Tardive Dyskinesia: Epidemiological and Clinical Presentation

John M. Kane, M.D.

Tardive dyskinesia (TD) continues to be an important concern in the long-term use of antipsychotic (neuroleptic) medication. Considerable research has been conducted on the clinical presentation, epidemiology and risk factors associated with the development of TD. There is certainly increased awareness of the different of ways in which this condition can manifest itself and the extent to which it can produce a variety of disabilities, either directly or indirectly. (SeeThe Pharmacotherapy of Acute Anxiety: A Mini-Update and Issues in long-Term Treatment of anxity Disorders for discussions of pathophysiology and treatment respectively.)

It is critical to emphasize that any and all abnormal involuntary movements experienced by neuroleptic-treated patients should not necessarily be considered to be TD. There are numerous clinical conditions which may produce abnormal involuntary movements that may be mistaken for TD.

The word dyskinesia refers to a broad range of abnormal involuntary movements including chorea, athetosis and dystonia. There are a variety of abnormal movements which would not be considered dyskinesias, including: akathisia, compulsive movements, mannerisms, stereotypic movements, tics or tremors.

Neuroleptic drugs can produce dyskinetic movements that are not tardive (late occurring), but rather acute. Ayd (3) reported dyskinesias in 2.3% of patients early in the course of neuroleptic treatment. Dystonic reactions may be part of acute dyskinesia. These intermittent or sustained muscle contractions can involve the eyes, neck, diaphragmatic muscles, jaw, larynx or truncal musculature. This reaction responds well to acute treatment with anticholinergic or antihistaminic agents. A phenomenon known as tardive dystonia may be identical phenomenologically, but is late-occurring and is considered to be a subtype of TD. This will be discussed subsequently.

Dyskinesia may be produced or precipitated by a variety of drugs other that neuroleptics. Caffeine, phenytoin, estrogens and other drugs can produce reactions which are rapidly reversed when the drug is discontinued. Movements which mimic TD can frequently be seen with L-dopa treatment of Parkinson's disease and have been reported in isolated cases with an array of other drugs such as tricyclic antidepressants and antihistamines.

Disorders of the basal ganglia can produce an array of abnormal involuntary movements, including chorea, athetosis and dystonia. Examples include Huntington's disease, Wilson disease, Sydenham's chorea, and Fahr's syndrome. In most cases, an appropriate evaluation can identify such causes of abnormal movements. Since patients with Huntington's disease and Wilson's disease may have psychotic symptomatology, the differential diagnosis is particularly critical in the context of the present discussion. Since patients with Huntington's disease may receive neuroleptic drugs, and the possibility of concurrent abnormal movements with different etiologies exists. Patients with Huntington's disease may have marked postural instability and/or atrophy of the caudate nucleus on brain imaging, both of which would not be likely in true dyskinesia. Wilson's disease can be diagnosed, if suspected, with a serum ceruloplasmin.

Hyperthyroidism, hypoparathyroidism and systemic lupus erythematosus are also capable of producing dyskinesias. Tourette's disorder, torsion dystonia and spasmodic torticollis are all movement disorders which should be differentiated from TD based on history and clinical presentation. Meige syndrome is an idiopathic condition which can produce dystonic facial movements and blepharospasm. When these phenomena are seen subsequent to neuroleptic treatment, the term tardive facial dystonia should be employed.

Critical to any assessment of abnormal involuntary movement disorder in a patient receiving neuroleptics is a good history as to whether or not movements were present prior to the initiation of antipsychotic medication. Dyskinesias can occur which are not attributable to known causes. Some may occur in patients with psychiatric illnesses who have not been treated with neuroleptics or may be a manifestation of a conversion reaction. In addition, some elderly individuals may have so-called spontaneous dyskinesias with no obvious cause. Some individuals with ill-fitting or absent dentures also may have abnormal movements of the tongue or lips.

Prior to the introduction of neuroleptic drugs, there were a number of descriptions of abnormal involuntary movements occurring in patients with apparent schizophrenia (21,44). These descriptions included tic-like movements, grimacing, and stereotypic movements, as well as occasional choreiform movements and orofacial dyskinesias. However, choreiform or choreoathetoid movements were felt to be fairly rare in chronic psychiatric patients prior to the neuroleptic era (48).

Given problems in diagnostic reliability and validity of diagnoses of both psychotic and movement disorders in historical reports, attempts have been made to identify special populations which might help shed some light on this issue. Owens et al. (53) reported a prevalence of dyskinesia (defined as a rating of 2 or more on one or more AIMS items) of 53% among a group of patients with chronic schizophrenia who apparently had never been treated with neuroleptic drugs. This cohort of 47 individuals satisfied Feighner criteria for schizophrenia and had an average age of 67 years. In comparison, a 67% prevalence of dyskinesia was found among a somewhat younger (mean age 57 years) group of chronic schizophrenic patients who had received neuroleptic drugs.

