CHAPTER TWO

The psychopharmacology of aggression

In this chapter, we review some of the aspects of the psychopharmacology of aggression. Anatomical topics are introduced first as they are relevant to the biochemical actions of various drugs on aggression. The importance of 5-HT is apparent and a review is presented of the effects of serotonergic drugs on aggression. An outline is also given of the clinical actions of various psychotropic drugs on aggression.

BRAIN MECHANISMS

One of the major attempts to understand the role of brain mechanisms in the regulation of aggressive behaviour was made by Moyer (1976). He suggested that in both humans and lower animals, functionally different structures in the CNS were related to different types of aggression. Thus, Moyer (1976) suggested a scheme to explain aggression that would involve: (1) a number of innate systems of neural organisation in the brain, with one such pattern for each kind of aggression; (2) activation of the innate systems by appropriate stimuli; and (3) a system for the generation of arousal (the reticular activating system) that affects both the organism's reactivity to the aforesaid stimuli and the intensity of the aggression induced by the innate neural system. This innate system would also be sensitive to other systems in the body, and would be linked to specific patterns of motor behaviour by which the aggression is acted out (Moyer, 1976).

This classification has been criticised by Herbert (1993) as confounding stimuli, contexts, and functions. In applying reductionism to the experimental study of any behaviour, we risk neglecting important contextual cues, but this may be more important with aggression as Herbert (1993) claims that it usually occurs as part of another behaviour. Aggressive behaviour is not a goal in itself but is used as a means to an end. Aggressive behaviour is fairly stereotyped and species-specific, which indicates that it must be organised by a neural system, but in trying to seek this system it is important to recognise the other behaviour of which aggression is a part. As aggression is an optional component, the mechanism involved may determine whether it is appropriate in the circumstances. Herbert (1993) entitles this "a tactical decision".

Efforts have been made to fit specific areas of the brain to the expression of aggressive behaviour. Various parts of the limbic system have been implicated, in particular the amygdala. Both the temporal lobe cortex and the medial limbic structures such as the hypothalamus are linked to the amygdala. Lesions of the amygdala in both primates and humans produce calming or "taming" effects. Conversely, electrical stimulation of the amygdala or hypothalamus can precipitate aggressive responses.

Data related to the involvement of the amygdala in aggression in humans come from reports dealing with surgical intervention aimed at the relief of organic brain disease not responsive to other clinical treatment. Control of violent and homicidal outbursts in previously intractable psychiatric patients after amygdalectomy was claimed in many studies in the 1960s and 1970s. However, a review of these studies (O'Callaghan & Carroll, 1982) shows many to be flawed and the overall results for amygdalectomy to be unimpressive. There are grave doubts about the specificity of operative effects. Surgery often results in a general change and any positive effects on aggressive behaviour may occur in the context of a general dampening of function.

Nevertheless, animal work has continued to link the amygdala to aggressive behaviour. Neurotoxin-induced lesions in the basolateral nuclei have been shown to reduce aggressive behaviour in the rat (McGregor & Herbert, 1992). The amygdala has been represented as linking cognitive and limbic function, thus allowing affective value to be assigned to stimuli. LeDoux (1989) suggests that there may be various pathways leading to different levels of processing. The projections from the thalamus would be involved in processing the affective significance of simple sensory cues, whereas those from the cortex would be involved in the processing of more complex stimuli. These would be allied to Lazarus' primary and secondary appraisal patterns (Lazarus, 1991), i.e. first is there a threat?, and second, how can I cope? It has been suggested that the amygdala contains the neural representation of fear. If amygdaloid lesions reduce fear and therefore threat, then the likelihood of an aggressive response may be reduced. The amygdala may then be involved in "tactical decisions" on the use of or response to aggressive behaviour (Herbert, 1993).

