptic dose reduction may benefit both TD and clinical status (55). Alternatives to NL treatment, including mood stabilizers (lithium carbonate, carbamazepine, divalproex sodium), benzodiazepines, or antidepressants (58) are sometimes effective clinically and may lead to amelioration of TD in the absence of NLs. During the last few years, there has been a shift from traditional NLs towards the newer, atypical NLs for the treatment of patients with psychotic disorders. New epidemiological studies will be needed to confirm the clinical wisdom that these atypical NLs produce not only less acute EPS but a lower incidence of TD as well. At the same time, the newer atypical NLs, particularly clozapine, have emerged as potentially effective treatments for TD.

Specific step-by-step recommendations for the clinical management of patients with TD, including algorithms, have been offered (6,18, 38). The doctor-patient relationship and the provision of adequate information to patient and family are also key aspects of the appropriate clinical management of TD.



Methodological Issues

Studies often distinguish between reversible and irreversible TD, but no agreement exists as to whether reversible TD implies a complete disappearance of all TD symptoms. Similarly, the issue of whether the definition of irreversible TD allows for improvement in TD with continued TD symptoms remains unresolved (18). Persistent versus remitting TD is a more flexible dichotomy, and, when coupled with operational criteria such as Schooler and Kane's research diagnoses for TD (83), a more accurate description of TD outcome may result. However, to assess treatment response, a criterion for improvement is also needed. A statistically significant treatment effect does not always reflect clinically significant improvement in particular patients (51). The criterion of at least 50% improvement was chosen as the indicator of positive drug response in the treatment studies to be cited.

Placebo Response

A sizable placebo response rate was found in double-blind studies: groups of placebo patients showed mean improvement in TD of as much as 50% (89), and the proportion of placebo patients improving by > 50% could be as high as one-third (50). The placebo effect makes interpretation of uncontrolled positive studies problematic.

Shifting Baseline

Tardive dyskinesia severity is often subject to diurnal variations, as well as to apparently spontaneous fluctuations over days or weeks (63) which inflate the variance in treatment studies and may lead to Type II errors. A different problem causing Type I errors may arise when, as is often the case, patients enter a treatment study at or near their worst level of TD. Such patients sometimes improve spontaneously no matter what the treatment and will inflate both the drug and placebo response rates—although not the drug-placebo differences.

Other methodological issues which are not specific to TD treatment studies are carry-over effects, which may confound cross-over studies, as well as problems handling dropouts.

Criteria for an Adequate Clinical Trial

Compounds which may be effective in treating TD should prove their therapeutic efficacy in well-designed studies. Based on Mackay and Sheppard's (71) recommendations, the following criteria for a methodologically sound therapeutic trial in TD are suggested: double-blind design, an adequate number of cooperative patients, clear diagnostic and descriptive definitions, valid and reliable rating scales, attention to the timing of ratings, and holding non-research medications constant. While few studies in the literature measure up to these standards, studies were judged on how clearly they approximated the ideal design.

Agents Used for Treating Tardive Dyskinesia

Until recently, Casey's assertion (19) that no drug was safe and effective over extended periods in the treatment of TD was unquestionably true. With the recent arrival of clozapine and other atypical NLs, and possibly Vitamin E as well, specific therapies which are safe and effective for TD are becoming available.

Typical Neuroleptics

The administration of traditional NLs to patients with TD often leads to a decrease in the severity of dyskinetic movements; studies in which NLs were employed as treatment for TD confirmed these observations. Jeste and Wyatt (52) found an average improvement rate of 66.9% in 501 NL-treated patients, while the first APA Task Force Report (55) similarly found that NL treatment produced >50% improvement in 60% of TD patients.

The reduction of TD severity by typical NLs may be movement suppression rather than a true antidyskinetic effect for the following reasons. The decrease in TD may be temporary and may be accompanied by a simultaneous increase in parkinsonian symptoms (57). Furthermore, after NL withdrawal, a rebound increase is often observed in proportion to the TD suppression produced by the NL (42).

