Schizophrenia is one of the most severe and debilitating major psychiatric diseases, with a yearly prevalence of 0.4–0.8% in the general population (Dohrenwend, 1980; Flekkoy, 1987) and a lifetime risk of 1–1.5% (Regier et al, 1984; Robins et al, 1984). It is the prototype of a severe mental illness, with the capacity to disrupt routine daily functions in all areas of life, but especially work, social relationships, and self-care. Schizophrenia also increases the vulnerability of a patient to physical and other mental disturbances (Koran et al, 1989). Until recently, the clinical course of this disease has been far worse than that of most other psychiatric disorders, thus making it devastating for many individuals and costly to society (McGlashan, 1988). Schizophrenic patients occupy 20–25% of all the beds available for psychiatric in-patient care and account for 40% of all long-stay hospital days (Talbott et al, 1987; Meise et al, 1992). It is not so much the direct costs associated with treatment, but rather the indirect costs, such as those that arise from morbidity and mortality, that place a severe burden on the social economy (Gunderson & Mosher, 1975; Hall et al, 1985; Rice et al, 1992).

The drugs currently used to treat the negative symptoms of schizophrenia are inadequate, and the methods used to evaluate their efficacy impractical and conceptually flawed. I will attempt to clarify these issues, but they are complex, and the solutions I suggest will not be completely satisfactory.

The primary basis for the action of neuroleptic drugs has been suggested to be a blockade of D2 receptors in the mesolimbic system, with subsequent decrease in the firing rate of ventral tegmental (A10) dopamine neurons by the process of depolarisation inactivation (Matthyssee, 1974; Bunney et al, 1991). The main evidence for this hypothesis is that: the affinities for the D2 receptor of all effective antipsychotic drugs are positively correlated with their average clinical dose (Seeman & Lee, 1975; Creese et al, 1976); and chronic administration of antipsychotic drugs produces nearly complete inhibition of the firing of ventral tegmental (A10) dopamine neurons that project to the limbic forebrain (Chiodo & Bunney, 1983). Clozapine, the prototypical atypical antipsychotic drug, also satisfies both criteria (Seeman & Lee, 1975; Chiodo & Bunney, 1985). However, the demonstration that clozapine is more effective than other neuroleptics for the treatment of both schizophrenic patients who are responsive to typical neuroleptics (Meltzer, 1992) as well as those who are resistant (Kane et al, 1988) suggests that this simple dopamine hypothesis of neuroleptic drug action is insufficient. Indeed, there is evidence that clozapine at clinically effective doses produces less D2 receptor occupancy and, hence, less antagonism of D2 receptors in the striatum, and probably also in the limbic system, than typical neuroleptic drugs (Farde et al, 1992); this further challenges the adequacy of the dopamine hypothesis to explain the greater efficacy of clozapine. This leaves the need to consider what else besides a D2 receptor blockade may explain the action of clozapine and, indeed, whether limited blockade is superior to a more complete one.

Since neuroleptic drugs were introduced in the 1950s they have become the primary mode of therapy for managing both acute and chronic psychoses. Although some antipsychotic drugs are free of disturbing extrapyramidal side-effects (EPS), most have a very narrow therapeutic/toxic index. Clozapine is the best example of an antipsychotic with a wide separation between antipsychotic actions and liability to EPS, but its side-effect profile of sedation, salivation, seizures, and agranulocytosis are factors that limit its use, and clearly indicate the need for further advancement in neuroleptic therapy.

Several types of neuroreceptors are of interest with respect to antipsychotic activity, in particular the D2, D1, and 5-HT2 receptors. Among currently prescribed antipsychotic drugs, some have an affinity for a broad range of neuroreceptors, while others are more selective for the D2 receptor (Hytell et al, 1985). The most widely accepted hypothesis of neuroleptic drug action is that antipsychotic effects are mediated by a blockade of the dopamine receptors (Carlsson & Lindqvist, 1963; van Rossum, 1966; Creese et al, 1976; Seeman et al, 1976; Peroutka & Snyder, 1980). This hypothesis has been supported by consistent findings of high D2 receptor occupancy in positron emission tomography (PET) studies of patients treated with antipsychotic drugs (Farde et al, 1986; Smith et al, 1988; Baron et al, 1989). At the same time, the risk of extrapyramidal side-effects (EPS) seems particularly high in patients with occupancy above 80% (Farde et al, 1992) (Fig. 1).

The therapeutic effects of ‘classic’ (typical) antipsychotic agents lie in their ability to block central dopaminergic receptors – a property that is also responsible for the frequent occurrence of undesirable extrapyramidal side-effects (EPS). In contrast to these typical agents, clozapine alone has distinguished itself in humans – by virtue of its enhanced antipsychotic action and lack of concurrent EPS – as an atypical antipsychotic. However, the use of clozapine has been limited by the occurrence of agranulocytosis and, to a lesser extent, seizures (Alvir et al, 1993; Haring et al, 1994). The mechanism underpinning the atypical profile of clozapine remains elusive. One hypothesis suggests that it lies in clozapine's higher serotonin 5-HT2: D2 binding ratio, when compared with typical agents – a factor being considered as a predictor of atypicality (Meltzer, this issue; Meltzer et al, 1989). However, an emerging view is that it is not a single pharmacological action, but rather multiple properties that may define an atypical, clozapine-like compound (Lieberman, 1993).