Animal work has also linked the temporal lobe to aggression. In rats, increased aggression is seen with experimentally induced foci in temporal lobe structures (Pinel, Treit, & Rovner, 1977). In humans the association of the temporal lobe and aggression has come from studies of patients suffering from epilepsy (Devinsky & Bear, 1984; Mark & Ervin, 1970; St. Hilaire, Gilbert, Bouvier, & Barbeau, 1980). Violent aggressive behaviour reported concomitant to seizures is usually stereotyped, unsustained, and not purposeful in nature (Delgado-Escueta, Mattson, & King, 1981). Ictal aggression is rare. However, it depends on the thought content of the patient at the time and is thus much more likely to be a problem in the community than when being carefully recorded in hospital (Fenwick, 1986).

It has been suggested that the relationship between aggression and epilepsy is a consequence of temporary disconnection between medial limbic structures and cortical centres of control (Ferguson, Rayport, & Corrie, 1986), which could disrupt appraisal systems. Moreover, aggressive outbursts in patients suffering from temporal lobe epilepsy have been viewed as a trait-like behaviour, which usually appears in the nonictal phase (Ramani & Gumnit, 1981). This might reflect decreased perfusion and metabolism in the area of the epileptic foci (Valmier, Touchon, Daures, Zanca, & Baldy-Moulinier, 1987). Based on the asymmetry of the human brain, a comparison of patients with unilateral right or left anterior temporal spiking foci has indicated that the right focus is associated with elation and optimism, and the left focus facilitates humourlessness, sadness, obsessiveness, anger, and aggressiveness (Bear & Fedio, 1977; Mandell, 1978).

Episodic dyscontrol has been suggested to be a consequence of neurophysiological dysfunction of the limbic system (limbic ictus) (Monroe, 1970). Monroe (1982) argues that although an EEG abnormality is usually not found, support for his theory of limbic ictal phenomena rests with the response of individuals with the dyscontrol syndrome to anticonvulsant treatments. However, Leicester (1982) examined 500 cases referred to a neurologist and found that of the 17 patients referred for temper tantrums, none had organic factors (epilepsy or episodic dyscontrol syndrome). Rather, the violent episodes were the result of psychological factors. Furthermore, Hermann and Whitman (1984) reviewed 64 studies conducted in the previous 20 years that assessed the relation between temporal lobe epilepsy, aggression, and other forms of psychopathology. They focused on studies of the interictal period in terms of irritable, aggressive, or hostile behaviour using neurosurgical and non-neurosurgical patients, as well as surveys of prison populations. They concluded that controlled investigations showed no overall differences in the levels of violence between persons with and without epilepsy. Any association between aggression and epilepsy is likely to be due to nonspecific factors, including associated brain damage (Treiman, 1991).

Most of the classic neuroanatomical studies are conducted in lower animals, but as they also intervene in the normal functioning of the CNS, i.e. they are invasive, we must be cautious in interpreting and generalising the results to humans. New techniques using noninvasive images of the brain, such as positron emission tomography (PET), single photon emission tomography (SPET), or magnetic resonance imaging (MRI), are likely to provide us with much more information in the future. Thus, brain abnormalities not detected by conventional EEG technology can already be picked up by brain imaging. To date the number of violent individuals studied has been small but damage has been shown in the predicted areas, i.e. the amygdala, the temporal lobe and left hemisphere dysfunction seems to be more common.

NEUROCHEMICAL ASPECTS

Neurochemical systems in the body may sensitise the organism to various conditions for aggression and make such behaviour more likely. Attempts have been made to link modifications in various neurotransmitter systems with changes in various types of aggressive behaviour. Research has focused on five CNS neurotransmitter systems, i.e. 5-hydroxytryptamine (5-HT), gamma-amino butyric acid (GABA) and dopamine (DA) that are largely inhibitory, and noradrenaline (NA), largely excitatory, as well as acetylcholine (ACh).