Catecholamine Depleters

These compounds deplete catecholamines from nerve terminals and have been employed as antipsychotic as well as antidyskinetic drugs for several decades.

In a double-blind, placebo controlled study, Huang et al. (47) obtained a statistically significant improvement with reserpine 0.75–1.5 mg daily; 5 of 10 patients who received reserpine showed >50% improvement. More recently, Fahn (34) used reserpine in doses up to 8 mg/day and found marked and long-lasting improvement in 8 of 17 patients. Reserpine appears more effective as an antidyskinetic drug without concurrent NL therapy (34). Side effects such as hypotension, sedation, parkinsonism, and depression limit the clinical usefulness of this compound.

Tetrabenazine, a nonindole benzoquinolizine, is a presynaptic monoamine-depleting agent as well as a postsynaptic dopamine receptor blocker (72). It is not marketed in the United States. Tetrabenazine in typical doses of 150–200 mg daily produced positive results in both open and controlled studies, but side effects of parkinsonism, depression, drowsiness, and akathisia restrict the drug's therapeutic potential (72).

Although little recent research has been carried out with these compounds, they remain potentially useful and are still frequently employed by neurologists (58).

Atypical Neuroleptics

A new generation of antipsychotic drugs has started to replace traditional NLs. These atypical NLs show novel psychopharmacological activity and have less potential for inducing acute EPS (32).

Clozapine is a dibenzodiazepine antipsychotic with relatively weak D1 and D2 receptor binding activity, as well as blocking of 5-HT2, a1-adrenergic, H1 histaminergic and muscarinic acetylcholine receptors. Clozapine may selectively block dopamine receptors in the limbic system of the forebrain (7).

The favorable EPS profile of clozapine was corroborated by a review of studies involving 1300 patients in which prevalence rates for EPS were found to be 0–20% (20). Gerlach and Peacock (41) compared 100 patients on clozapine with 100 patients on typical NLs and found less EPS in clozapine-treated patients: tremor was seen in 3% (vs 11% in control patients), rigidity in 0% vs 19%, and no dystonia on clozapine versus 13% in control patients. Unequivocal cases of TD caused by clozapine are extremely rare. There are, however, reports implicating clozapine in hyperkinesia (9), jaw dyskinesia (28), myoclonic jerks and drop attacks (5,8,43), and exacerbation of dyskinesia.(27).

The therapeutic effects of clozapine in TD have been demonstrated in several clinical studies. Open studies explored effective dose ranges and duration of treatment. Simpson et al. (87) found statistically significant reductions in TD scale items after 18 weeks of treatment with clozapine at dosages of up to 500 mg daily. Gerbino et al. (40) demonstrated that after >1 year on an average dose of 650 mg/d of clozapine, 14 of 17 patients reached at least 90% improvement. Cole et al. (26) treated 27 TD patients with clozapine. Those patients who remained on the drug for more than 3 months showed the greatest improvement in their TD symptoms. In a double-blind study involving 241 patients (54), there was a trend (p=0.09) for AIMS total scores to decrease more with clozapine than with chlorpromazine. Lieberman et al. (65) treated 37 treatment-resistant schizophrenics (30 with TD) with clozapine for an average of 25.7 months. Forty-three percent achieved remission, most within the first 6 months of treatment. Tamminga et al. (91) compared clozapine with haloperidol in 32 patients with moderate to severe TD in a double-blind protocol. The clozapine group (N=19) showed a significantly greater decrease in TD over the 12-month treatment course than the haloperidol group (N=13). At the end of the 12-month treatment period, only the haloperidol group showed exacerbation of dyskinesia after drug withdrawal.