Chorfi and Moussaoui (16) did not find any dyskinesias in never-medicated schizophrenic patients. Fenton et al. (22) reviewed detailed medical records for 100 lifetime neuroleptic-naive schizophrenic patients treated at Chestnut Lodge. Descriptions of abnormal movements were reviewed and rated by investigators who were blind to the treatment history of the patient. Fifteen percent of the records documented oral/facial dyskinesias with sufficient detail to be considered definite. If less stringent criteria are employed, 28% of the patient records documented some form of movement disorder.

Fenton et al. (22) reported that patients with spontaneous dyskinesia received significantly greater ratings on negative symptoms and conceptual disorganization than did those with no movement disorders. Owens et al. (53) have suggested that, in some patients, a syndrome of abnormal involuntary movements reflects the cerebral pathology underlying severe, chronic schizophrenia. Lidsky et al. (46) postulated that some repetitive motor acts exhibited by patients with schizophrenia could be a manifestation of an underlying dysfunction of the basal ganglia associated with the condition.

Yarden and Discipio (75) described a sample of young, drug-free, schizophrenic patients who exhibited a variety of abnormal involuntary movements (choreiform, athetoid, tics, stereotypies and mannerisms). After following-up these patients 2.5–3.5 years later, the investigators concluded that these patients constituted a group with early onset and a steadily progressive course. In addition, they showed marked thought disorder and poor response to pharmacologic treatment. Manschrek et al. (47), in studying disturbances in voluntary motor activity apparently unrelated to drug effects, also reported an association between movement disorders, cognitive impairment, affective blunting and formal thought disorder.

Chatterjee et al. (15) assessed the prevalence of extrapyramidal signs and spontaneous dyskinesia among neuroleptic-naive patients participating in the Hillside Hospital study of first-episode schizophrenia. Seventeen percent of this sample had extrapyramidal signs, but only one had spontaneous dyskinesia. Presence of extrapyramidal signs was correlated with more negative symptoms and poorer treatment outcome. In addition, the patients with extrapyramidal signs were more likely to develop Parkinsonian side effects after eight weeks of neuroleptic treatment.

These findings complicate suggestions that patients with negative symptoms, cognitive dysfunction and possibly relative abnormalities on neuroimaging are at increased risk for developing TD as some of the movements attributed to neuroleptic treatment may in fact be spontaneous dyskinesias associated with the schizophrenic illness itself. Waddington (69) has suggested that neuroleptic drug treatment may precipitate or hasten the appearance of dyskinesias in those people who may have ultimately developed the movements "spontaneously".

Waddington (70) has more recently modified his earlier focus on a simple disease-related organic vulnerability model and emphasized a disease process model suggesting that the greater are certain early signs of disease severity (particularly those that could reflect fronto-striato-pallido-thalamic dysfunction), the greater may be the risk of TD, and the subsequent confluence of movement disorder with disorganization, psychomotor poverty and cognitive dysfunction.

As suggested by the report of the American Psychiatric Association Task Force on Tardive Dyskinesia (40), the diagnosis of TD should not be solely one of exclusion but should also be based on satisfying some criteria. The following were suggested by the Task Force.

Nature of abnormal movements: Choreiform, athetoid, or rhythmic abnormal involuntary movements are reduced by voluntary movements of the affected parts and increased by voluntary movements of the unaffected parts.

Other characteristics: The abnormal movements increase with emotional arousal and decrease with relaxation and volitional effort. The movements are absent during sleep.

Specific localization of neuroleptic-induced TD: The tongue, jaw or extremities are involved in most cases.

Severity: At least "moderate" abnormal, involuntary movements in one or more body areas or "mild" movements in two or more body areas (face, lips, jaw, tongue, upper extremities, lower extremities and trunk) are present. Because of the variability in the manifestation of movements associated with TD, if the examination reveals movements that are only "minimal" or "mild" in only one body area, the examination should be repeated within one week to confirm their presence. Determination of the presence of these movements should be made using a standardized examination procedure and rating scale (e.g., the AIMS or the Rockland/Simpson Tardive Dyskinesia Rating Scale).

Duration of dyskinesia: The abnormal movements should be present continuously for at least four weeks.

Neuroleptic treatment: A history of at least three months of total cumulative neuroleptic exposure. Exposure may be continuous or discontinuous. Patients who fail to meet the criterion for duration of neuroleptic exposure should receive the appropriate diagnosis with the qualification "less than three months of neuroleptic exposure."

Onset of dyskinesia: The onset of the dyskinesia should be either while the patient is on neuroleptics or within a few weeks of discontinuing neuroleptics.

Antiparkinsonian agents: Antiparkinsonian agents usually have no effect or may even aggravate TD (although they may improve tardive dystonia).

Changes in neuroleptic doses: Increasing the dose of neuroleptics usually suppresses dyskinesia. Reduction or discontinuation of neuroleptics may worsen the symptoms temporarily (but are likely to result in improvement over the long term).