Behavioural paradigms to measure aggression in animals grew out of experimental psychology and ethology (see Table 2.1). Exposure to aversive living conditions, such as deprivation of social contact or crowding, or the administration of aversive stimuli, such as shocks, are used to induce aggressive behaviour in habitually nonaggressive animals. Alternatively, predatory behaviour, such as mouse-killing by rats, is studied. Reis (1974) has suggested two major categories of animal aggression: predatory, such as the mouse-killing behaviour of rats, and affective, such as isolation-induced fighting in mice, intruder aggression, and pain-induced or shock-induced fighting in rodents. These standard laboratory paradigms have been used to study the activity of the different neurotransmitter systems on aggressive behaviour.

TABLE 2.1
Paradigms used to study behavioural aggression in animals

Aggressive behaviour appears to be enhanced by activation of the cholinergic system (Grossman, 1963). Aggressive behaviour can be produced by an injection of carbachol, a cholinomimetic drug, into the amygdala of cats. If the same drug is injected into the lateral hypothalamus of rats, predatory aggression will be increased. The behaviour is suppressed when methylatropine is applied centrally (Smith, King, & Hoebel, 1970). Also the injection of physostigmine, a cholinesterase inhibitor, into the amygdala of rats increased aggressive behaviour during a shock-induced fighting paradigm (Rodgers & Brown, 1976). In the same paradigm, aggression was abolished by cholinergic blockade (Powell, Milligan, & Walters, 1973). In humans, enhanced cholinergic receptor sensitivity has been linked to affective disorders (Janowsky & Risch, 1987), but may be associated with more general negative mood or dysphoria. In normal males, dysphoric responses to physostigmine have been shown to correlate with traits of irritability and emotional lability (Fritze et al., 1990), and in personality-disordered patients, dysphoric responses are most prominent in those with affective instability (Coccaro & Siever, 1995).

Central dopamine also appears to stimulate fighting in laboratory animals. Drugs that increase dopamine levels in the brain, like levodopa and apomorphine, a dopamine agonist, have been found to produce aggressive behaviour in rodents and defensive "boxing" behaviour in rats, respectively (Lammers & van Rossum, 1968). Dopamine antagonists are effective in reversing aggression (Yen, Stangler, & Millman, 1959) in shock-induced fighting in rats previously treated with an infusion of dopamine into the cerebral ventricles (Geyer & Segal, 1974).

The involvement of the GABA-ergic system in aggression has also been studied. Puglisi-Allegra and Mandel (1980) found that activation of central GABA activity decreases isolation-induced fighting in mice and shock-induced fighting in rats (Puglisi-Allegra, Simler, & Kempf, 1981). Thus, enhanced GABA activity inhibits aggression. Conversely, decreasing GABA activity may increase aggressive behaviour. Central GABA activity is augmented through activation of benzodiazepine receptors and so treatment with benzodiazepines would be expected to reduce aggressive behaviour. In line with this, benzodiazepine treatment has been associated with successfully controlling aggressive behaviour in brain-lesioned rats (Randall, Schaller, & Heise, 1960) and shock-induced fighting in rodents (Christmas & Maxwell, 1970). In contrast with these data, benzodiazepines have also been shown to increase aggressive behaviour under various conditions (Mos & Olivier, 1987). Benzodiazepine-induced aggression seems to differ according to species and compound tested (Krsiak & Sulcova, 1990). It is not clear if pro-aggressive effects are exclusively benzodiazepine receptor-mediated, but the benzodiazepine-GABA_A-C1 ionophore complex also seems to be an important site for the effects of alcohol on aggression (Miczek, Weerts, & DeBold, 1993). Alcohol-mediated aggression can be potentiated by benzodiazepine agonists and reversed by antagonists. Work in humans will be discussed later under benzodiazepines and in Chapters 6 and 7.