The current status of clozapine treatment in TD may be summarized as follows. Clozapine is an important advance in the treatment of TD. Since it practically never causes TD, it can be prescribed as an antipsychotic which will likely prevent the development of TD. The studies cited above also demonstrate the ability of clozapine to ameliorate existing TD of all severity levels. Accordingly, clozapine can play positive roles in the primary, secondary, and tertiary prevention of TD. The consensus of research suggests the advisability of long-term treatment with the drug, as the remission of TD symptoms (as well as some psychotic symptoms) is slow and continues for many months. The effects of clozapine dosage remains to be clarified: some studies suggest that higher doses are more effective (35), while European studies claim that lower doses are just as effective in treatment-resistant schizophrenics, although data on TD were not published (35). The notion that clozapine exhibits specific antidyskinetic effects, as opposed to the suppressant effects of typical NLs, is supported by the time- and dose-dependent efficacy and the lack of withdrawal dyskinesia (65).

The neuropharmacological basis for the antidyskinetic effect of clozapine remains undetermined. Aspects of clozapine's pharmacological profile which may be relevant include lower D2 and higher D1 dopamine receptor binding, significant 5-HT2 antagonism, potent a2-noradrenergic antagonism, its marked affinity for the D4 dopamine receptor site, and activity at the M1, muscarinic receptor (92).

The benzamides are selective D2-receptor antagonists without significant actions on receptors of other brain neurotransmitter systems (24). Sulpiride, in doses of 400–1200 mg/day, was found to reduce TD significantly (p <0.01) without affecting parkinsonism in a placebo-controlled trial with 11 elderly patients. However, there was re-emergence of TD symptoms after sulpiride withdrawal (21). Haggstrom (45) found a dose-dependent reduction in TD and psychosis by sulpiride (200–1200 mg/day) in six patients with marked TD. Tiapride, another substituted benzamide, showed an antidyskinetic effect in two controlled trials with 12 (16) and 5 (21) patients, respectively.

Risperidone, a benzisoxazole derivative, has a high binding affinity for both serotonin and dopamine receptors (49). It has shown efficacy for positive as well as negative symptoms of schizophrenia (12,23,79). It caused less parkinsonism, dystonia and akathisia at all doses tested than haloperidol 10 mg/day in a large-scale double-blind trial (79). Other controlled studies confirmed the lower propensity of risperidone to cause EPS (12, 23). Two confirmed cases of TD have been reported with risperidone after more than one year of therapy (44). Improvement of dyskinesia by risperidone initially was reported in individual patients with TD (59,86). In a double-blind, controlled study, Borison et al. (12) found that risperidone 2–10 mg daily (N=12) produced a statistically significant suppression of dyskinesia compared with placebo. In a double-blind, multicenter study of 155 schizophrenic inpatients at doses of 6–16 mg daily, risperidone significantly lowered dyskinesia scores, compared with placebo (23). Patients with TD treated with the optimal dose of 6 mg/day risperidone showed a significantly greater decrease in dyskinesia scores than did haloperidol-treated patients. The greatest improvement was observed in patients with severe TD treated with 6 and 10 mg/day of risperidone. The liability of risperidone to cause TD, as well as its potential for the treatment of TD, remains to be established. The same applies to olanzapine—a recently approved, atypical NL with a pharmacological profile similar to clozapine.

Other Catecholamine Antagonists

Based on prevailing theories that striatal dopamine and/or norepinephrine overactivity may be associated with TD, a host of dopamine and norepinephrine antagonists in addition to those already discussed above have been tested for potential antidyskinetic effects. Most of them are of theoretical or research interest only.

Alpha-methyl-paratyrosine (AMPT) inhibits tyrosine hydroxylase, the rate-limiting enzyme in the synthesis of dopamine. Modest improvement in TD by AMPT was seen in several studies, but the samples were small and the period of observations varied between a few days and 4 weeks (52,55). Other putative dopamine antagonists—a-methyldopa, papaverine, metoclopramide, droperidol, and oxiperomide—have shown unimpressive antidyskinetic effects in small and/or uncontrolled studies.