Tardive dystonia is viewed sometimes as a subtype of TD and sometimes as a distinct entity. Burke and Kang (8) have suggested that tardive dystonia is distinguished from the classic oral-buccal-lingual choreiform movements of TD not only by the dystonic nature of the movement, but also by the frequency with which it causes significant neurologic disability. The movements seen in tardive dystonia are not dissimilar to those presenting in primary torsion dystonia. The face and neck are the most frequently involved body regions. Some patients may exhibit aspects of both tardive dystonia and TD either concurrently or sequentially. Since the dystonic disorder is in many cases more disabling, the rational for neuroleptic discontinuation may be even greater. Tardive dystonia has been shown to benefit from anticholinergic medications, whereas in general TD does not (6).

Acute dystonic reactions usually occur in the first 24–48 hours of treatment with a dopamine D2 antagonist, whereas tardive dystonia tends to occur much later. In a series of 67 patients, Kang et al. (42) reported that tardive dystonia occurred after a median of five years of drug exposure.

There are several factors specific to TD that make accurate assessment a particular problem. Although diagnostic criteria have been developed, investigators may differ in the threshold they set for defining a case. On a cross-sectional assessment, it is obviously difficult to rule out other conditions discussed previously, and rating instruments or assessment measures cannot be utilized to make a diagnosis, though they may be useful as a screening tool.

There is considerable variability (even within patients) in the severity and presentation of abnormal movements over time. Some changes in severity may also be apparent in relation to changes in level of arousal or focus. Anxiety may aggravate the dyskinesias, whereas relaxation may improve them. However, these influences are not consistent across patients.

A critical source of variability in the presentation of TD is the role of medication, as changes in antipsychotic drug dosage may influence the severity of the movements. In addition, the administration of anticholinergic medication can influence the presentation of TD.

The most frequently employed methods of assessment are instrumentation, frequency counts, and multi-item rating scales.

Instrumentation (e.g., electromyography, accelerometry, force procedures or ultrasound) has been employed in an attempt to provide objective and automated assessments. Some studies have found quantitative instrumental methods to correlate with traditional rating scales. These techniques may also prove useful in helping to identify the early or subclinical case and to help differentiate TD from other movement disorders. For example, Paulsen et al. (54) reported that postural tremor of the hand and upper extremity was a significant predictor of limb-truncal (but not orofacial) TD in a prospective study of older patients.

Frequency counts of abnormal movements provide the number of movements occurring within a certain time frame. This method can be difficult to employ when multiple movements are present, and there are inherent problems in comparing patients who may have different types of movements. In addition, a frequency count may not be an accurate reflection of the overall severity of the condition in terms of body areas involved and the magnitude of the movements themselves.

Multi-item scales remain the mainstay in clinical research. A variety of scales are available (40). Many instruments have also been developed with a standard examination format for increased standardization. Brief multi-item scales can be employed in the context of routine clinical care to monitor patients for the development of abnormal involuntary movements. As emphasized previously, these instruments should not be used to make a diagnosis, though they can be used to screen patients and threshold criteria can be developed for defining the severity of movements necessary to be identified as a definite case.

A variety of problems remain in using rating scales. It is not always clear where one draws the line between normal and abnormal movements (e.g., occasional tongue protrusion or movements of the lips may not constitute TD). The perceived severity of movements could be based on quality, frequency, duration or amplitude of the movement and it is far from clear how to integrate these different aspects.

Satisfactory levels of inter-rater reliability have been demonstrated for several widely used scales. Attempts to validate scales have not been extensive, particularly since there are no established criteria for the condition that could be used to demonstrate concurrent validity.

Special populations present particular challenges in assessing abnormal involuntary movements. As previously discussed, among patients with schizophrenia, some proportion of abnormal involuntary movements may be spontaneous. This also applies to elderly populations, where a relatively high prevalence of a variety of neuromedical conditions may also contribute to the presence of abnormal movements.

Antipsychotic medications are frequently employed in those with developmental disabilities. In this population of patients, a variety of abnormal movements (stereotypies, mannerisms, adventitious movements) may be seen that are unrelated to exposure to medication. Clearly, the training of raters in the assessment of abnormal movements takes on an added importance when special populations are involved.

Some debate is ongoing as to what extent neuroleptic treatment is either necessary or sufficient to produce abnormal involuntary movements. The consensus at present remains that antipsychotic drugs do play an important role in producing or evoking abnormal involuntary movements. Studies of the epidemiology of TD have had to confront a variety of methodological problems, including differential diagnosis, fluctuations in symptomatology, the potential masking effect of antipsychotic drug treatment, and the absence of validating criteria.

Numerous prevalence surveys of TD have been conducted and extensively reviewed (31,36). Prevalence estimates have been useful in identifying populations at particular risk and providing some sense of the overall scope of the problem. Clearly, prevalence estimates will vary depending upon the population studied (e.g., their age, sex, length of neuroleptic treatment), the assessment strategy utilized, and the criteria employed to define caseness.

Yassa and Jeste (79) reviewed 76 published studies on the prevalence of TD, representing a total of 39,187 patients. The reported prevalence ranged from 3–62%, with a mean of 24%.