Noradrenaline has different effects on aggressive behaviour depending on the paradigm used. The enhanced central level of NA produced by tricyclic antidepressants exerts an inhibitory effect on predatory aggression (mouse-killing behaviour, cf. Katz, 1976). In contrast, most data support a facilitory role in affective or irritable aggression (Stolk, Conner, Levine, & Barchas, 1984). For example, artificially induced stress in rats (depriving them of REM sleep, or immobilisation for several hours a day for a month) produces irritability and shock-induced fighting after several days in association with altered central noradrenergic receptors (Eichelman & Hegstrand, 1982; Lamprecht, Eichelman, & Thoa, 1972). It is argued that this effect is caused by down-regulation of cortical beta-adrenergic receptors and increased activity of the noradrenaline synthesis enzyme tyrosine hydroxylase in the brain stem (Eichelman & Hegstrand, 1982) and hypothalamus (Lamprecht et al., 1972). On the other hand, intraventricular infusion of noradrenaline into normal rats reduces shock-induced fighting (Geyer & Segal, 1974). Further, depletion of central noradrenaline with the neurotoxin 6-hydroxydopa enhances shock-induced fighting in the rat (Thoa, Eichelman, & Richardson, 1972). Considering that this effect takes several days to develop fully, it has been attributed to a supersensitivity to endogenous noradrenaline. Human work has confirmed the association between enhanced noradrenergic activity and increased irritability. A positive correlation has been found between irritability, impulsivity, andCSF concentrations of 4,5-methyl-hydroxyphenylglycol (MHPG), a major metabolite of noradrenaline (Roy, De Jong, & Linnoila, 1989). Also, an augmented growth hormone response to clonidine has been shown to correlate with subscales of the Buss Durkee Hostility Inventory (Coccaro et al., 1991; Trestman et al., 1992). It has thus been suggested that the noradrenergic system plays a major role in responsiveness to the environment and reactivity to threatening stimuli.

Among all the neurotransmitter systems involved in aggressive behaviour, convergence of animal and human research is best exemplified by the serotonergic system. 5-HT receptors are now known to constitute a complex series of subtypes, e.g. 5-HT₁, 5-HT₂, and even sub-subtypes, e.g. 5-HT_1A (Harrington, Zhong, Garlow, & Ciaranello, 1992). At least 15 receptor types are now proposed based on evidence of cloning but the number increases rapidly. Attention has focused on 5-HT_1A, 5-HT₂, and 5-HT₃ receptors and various agonists and antagonists are being evaluated in a wide range of psychiatric disorders including aggression. However, the physiological specificity of these receptors is still being clarified. Lowering central serotonin levels enhances both affective and predatory aggression in laboratory animals (Eichelman, 1988). One important role of a normally functioning serotonin system is to facilitate appropriate delays in behavioural responses to various environmental stimuli (Sanger & Blackman, 1976), as well as the release of inhibition of "punished behaviours" (Tye, Everitt, & Iversen, 1977; Tye, Iversen, & Green, 1979). An extrapolation of similar behaviour in humans would be the relative incapacity of aggressive individuals to "delay gratification" (cf. Soubrie, 1986). Abnormalities of the serotonergic system in patients with different psychiatric diagnoses correlate with impulsive inward and outward directed aggressive behaviours (Asberg, Schalling, Traskman-Bendz, & Wagner, 1987) but not with premeditated acts of violence (Brown & Linnoila, 1990; Linnoila & Virkkunen, 1992). This suggests that violence itself is not necessarily associated with decreased central 5-HT activity. Coccaro, Kavoussi, and Lesser (1992) have suggested that central 5-HT dysfunction may represent a behavioural trait (irritability), which could be characterised by a "tendency to respond" aggressively to adverse stimuli, rather than the actual act of aggression, and others have suggested that it is impulsivity rather than aggression that is correlatd with decreased central 5-HT activity. Serotonin may in fact be involved in the appraisal process so that abnormalities of the system mean that individuals are both more likely to perceive threat and to attribute blame. Interrelationships have in fact been found between violence risk and suicide risk, anger, impulsivity, and anxiety (Apter et al., 1990).