Noradrenergic antagonists, primarily the b-adrenergic blocker propranolol and the a2 agonist clonidine, have been tested in several treatment studies. Propranolol produced mixed results in doses of 20–40 mg/day in open studies. A double-blind study with intensive case design using up to 800 mg/day of propranolol in four TD patients suggested that therapeutic effects may appear slowly: two of the four subjects showed a substantial therapeutic response during weeks 20–50 of observation (84). In a double-blind study of eight TD patients, clonidine 0.2–0.9 mg/day produced a mean improvement of 63% (36), whereas in another double-blind, placebo-controlled study all seven patients improved on 0.4 mg/day (13). Significant side effects, especially hypotension, restrict the clinical usefulness of clonidine for TD. Disulfiram and the dopamine beta-hydroxylase inhibitor fusaric acid are of theoretical interest, but appropriate treatment studies have not been conducted.

Catecholamine Agonists

Apormorphine, hydergine, piribedil, methylphenidate, amphetamine, and amantadine have been tried for TD without notable success. Bromocriptine, an ergot alkaloid derivative, possesses both dopamine agonist and antagonist activities. In doses of 0.75–7.5 mg daily, it failed to produce satisfactory improvement in TD in small open (50) and double-blind (61) studies. In a placebo-controlled study of 11 TD patients, Lieberman et al. (64) added increasing doses of bromocriptine (15–60 mg/day) to NLs for 10 weeks, followed by 8 weeks of observation off bromocriptine. No significant therapeutic effects were found. The Receptor Sensitivity Modification Hypothesis (3) postulates that dopamine agonists in relatively high doses may reverse postsynaptic dopamine-receptor hypersensitivity. Positive results with L-DOPA treatment (70) were not supported by several negative studies (19,50,88). The potential for increased psychotic symptoms further limits the clinical applicability of L-DOPA therapy. L-Deprenyl has recently attracted interest for neurological conditions, including TD (96), but controlled studies are needed. Catecholamine agonists appear to be more useful as tools to elucidate the pathophysiology of TD than as reliable treatments for patients.

Cholinergic Treatments

The application of centrally acting cholinergic compounds is based on the hypothesis that the pathophysiology of TD is linked to a relative imbalance between cholinergic and dopaminergic activity within the striatum (4).

Most of the relevant research was carried out in the 1970s and has been reviewed by Alphs and Davis (4). Intravenous physostigmine, which elevated brain acetylcholine levels, was given in 11 studies: results ranged from marked improvement (>50%) to no change or worsening, with no clear differences between double-blind, single-blind, and open studies (4). Because the drug is short-acting and toxic, its main clinical application appears to be as a pharmacological probe and as a potential predictor of response to other cholinergic therapies (4). Deanol, an orally administered putative cholinergic drug, enjoyed intense but brief popularity. Alphs and Davis (4) reviewed no fewer than 14 double-blind studies involving 199 patients, as well as numerous single-blind and open studies. The prolific research on deanol is astonishing, considering that only one of the double-blind studies showed it to be significantly superior to placebo. Furthermore, serious doubt exists as to whether and how deanol increases central acetylcholine content (4). Choline chloride (3–18 g/day) demonstrated therapeutic efficacy in one double-blind and four other trials, but distressing side effects severely limit its clinical applicability (4). Lecithin is much better tolerated than choline. Open treatment trials tended to produce positive findings, whereas three double-blind studies reported mixed results (4). The clinical use of lecithin is hampered by the fact that phosphatidylcholine content varies greatly between products of different manufacturers; therefore, accurate dosing is difficult to achieve. Another problem is the need to prescribe large doses (sometimes >50 g daily), which represent a considerable increase in caloric intake. In a double-blind, cross-over study, Gelenberg et al. (39) gave 20 g lecithin daily to 21 patients for 8 weeks and obtained statistically significant drug-placebo differences. Clinically, however, the lecithin effect was negligible.

Overall, the early encouraging results with cholinergic compounds have not held up, and cholinergic drugs are rarely used in the treatment of TD.