Woerner et al. (72) completed a large-scale prevalence survey designed to address a variety of methodological concerns by including a wide range of clinical populations, while employing the same raters, assessment techniques and definition of caseness throughout. The prevalence of presumptive TD varied enormously across the different populations: 13% in a voluntary hospital with a relatively young patient population; 23% in a Veterans Administration Hospital; and 36% in a state hospital. The proportion of apparent cases of TD that could be attributed to other neuromedical conditions (following clinical laboratory evaluation and examination by a neurologist experienced in the assessment of movement disorders) was surprisingly low (0.1%). A somewhat larger group (3.7%) had abnormal movements in the context of a variety of neuromedical conditions that might have played an etiologic role. As a result, the most conservative estimate of overall prevalence among the 1,441 patients examined was 19.6%.

In order to estimate the rate of false negative cases due to masking by the neuroleptic, a subgroup of patients with no evidence of TD was withdrawn from neuroleptic drugs and examined weekly for three weeks. Seventy subjects were discontinued; of these, 24 (37%) showed emergent dyskinesias. The rate of emergent dyskinesia at the state facility was strikingly higher (67%) than at the voluntary hospital (18%) or the VA facility (17%).

Patients discontinued from fluphenazine decanoate (a minimum of five weeks from the last injection) were less likely to show dyskinesia than patients discontinued from oral medication, confirming a previous finding by Levine et al. (45). The dose of antipsychotic medication prior to drug discontinuation was not related to the emergence of TD; however, a history of treatment with very high doses (3,000 mg/day in chlorpromazine equivalents) was negatively correlated with covert dyskinesia. Age and total months of antipsychotic drug treatment were found to be positively correlated with emergent TD.

Interpretation of the results of prevalence surveys, especially with regard to analysis of potential risk factors, is complicated by the fact that prevalence estimates are influenced by the persistence of the condition within a given population. Many cases of TD will remit over time, particularly with drug discontinuation or dosage reduction, while others persist. Therefore, any analysis of risk factors in a prevalence survey will be complicated by those factors that influence course. If, for example, advanced age is a risk factor for the development of TD and a risk factor for the relative persistence of TD, then the prevalence would be particularly high in an elderly population.

The approach involved in prevalence surveys has definite limitations, and further progress in understanding and preventing TD will require more sophisticated methodology.

Several prospective studies have been conducted in the past decade, and this strategy overcomes many, but not all of the methodological problems inherent in cross-sectional prevalence surveys.

Kane et al. (38,39) have reported interim results from a long-term, prospective study of TD involving more than 850 patients. The design of the study allows for reassessment every three months, and patients are selected without regard to their psychiatric diagnosis or history of antipsychotic drug treatment. A subgroup of approximately 100 patients recruited into the study have never received antipsychotic drugs. The average age of the cohort at entry was 29 years, and 43% were female. Of those individuals who did have a history of antipsychotic treatment at study entry, the median length of lifetime exposure was 12 months. Therefore, patients are followed from a relatively early stage in their drug treatment experience. The findings regarding incidence thus far indicate that the cumulative incidence of TD is 5% at one year, 10% at two years, 15% at three years and 19% after four years of antipsychotic drug exposure. For the fifth and sixth years, the figures have continued to increase, to 23% and 26% respectively. These results certainly support the notion that an increasing duration of antipsychotic drug exposure is associated with an increasing cumulative incidence of TD (at least for the first several years of such exposure). It remains to be seen whether or not the risk decreases after some point in time.

Yassa et al. (76,77) have carried out a prospective study involving 108 patients (55 men and 53 women) who were assessed over a two-year period. Sixty-five percent of the patients had a diagnosis of schizophrenia. In eight patients, the TD persisted through at least two separate examinations at some time during the two years. The condition was considered to be of moderate severity in one case and mild in the remainder. An incidence of 7.4% for persistent TD was reported by the investigators, but this figure was not based on a life-table analysis or a cumulative proportion remaining free of TD, which would allow inclusion of the dropouts in the incidence calculation.

Glazer et al. (30,49) have reported prospective data on a cohort of 362 chronic psychiatric outpatients who were free of TD at baseline and were maintained on neuroleptic medication. Many of these patients had lengthy neuroleptic treatment prior to baseline (mean: 8 years). However, the investigators made every attempt to rule out prior evidence of TD.

This study demonstrated an average incidence rate of 5% per year during the five-year follow-up, with a cumulative incidence of 20% at the end of the five years. By combining data from patients with different durations of neuroleptic exposure at baseline, the investigators also estimated the risk of developing TD for exposure periods that exceeded the observed follow-up duration of five years. Using this strategy, they estimated the 10-year risk to be 49% and the 25-year risk to be 68%.

Chakos et al. (14) reported prospective data involving 118 patients who participated in the Hillside Hospital study of first-episode schizophrenia. Seventy-three percent of the patients had no prior antipsychotic drug treatment, and the remainder had less than 12 weeks of exposure at study entry. Patients received standardized antipsychotic drug treatment and were evaluated for up to 8.5 years. The cumulative incidence of presumptive TD 6.3% after one year and 11.5% after two years. These data underscore the fact that TD can occur relatively early in treatment.