In humans, evidence supporting an association between serotonergic system dysfunction and aggression comes from five main sources: studies in suicidal populations (e.g. Asberg, Thoren, & Traskman, 1976a); with patients with impulsive aggressive behaviours (e.g. Brown, Goodwin, Ballenger, Goyer & Major, 1979; Brown et al., 1982); 5-HT receptor-related studies (e.g. Coccaro & Astill, 1990); familial and genetic factors related to the 5-HT system (e.g. Linnoila, De Jong, & Vikkunen, 1989); and neuroendocrine challenges with drugs that interfere with 5-HT activity (e.g. Meltzer & Lowe, 1987).

Evidence for a low cerebrospinal fluid (CSF) 5-hydroxyindoleacetic acid (5-HIAA) concentration in patients with depression and with associated vegetative symptoms was first observed by van Praag and Korf (1971). In a subsequent study, Asberg, Traskman, and Thoren (1976b) demonstrated an association between low CSF 5-HIAA concentrations and violent suicide attempts in depressed patients. Subsequently many other CSF studies (e.g. Banki, Arato, Papp, & Kurcz, 1984; Traskman, Asberg, Bertilsson, & Sjostrand, 1981) have confirmed the relation between CSF 5-HIAA and attempts at suicide in patients with unipolar depression, personality disorder, or schizophrenia (Lester, 1995). Furthermore, low brain serotonin turnover, as indicated by low CSF 5-HIAA levels, correlates negatively with scores on the impulsiveness scale of the Karolinska Scale of Personality (Schalling & Asberg, 1984). Consequently, impulsive or violent suicidal behaviour may be considered as the product of inhibiting (mediated by 5-HT) and activating (mediated by other neurotransmitter systems, such as noradrenaline and dopamine) a neuronal system (Depue & Spoont, 1986).

Abnormalities of the serotonin system may have a wider role in relation to aggression and violence. Brown et al. (1979, 1982) found strong negative correlations between 5-HIAA levels and a history of aggressive behaviour, suicide attempts, and scores on the Buss-Durkee Hostility Inventory, and the psychopathic deviate scale from the Minnesota Multiphasic Personality Inventory, in patients with personality disorder. Further analyses of these data suggest that individuals with low CSF 5-HIAA had shown disturbed behaviour during childhood (Brown & Goodwin, 1984). Linnoila et al. (1983) also found a linear association between low CSF 5-HIAA and murder and attempted murder in patients whose crimes were assessed as impulsive, by comparison with nonimpulsive violent offenders who had premeditated their crimes. Moreover, offenders who had committed more than one violent crime had lower levels than those who had committed only one such crime, and impulsive offenders who had attempted suicide had lower levels than violent offenders who had never attempted suicide. In addition, CSF 5-HIAA levels are lower in arsonists than in nonimpulsive violent offenders or normal controls (Virkkunen, Nuutila, Goodwin, & Linnoila, 1987).