The same rationale that led to the optimism about use of cholinergics for TD would predict worsening by anticholinergics. Although this expectation is generally borne out, there are reports of TD improving in a minority of affected patients. Jeste and Wyatt (52) reviewed 14 studies involving 177 patients in which anticholinergic drugs were given systematically for TD and found that 7.3% of patients improved. The anticholinergics included benztropine (2 mg IV and 1–5 mg PO), trihexyphenidyl (2–27 mg/day), biperiden (6–18 mg/day), and procyclidine (30 mg/day); the duration ranged from single-dose studies to 2 months of observation. One explanation for the improvement of TD is that antimuscarinic, antiparkinsonian agents such as benztropine have potent dopaminergic and noradrenergic agonist activities in the central nervous system (73).

GABAergic Drugs

The inhibiting effect of g-aminobutyric acid (GABA) on dopamine neurons provides the rationale for treating TD with drugs that increase GABAergic influences (19). Thaker et al. (93) traced the evolution of GABAergic treatments in TD. Muscimol, the first specific GABA-A agonist to be tried for TD, produced a 48% decrease in TD symptoms in doses up to 9 mg/day in a placebo-controlled trial of seven NL-free schizophrenics. Another GABA agonist (tetrahydroisoxaxolopyridinol) was, at best, slightly beneficial in two small studies (93). Progabide, which is active at both GABA-A and GABA-B receptors, yielded 40–60% improvement in TD in one open and two double-blind clinical trials (93). The GABA transaminase inhibitors GABA-acetylenic GABA and g-vinyl GABA were tested in small-scale trials and resulted in modest improvement, which in some patients was clinically significant (93). Distressing side effects of dizziness, sedation, confusion, myoclonic jerks, and increases in parkinsonian and psychotic symptoms impose limitations on the therapeutic potential of these specific GABAergic drugs.

Sodium valproate may increase brain GABA via GABA-transaminase inhibition, although it is unclear whether in usual clinical doses this does in fact occur. Double-blind studies in which 900–2500 mg sodium valproate was added to ongoing NL treatment failed to improve TD in the majority of patients (67,76). Baclofen, a structural analogue of GABA, has shown variable effects on TD in double-blind studies (19).

Benzodiazepines may enhance GABA function (19). These widely used and generally safe compounds have been extensively used in patients with TD, although little controlled research has been published. In open studies, an average of 58% of patients improved, while in double-blind studies the improvement rate was 43% (17,19,52). Despite the high rate of improvement, benzodiazepines have not been established as specific treatments in TD. The mechanism of action and the influence of nonspecific factors, such as sedation or the antianxiety effect, need to be established (13). Nonetheless, benzodiazepines enjoy widespread use in the clinical management of TD of all types and severity.

Calcium-channel Antagonists

Case reports of unexpected benefit in TD patients stimulated clinical trials of various calcium-channel antagonists. A rationale was provided by data from animal experiments that demonstrated that calcium-channel blockers possess dopamine antagonist properties (11,69). However, Borison et al. (11) stressed the pharmacological differences among the many calcium-channel antagonists. Verapamil, for instance, crosses the blood-brain barrier more readily than diltiazem or nifedipine and was also found to exhibit dopamine-antagonist properties. Clinical studies tend to bear out these differential effects. Adler et al. (1) added up to 80 mg q.i.d. of verapamil or up to 60 mg q.i.d. of diltiazem to ongoing antipsychotic drug treatment in 21 patients with TD. AIMS ratings were done blindly. There was a 19% decrease in AIMS scores (p<0.05) in the verapamil patients, whereas the 11% decrease in AIMS scores in the 12 diltiazem patients was not statistically significant. Borison et al. (11) treated 13 treatment-refractory male schizophrenics with verapamil 80 mg t.i.d. and chlorpromazine 600 mg daily. Average AIMS scores decreased by over 50% (p<0.001) after 3 weeks and rebounded after verapamil discontinuation. By contrast, a double-blind, placebo-controlled crossover study found diltiazem ineffective (69). Studies with nifedipine have likewise been disappointing.