Saltz et al. (59) reported on a prospective study of TD development in a large cohort of elderly subjects. The mean age of the study group was 77 years, and none of the subjects had any evidence of abnormal involuntary movements prior to antipsychotic drug treatment. Fifty-eight percent of this sample received a diagnosis of organic mental syndrome, and 42% had a primary psychiatric diagnosis. The incidence of TD observed after 43 weeks of cumulative antipsychotic drug exposure was 31%.

Jeste et al. (35) have reported data on 266 elderly patients followed prospectively. Twenty-one percent of the patients received a diagnosis of schizophrenia; 24% Alzheimer's disease; 21% affective disorders; and the remainder had miscellaneous diagnoses. The mean age of this cohort was 66 years, and patients were enrolled early in the course of neuroleptic treatment. Twenty-five percent were neuroleptic-naive, while 45% had less than three months of total lifetime exposure at baseline.

The overall cumulative incidence of TD was 26% by the end of one year and 52% and 60% after two and three years, respectively. The investigators found no significant difference among the four diagnostic groups in terms of TD incidence.

The elucidation of factors that contribute to individual vulnerability to TD should be of help in the development of preventive strategies. However, there is a paucity of data from long-term prospective studies which could test the predictive value of potentially relevant risk factors.

Age is one of the most consistently replicated risk factors, particularly in prevalence surveys. Extensive reviews of the topic have concluded that increasing age remains the most consistently implicated factor for the risk of TD development. In addition, advanced age is associated with the development of more severe and persistent types of the disorder. Data from the Saltz et al. (59) and Jeste et al. (35) prospective studies confirm a strikingly high incidence of TD within the first year of neuroleptic treatment. The prospective study reported by Glazer et al. (30) and Morgenstern and Glazer (49) also found that age was a risk factor for TD occurrence.

Clearly, the elderly are also at risk for the development of spontaneous dyskinesias or abnormal involuntary movements associated with a variety of neuromedical conditions independent of their exposure to antipsychotic drugs. At the same time, prevalence surveys of healthy elderly volunteers suggest that spontaneous dyskinesias are not that common in the absence of some form of brain disease or mental disorder, even in late life (36,72).

The potential mechanism(s) mediating the relationship between age and risk of TD remains speculative. It is possible that neuronal damage, degeneration, age-related changes in receptor number, sensitivity or plasticity, and the reduced efficiency of restorative processes may be relevant.

Studies of neuroleptic effects in children indicate that they are also vulnerable to TD and even persistent dyskinesias (9).

Sex differences in the prevalence of TD have been reported by a number of investigators. Yassa and Jeste (79) performed a meta-analysis of published reports that provided prevalence data in men and women. The mean prevalence rate was found to be 26.6% in women and 21.6% in men. In addition, women were found to have more severe TD and a higher prevalence of spontaneous dyskinesia in comparison to men. They also found an interaction between age and sex as suggested by numerous individual prevalence surveys. The prevalence of TD appeared to reach its peak in the sixth and seventh decades among men, whereas it continued to rise after the seventh decade in women. It remains unclear as to what extent these sex differences are a result of biological variables or how much they are due to external risk factors, such as differences in neuroleptic treatment, dosage or duration.

As previously discussed, a variety of abnormal involuntary movements have been described in psychiatric patients, particularly those with schizophrenia, long before the introduction of antipsychotic medication.

Another line of evidence supporting the possibility that schizophrenia includes motor disturbances is the occurrence of abnormal movements in amphetamine-induced psychosis. Chronic use of amphetamines can produce a paranoid state almost indistinguishable from schizophrenia, in which patterns of perseverative, stereotyped behavior and grimacing, chewing, and twisting of the trunk and limbs may occur (60).

Tardive dyskinesia may be more common in patients with predominant negative features (4,34,65). There are a variety of caveats that should be kept in mind in drawing firm conclusions regarding this association. It is particularly important to recognize the inherent difficulty in differentiating negative symptoms from depression or drug induced Parkinsonism, especially on a cross-sectional basis. It may be particularly complicated if affective symptoms and/or vulnerability to drug-induced Parkinsonian symptoms convey a higher risk for the development of TD among neuroleptic-treated patients.

In a prospective study of elderly neuroleptic-treated patients, Saltz et al. (59) found a higher incidence of TD in elderly individuals with psychiatric diagnoses, compared with elderly individuals having organic brain syndromes. Jeste et al. (35), however, did not. This finding runs counter to the notion that "organicity" increases the risk for TD, but supports the notion that some psychiatric disorders may predispose to the development of abnormal involuntary movements following neuroleptic treatment. The dichotomy is outmoded in that schizophrenia is also now considered to be a largely "organic" disorder.