Adequately controlled, long-term studies have yet to be done. All these drugs, however, have serious limitations: hypotension, increase in anxiety, hostility and depression, and the dissipation of antidyskinetic effect after 1–3 months.

Vitamin E (Alpha-tocopherol)

Vitamin E (a-tocopherol) is a lipid-soluble antioxidant located on cell membranes near enzymes that produce free radicals (90). Neuroleptics increase catecholamine turnover and may increase free radical formation that, in turn, may cause neurotoxicity and may induce TD. Szymanski et al. (90) reviewed the double-blind and placebo-controlled treatment studies of vitamin E in TD. Lohr et al. (68) treated 15 patients with persistent TD with 1200 I.U. of vitamin E daily and found an average 43% decrease in AIMS scores. The seven patients who showed a > 50% response had a shorter duration of TD than did the eight patients with <50% improvement. Elkashef et al. (31) treated eight schizophrenics with TD with 1200 I.U./day for 4 weeks, in addition to their regular psychotropics; despite a statistically significant drug-placebo difference, only one patient showed >50% reduction of AIMS scores. Egan et al. (30) gave 1600 I.U./day of vitamin E to 21 TD patients. Data from the 18 patients with high blood levels of vitamin E showed no significant drug-placebo differences, except in the subgroup of 9 patients who had had TD for 5 years or less, where an 18.5% average decrease in AIMS scores was obtained. Adler et al. (2) administered 1600 I.U. of vitamin E daily in a treatment study of TD patients. During the initial 8-week trial, 2 of 16 patients showed >50% improvement. When the trial was extended to 36 weeks, 4 of 8 patients on vitamin E and none of 9 placebo patients improved by >33%. Junker et al. (53) used 1200 I.U./day of vitamin E in 16 TD patients and found significant improvement only in patients over the age of 40 years. A double-blind, crossover study with 6-week treatment periods of vitamin E 1200 IU/day or placebo was carried out on 27 outpatients with TD by Shriqui et al. (85); no significant drug-placebo differences were seen in AIMS scores.

The impression gained from these studies is that, while vitamin E is safe and well-tolerated, it confers only modest benefits. Those with shorter duration of TD improve more, but the usual response criterion of 50% improvement is reached only infrequently. Longer-term trials may yield better results.

An unresolved paradox in the application of vitamin E for TD therapy is the notion that this compound is purported to prevent neuronal damage and not necessarily to repair already damaged nerve cells. Similar reasoning may be applied to the atypical NLs, which may be assumed to prevent the development of TD more readily than to reverse already existing TD. After further demonstration of antidyskinetic efficacy, additional research will need to address the mechanism responsible for the therapeutic effects of vitamin E on TD.

Miscellaneous Treatments

Electroconvulsive therapy (ECT) has not been studied systematically for the treatment of TD. Hay et al. (46) reviewed 22 published cases of ECT in patients with TD and noted 5 cases of disappearance of TD, with improvement in another 6 patients. Nine of the 11 patients had a diagnosis of depression; the age range was 49–92 years, and 10 of the 11 patients were NL-free during ECT treatment.

Lithium carbonate. A series of case reports and small uncontrolled studies suggested that lithium carbonate benefited TD as well as Huntington's chorea, tics, and dystonias (25). Systematic open and double-blind studies, however, revealed only a modest benefit resulting from lithium carbonate treatment, whether used as the sole therapeutic agent or when added to NLs (25). There was an apparent trend in these studies for serum lithium levels of <0.8 mEq/liter to be more likely to reduce TD than lithium levels of over 0.8 mEq/liter. Some investigators reported worsening of existing TD or development of new TD on lithium carbonate, especially at toxic lithium levels (25). The lack of clear therapeutic efficacy still leaves open the question of whether lithium carbonate could prevent TD development (25).

Amino acids. Amino acids such as phenylalanine, tyrosine, and tryptophan are precursors of neurotransmitters that may be involved in the pathophysiology of TD. Recent research by Richardson and colleagues (81) demonstrated remission of TD symptoms by a breakfast meal containing branched-chain amino acids. Sandyk and Fisher (82) pioneered a dietary approach of L-tryptophan (8 g/day) combined with nicotinic acid (25 mg/day) [the latter to slow hepatic degradation of L-tryptophan], and they devised a high-carbohydrate-low-protein diet to further increase brain levels of serotonin. Improvement in TD was obtained in three patients (82). The dietary manipulation of neurotransmitter precursors is a novel approach that awaits experimental and clinical confirmation. The potential for dietary interventions to influence psychiatric disease and TD is yet to be realized (80).

Neuropeptides have received attention because of the connection of endogenous opiates with brain dopamine systems and with motor behavior. An acute study with eight TD patients showed weak antidyskinetic effects by morphine (10 mg SC) and by the synthetic enkephalin FK 33-824 (3 mg IM), but no effect by naloxone (0.8 mg IM) [10]. Ceruletide (a cholecystokinin octapeptide [CCK-8] analogue), in doses of 0.8 mg/kg, was found to reduce TD in a small number of patients (77).

Prostaglandin precursor essential fatty acids (EFAs) were tested in 16 TD patients in a double-blind, placebo-controlled, 6-week trial. Six hundred milligrams of g-linolenic acid had no significant effects on TD (98).

Insulin (10 units SC) was tested in a double-blind, placebo-controlled study in 20 DSM-III schizophrenics with persistent severe TD (mean AIMS at baseline: 13.7). Daily injections for 15 days were followed by injections every other week for a total of 90 days. At day 7, the insulin group (N=10) showed a significant decrease in TD scores in comparison with placebo group (N=10); the significant drug-placebo differences persisted throughout the study (p=<0.001) [75]. The authors believed that decreased glucose availability may have reversed receptor hypersensitivity and led to improvement in TD.

Pyridoxine (vitamin B6) has been considered for treatment of TD because of its role in the metabolic breakdown of L-DOPA. High doses of pyridoxine (1000–1400 mg daily) have been advocated as both therapy and prophylaxis for TD (29). However, controlled studies are lacking.

Buspirone, an azopirone antianxiety agent, exerts a partial agonist effect at the serotonin 5-HT1A receptor and exerts mixed agonist and antagonist effects at the dopamine D2 receptor. An open trial of increasing doses of buspirone in eight TD patients found the highest dose of 180 mg/day to be most effective: there was a mean improvement of 4.4 on the AIMS (p<0.01), and 3 of the 8 patients showed >50% improvement (74).


Several distinct forms of NL-induced involuntary movements have been recognized which may coexist with choreo-athetoid dyskinesia or may occur separately. Although much less prevalent than typical TD, they may become severe and disabling. The pathophysiology and treatment of TD variants show important differences from typical TD.

Tardive dystonia is a syndrome of sustained muscle contractions, frequently causing either (a) twitching and repetitive movements, or (b) abnormal postures (14). Blepharospasm, grimacing, torticollis, retrocollis, and the Pisa syndrome are characteristic movements of tardive dystonia. Its prevalence is estimated to be 2% in NL-treated inpatients (99).

Methodologically adequate treatment studies of tardive dystonia are conspicuously absent, probably because of the lack of a sufficiently large cohort of suitable patients at any site. The best information available on the treatment of tardive dystonia comes from series of patients treated at centers specializing in movement disorders (14,56). The general principles of therapy are similar to those for treating TD. Because tardive dystonia is an NL-induced condition, NL discontinuation is considered whenever feasible, but the results have been disappointing. In Burke's series (14), only 5 of 42 (12%) NL-withdrawn patients remitted, which was defined as prolonged cessation of dystonic movements.