Several investigators (12,57,58) have suggested that patients with affective disorder may be at increased risk for development of TD. This may be particularly true among patients with recurrent unipolar depressions. It is important to keep in mind that a diagnosis of depression may also be associated with intermittent antipsychotic drug administration, pharmacotherapy with tricyclic antidepressants or alcohol abuse, all of which may increase the vulnerability to TD. Even within the population of patients with schizophrenia, some evidence exists that the presence of a familial history positive for affective disorders may increase the risk of TD (56,71).

Morgenstern and Glazer (49) reported that nonwhite patients (largely African Americans) were nearly twice as likely to develop TD. This did not appear to be accounted for by differences in dosage or types of medication utilized. However, this finding has not emerged in many other studies.

Enormous attention has been directed toward the question of whether or not any specific antipsychotic drug or drug class differs from others in the propensity to produce TD. Unfortunately, a great deal of this attention has focused on data derived from animal models and preclinical studies that may not be entirely relevant to the clinical situation. At the present time, there are no compelling data from prospective clinical studies that any currently available antipsychotic drug has a lower risk for TD, with the possible exception of clozapine. Retrospective studies are difficult to interpret, since such factors as polypharmacy, noncompliance, co-administration of other drugs, etc, remain a problem. Some reports suggest that depot injectable forms of fluphenazine may carry an increased risk of TD (27,61). However, this could be a result of increased antipsychotic drug exposure due to guaranteed medication delivery overcoming noncompliance. There are many other difficulties in arriving at meaningful conclusions regarding relative risk with a specific drug. The ideal design for such studies would involve random assignment of patients to treatment groups, systematic evaluation, and control of such variables as age, length of drug exposure, and dose equivalents. Pharmaceutical companies have tended to shy away from conducting prospective studies involving the risk of TD development ,since the assumption has been that hundreds of patients would be required, and a very lengthy follow-up would be necessary. Based on previously discussed results from prospective studies, it would appear that relatively high-risk groups (e.g., elderly individuals with early signs of drug-induced Parkinsonism) could be recruited and might allow the development of a relatively rapid comparison between an experimental and a standard medication.

The best case for a reduced liability in producing TD has been made for sulpiride and clozapine. For both compounds, a variety of theories have been put forward to explain this potential advantage. The best claims have been made for clozapine. Recent data in the United States indicate that more than 120,000 patients have been treated with clozapine, and approximately 68,000 are currently receiving the medication (Michael Krassner, October 1996, personnel communication). To date, no confirmed cases of definite TD attributable to treatment with clozapine have occurred among these patients.

Casey (13) also concluded that this compound produces a very low incidence of Parkinsonism, and there are virtually no reports of akathisia or dystonia. Clozapine is associated with a relatively high rate of agranulocytosis. Experience in the United States to date reveals a cumulative incidence of 0.8% for this reaction after one year of treatment (2).

Although it must be true that at some level there is a significant relationship between neuroleptic dosage, duration of treatment and the risk of TD, it has been a very difficult relationship to establish. Since most patients do not develop TD, it may be difficult to establish such a relationship. It might only be apparent in those individuals who are in fact vulnerable to the disorder.

The overwhelming majority of cross-sectional studies and follow-up studies that have addressed this issue have not been able to identify a clear association between the risk of TD development and drug variables such as duration, total amount of drug administered, or type of drug or current dosage. One study (37) did suggest the possibility of a reduced incidence of TD among patients receiving very low doses of neuroleptic maintenance treatment, but this finding has not been consistently borne out in other studies of dosage reduction or intermittent treatment. Morgenstern and Glazer (49) did report a positive dose-response relationship for the development of TD in their prospective study.

Kane and Smith (36) pointed out that most studies reporting a significant relationship between TD and medication dosage or duration have included samples of patients in the relatively early years of their drug treatment history. It is possible that the relationship between these variables and TD risk is more apparent or more of a risk factor in the relatively early phase of development , as opposed to those cases which occur after 5–10 years of cumulative neuroleptic exposure, although the findings of Morgenstern and Glazer were among largely chronically treated patients. It is also important to keep in m that higher doses of neuroleptic drugs may also produce a masking effect in some patients by reducing abnormal involuntary movements. In fact, in the Morgenstern/Glazer study, a dose increase between visits tended to lower the rate of TD detected at the later visit.

Several recent studies have employed an intermittent or targeted strategy to attempt to reduce cumulative neuroleptic exposure. One important outcome measure in these studies was the potential impact on TD. To date, none of these studies has demonstrated significant advantages for intermittent treatment in reducing the risk of TD (41).

In exploring the relationship between dosage, duration and TD, risk analyses are also complicated by the fact that there are enormous individual differences in the absorption and metabolism of antipsychotic drugs. Therefore, patients receiving similar oral doses of a specific drug may have very different levels of active drug in relevant brain areas. To date, results are not consistent across studies in suggesting that patients with TD experience higher blood levels following equivalent doses of antipsychotic drugs. Most of the studies have not found a significant correlation (20,32). Some other investigators, however, have reported that patients with TD had higher serum drug levels (33,80). A comparison between prescribed dosage regimens in oral form is a far cry from an accurate reflection of the drug exposure at a specific receptor. Is not surprising, therefore, that relationships between drug treatment variables and TD have been difficult to demonstrate. It remains unclear as to whether or not drug blood levels are a critical predictive factor, but, as indicated in the previous discussion of dosage and duration, it may be that we have not applied adequate methodology to establish the point in time and for which patients such a relationship might exist.

Many observers have noted that the addition of anticholinergic medication can exacerbate existing TD, and that discontinuing anticholinergic drugs may improve the condition. The observation has led to the hypothesis that these compounds might also contribute to the development of TD. However, no clear-cut evidence has been presented that an increased incidence or prevalence of TD is seen among patients chronically treated with antiparkinsonian drugs (27). It may also be possible that any association observed between vulnerability to TD and anticholinergic administration is really an epiphenomenon, since those individuals who are vulnerable to drug-induced extrapyramidal side effects may also have an increased risk of developing TD. Since such patients are probably more likely to receive anticholinergic medication, this relationship may suggest anticholinergic drugs as a risk factor when in fact it is the extrapyramidal side effects that are increasing relative risk (36).

In the case of TD, it is clearly difficult to make inferences regarding etiology based on the effects of a particular manipulation on preexisting TD. The ability of antipsychotic drugs to mask the condition is an example, since we assume antipsychotic drugs play a critical role in etiology. The chronic administration of cholinergic receptor antagonists has been shown in animal models to increase antagonist -induced supersensitivity of dopamine receptors. There is also considerable difference among commonly used anticholinergic drugs in their relative affinities for different subtypes of muscarinic receptors.

Some animal studies have explored the effect of concomitantly administered lithium in models of TD (55). The results from these studies have been somewhat mixed, however, and it is difficult to draw conclusions. There are very few clinical data available, but Kane et al. (38) did find an effect of concomitant lithium administration in reducing the incidence of TD in a group of patients receiving neuroleptic treatment.

Some investigators have found that, among neuroleptic-treated patients, smoking was associated with an increased prevalence of TD. It is possible that nicotine stimulates dopamine release from nigrostriatal neurons (78).

Several investigators (18,19,35,52) have reported an increased prevalence of TD in subjects with a history of alcohol abuse. The mechanisms underlying this possible risk factor are unknown.

Several investigators have suggested that diabetes may increase the risk of TD (23,73,50). The possible mechanism underlying this apparent increased susceptibility remains unclear.

Extensive evidence is now available from neuropathological, neuroimaging and neuropsychological studies suggesting the potential importance of abnormalities in brain structure. The most consistent findings are a 5–10% reduction in cortical gray matter volume and overall brain volume. Statistically significant reductions in volume of the temporal lobe cortex, especially the mesial cortices in the region of the hippocampal formation and the lateral cortices of the superior temporal gyrus, have been described in many studies (8). Reduced volume in other cortical areas has been reported less frequently, and some also show abnormal morphometry in subcortical areas such as the basal ganglia and thalamus.

Given this, a number of investigators have hypothesized that such abnormalities, particularly in the basal ganglia, might bear some relationship to vulnerability to and/or the development of abnormal involuntary movement disorders such as TD. Overall, the findings on the relationship between abnormalities on brain imaging and the prevalence of TD have been mixed (1,43,63), and a variety of methodological problems complicate the drawing of firm conclusions. Waddington (68) has raised the possibility that associations between structural brain abnormalities and measures of abnormal involuntary movements may be more robust for orofacial dyskinesias, as has been suggested for some other risk factors.

Magnetic resonance imaging not only allows for high resolution structural images, but relaxation parameters can provide information in tissue characteristics such as hydration status, tissue lipid concentration or iron content.

Iron deposition has been studied in relationship to neuroleptic side effects such as akathisia and TD with mixed results. Buckley et al. (7) reported that, as a group, schizophrenic patients showed evidence of somewhat more prolonged T2 relaxation times in some regions of the basal ganglia, but there was no significant difference between patients with and without TD. More sophisticated neuroimaging techniques have not as yet provided any consistent, meaningful evidence of underlying cerebral abnormalities in TD.

The relationship between a variety of measures of cognitive functioning and TD has been the subject of several investigations. These considerations are obviously complicated by the fact that some degree of cognitive dysfunction is found in a substantial proportion of patients with schizophrenia. The question as to whether or not such dysfunction is more common in those patients who are most vulnerable to the development of TD has been a difficult question to pursue. It is also possible that those schizophrenic patients with cognitive dysfunction (and negative symptoms, etc.) are overall more vulnerable to developing TD, and that cognitive dysfunction represents one aspect of a particular schizophrenia subtype.