Specific therapies often involve dopamine depleting or anticholinergic drugs. Therapy with reserpine is typically started at a low dose of 0.25 mg daily and raised gradually to an average of 5 mg/day (occasionally to as high as 9 mg/day) [14]. Tetrabenazine has also been employed with some benefit. Anticholinergic treatment of tardive dystonia has yielded mixed results. Trihexyphenidyl, the anticholinergic studied most, is usually started at around 5 mg daily, and the dose is raised gradually until adequate benefit occurs or until distressing side effects intervene (33). Doses as high as 120 mg/day have been used, but the average dose of trihexyphenidyl for treating tardive dystonia is 20 mg/day (33). Ethopropazine, a phenothiazine derivative with antimuscarinic, anticholinergic effects, has also been used with some success in doses of 50–350 mg daily (33,78). The disadvantages of high-dose anticholinergic treatment are relapse after transient therapeutic effects, distressing anticholinergic side effects, and exacerbation of choreo-athetoid dyskinesia (97). Benzodiazepines such as diazepam, clonazepam, and lorazepam have sometimes been helpful (14). Occasional patients have benefited from baclofen, propranolol, bromocriptine, clonidine, carbamazepine, or ECT. Botulinum A toxin injections into affected muscles were found effective in a placebo-controlled trial in focal cranial dystonias such as blepharospasm and torticollis (48).

Burke (14) advocated dopamine antagonist treatment ("suppression") for either (a) severe generalized tardive dystonia causing muscle pain or muscle damage (with elevated CPK), or (b) instances where other drugs have failed. Clozapine may be a suitable alterative. Lieberman et al. (66) found that clozapine was particularly effective for TD patients showing dystonia. A published case report illustrated the effectiveness of clozapine (225–300 mg daily) in a 35-year-old man with severe tardive dystonia who failed to improve on reserpine 8 mg/d and benzatropine 4 mg/d (60). Clozapine has also been found effective in Meige Syndrome (95) which is characterized by blepharospasm and oro-mandibular dystonia, sometimes caused by NL.

Clozapine or other atypical NLs may emerge as the treatment option most likely to benefit tardive dystonia, particularly in the presence of an active psychotic disorder.

Tardive akathisia, consisting of both motor restlessness and subjective discomfort, is considered a subtype of TD (15). The treatment strategy is similar to that for TD and tardive dystonia. Neuroleptic discontinuation resulted in remission in a minority of patients only (8% in Burke's series) (14). The dopamine-depleting drugs reserpine and tetrabenazine in place of, or added to, NLs were found to be beneficial: of 15 patients treated with reserpine, three had full remission and eight showed marked improvement; tetrabenazine was effective in 7 of 12 (58%) patients (14). Beta-adrenergic blockers (such as propranolol), which show clear therapeutic efficacy in acute NL-induced akathisia, have produced mixed results in tardive akathisia. Occasional patients may improve on benzodiazepines, antiparkinsonian drugs, L-deprenyl, or clozapine. Successful treatment with clozapine in 3 patients with severe treatment-refractory akathisia was reported by Levin et al. (62). The remission of akathisia was maintained for 18–24 months with clozapine therapy.

Tardive myoclonus is characterized by sudden, brief, shock-like jerks occurring in NL-treated patients (14). It has also been described in clozapine-treated patients (5,8). The myoclonus tended to occur at clozapine doses >600 mg daily, was mostly seen on one side of the face or the other, and was often accompanied by drop attacks. The myoclonic jerks were considered by some to be related to myoclonic epilepsy and responded to treatment with carbamazepine (5,43). The concurrent administration of clozapine and carbamazepine is justifiable provided the white blood count is closely monitored (5). Clonazepam may also be effective for tardive myoclonus (14,94).

Tardive tics and Tardive Tourette's syndrome may be observed after chronic NL treatment. Burke (14), however, points to the difficulty of ascribing tics to NL therapy, because tics often appear and disappear spontaneously. No effective treatment has been identified.

The TD variants discussed above often coexist with typical TD, and—like typical TD—they do not have clearly effective therapies. Research into the treatment of TD variants may be particularly rewarding by elucidating the pathophysiological mechanisms of NL-induced movement disorders.


Partial support was obtained by NIMH grant #TDA-2 2 ROI MH32675-12.

published 2000