Waddington (65) reviewed 11 studies in which patients with schizophrenia underwent neuropsychological testing, and contrasts were made between those with and those without TD. These investigations involved a variety of test measures and very heterogeneous patient populations; however, in eight of these reports, evidence supported an association between cognitive dysfunction and TD. In many of these studies, the investigators failed to control for such potentially important variables as age and psychiatric diagnosis. Waddington et al. (65) also reported a significant relationship between TD and cognitive dysfunction in neuroleptic-treated bipolar patients.

Most of those studies that have included a wide range of cognitive functions in their assessment methodology failed to detect significant differences between individuals with and without TD (5,62). At the present time, it is difficult to draw firm conclusions regarding the relationship between cognitive dysfunction and drug-induced movement disorders.

The outcome of TD appears to be very heterogenous. Many patients, particularly younger individuals, will experience considerable improvement over time (10,74). In general, follow-up studies suggest that TD does not commonly progress, but tends to follow a fluctuating course with an overall trend towards improvement (26). Time may be the most important factor in outcome. The length of follow up in most studies is positively correlated with increasing improvement or remission. Therefore, studies examining outcome after more than five years show results that are more favorable than those obtained from shorter follow-up assessments. If drug treatment is stopped or if medication dosage is reduced, the TD has a more favorable outcome (12).

Kane et al. (unpublished data) have assessed the outcome of TD developing among patients followed in the prospective study previously alluded to (38). In this study, 144 patients with a mean age of 29 years were followed for an average of 4.3 years after the development of TD. Approximately 60% showed remission in TD at some point during the follow-up period. However TD can have a fluctuating course, and it is possible that some patients who experienced a remission will be at risk for subsequent reemergence of TD. Approximately 40% of patients had remission of TD with no recurrence after five years of follow up, Age, gender, diagnosis, and length of neuroleptic treatment prior to the diagnosis of TD bore no relationship to the outcome. On the other hand, the proportion of time on medication following the development of TD and the modal dose of neuroleptic following the diagnosis of TD did appear to be associated with outcome. Interestingly, those patients who were either on neuroleptics all of the time or less than 50% of the time had a significantly better outcome than those patients who were on neuroleptics between 51% and 99% of the time. This might suggest that intermittent treatment or on/off medication manipulations are associated with a worse outcome than either continuous treatment or relatively long-term drug withdrawal. This association will require further study, but it has important implications for the clinical management of TD, particularly in the context of those patients who are likely to benefit from or require continued neuroleptic treatment.

Glazer et al. (28,29) reported the results of a retrospective follow-up study of 192 patients seen two or more times during a 3–55-month period while continuing to receive neuroleptics. One hundred and twelve subjects (58% of the sample) demonstrated a "chronic persistent" pattern, in contrast to an intermittent pattern in the remainder of the patients. These investigators found that the most important predictors of chronic persistent TD included increased age and the presence of non-orofacial dyskinesia at baseline. Glazer et al. (28,29) reported results on a separate cadre of TD patients who were withdrawn from neuroleptics and were reexamined monthly for a mean of ten months. These investigators reported complete and persistent reversibility of TD in only 2% of these patients, although many patients showed substantial improvement without complete remission.

Gardos et al. (24) reported on a group of 63 neuroleptic-treated outpatients in Hungary who were assessed for TD over a ten-year period. The overall prevalence changed little over time—30% at baseline, 37% at five years, and 32% at ten years. However, 11 of 19 patients who had TD at baseline were in remission at year 10, and 12 of 44 patients who did not meet research criteria for TD at baseline had developed it at 10 years.

Chouinard et al. (17) reported on a 10 year reassessment of 98 patients and found the prevalence had increased from 22% at baseline to 44% at five years and 56% at 10 years.

Richardson et al. (56) reexamined 223 chronic inpatients after an average of 5.8 years. The prevalence of TD remained fairly stable; 90 patients continued to show involuntary movements, 64 remained free of TD, 30 patients were in remission, and 39 new cases developed. These data support the notion that TD can improve and even remit over time, even with continued neuroleptic administration. This is an important aspect of assessing the benefit-to-risk ratio of continued medication in a patient who develops TD.

If medication is indicated, the lowest effective dose should be administered, but complete discontinuation is likely to lead ultimately to psychotic relapse in most patients with schizophrenia (see Psychosocial Predictors of Outcome in Depression, this volume).

In moderate or severe cases, switching to clozapine has been shown to be of value. The role of other putative atypical antipsychotics in this context has yet to be established. (Treatment of TD is discussed in Issues in long-Term Treatment of anxity Disordersthis volume.)

Given the promise offered by clozapine and other newer medications, it will be essential to conduct adequately designed and controlled tests to establish the relative merits of new compounds in this regard.

As new compounds with more specific effects on particular receptors become available, new knowledge with regard to mechanisms of drug action will hopefully be forthcoming. It may also be particularly valuable to study those patients who appear invulnerable to TD despite years of neuroleptic treatment. These individuals provide many clues to pathophysiology and risk factors.

At present, there are no proven safe and effective treatments for TD. Therefore, identification of risk factors and the development of preventive strategies remain important goals.

This work was supported by NIMH grants MH-32369; MH-40015 and MH-41960.

published 2000