Post-mortem brain studies of suicide victims have also revealed pre-synaptic (Asberg et al., 1987) and post-synaptic indices of serotonin turnover to be changed (Arora & Meltzer, 1989). The index of pre-synaptic 5-HT function was the binding of tritiated imipramine to the brain tissue: A reduction of maximal binding of imipramine was found in various brain areas (Crow et al., 1984; Gross-Isseroff, Israeli, & Biegon, 1989; Paul, Rehavi, Skolnik, & Goodwin, 1984; Stanley, Virgilio, & Gershon, 1982), suggesting a possible loss of 5-HT pre-synaptic receptors. However, Owen et al. (1986) and Meyerson et al. (1982) reported respectively no alteration and an increase in the number of pre-synaptic serotonin receptors. In contrast, post-synaptic upregulation of the 5-HT₂ receptors in the frontal cortex has been reported in impulsive aggressive and suicidal individuals (Arora & Meltzer, 1989; Mann, Stanley, McBride, & McEwen, 1986; Stanley & Mann, 1983). In addition, a decrease of function mediated by 5-HT₁ receptors seems to occur (Gudelsky, Koenig, & Meltzer, 1986). This reciprocal effect has received support from behavioural responses to antidepressant treatment (Pericic & Manev, 1988), and animal studies that have indicated that 5-HT₁ receptors, in contrast to 5-HT₂ have an important role in the mediation of behavioural inhibition (McMillen, Scott, William, & Sanghera, 1987). On the other hand, the claim that increased 5-HT₂ activity would be a compensatory mechanism to down-regulation in the pre-synaptic terminals has not received empirical support. Increasing either the number of 5-HT₂ receptors or their responses to chemical stimulation is not followed by pre-synaptic 5-HT depletion (Blackshear, Steranka, & Sanders-Bush, 1981; Conn & Sanders-Bush, 1986; Quick & Azmitia, 1983).

Another method of studying 5-HT in the brain has been by examining a hormone response (prolactin) to a challenge with a 5-HT probe, e.g. with fenfluramine hydrochloride. Fenfluramine increases 5-HT system activity indirectly leading to the release of prolactin. Patients with various disorders such as borderline personality disorder, a history of suicide attempts, or episodic alcohol abuse exhibit reduced prolactin responses to this challenge compared with normal controls (Coccaro et al., 1989; Siever et al., 1987). Recently a similar association between reduced Cortisol responses to d-fenfluramine and the motor aggression factor on the BDHI has been found in healthy males (Cleare & Bond, 1997). Another study measured the prolactin response to a challenge with a post-synaptic 5-HT agonist, m-chlorophenylpiperazine (m-CPP) in men with antisocial personality disorder with substance abuse compared with healthy controls (Moss, Yao, & Panzak, 1990). They found that assaultive aggression, resentment, and irritability were associated with a diminished prolactin response to m-CPP. This work therefore confirms the correlation between reduced indices of 5-HT function and irritable, impulsive aggression.

The serotonin system is also affected by drugs leading to disinhibition, impulsivity, and aggression. Alcohol and benzodiazepines may produce behavioural dyscontrol in normal subjects. Although both these substances have effects on a range of neurotransmitter systems, a change in serotonin function could contribute to their disinhibitory effects. Moreover, a relationship between impulsive violent behaviour, low serotonin turnover, and type II alcoholism (i.e. male-linked alcoholism, inherited from fathers by sons, cf. Cloninger, Bohman, & Sigvardsson, 1981) has been found (Linnoila & Virkkunen, 1992; Roy, Virkkunen, & Linnoila, 1987).

In summary, there is animal evidence that many neurotransmitter systems may be involved in aggressive behaviour. In humans the current evidence supports a strong link between affective aggression and lowered levels of central serotonin, although other neurotransmitter systems are likely to be involved. Recent evidence suggests an associated increase in noradrenergic activity. In addition, studies have shown that these biological correlates of aggressive behaviour are under partial genetic control. Thus, heredity accounts for a significant proportion of the variation in biogenic amines in rhesus monkeys (Clarke et al., 1995). There were significant differences between sire-family groups for CSF levels of noradrenaline, the serotonin metabolite, 5-HIAA, and the dopamine metabolites, HVA and DOPAC. Indirect measures have been used to confirm these results in humans. Reduced prolactin responses to fenfluramine challenge have been found in the relatives of patients with personality disorder and a history of impulsive aggression (Coccaro, Silverman, Klar, Horvath, & Siever, 1994).

HORMONES

The consistency of sex differences in aggressive behaviour across species has led to the study of hormonal influences. The relationship between aggressive and criminal behaviour in men and the sex hormone testosterone has been explored. This hormone appears to be related to some indicators of aggression, but not others. In one study, testosterone levels in prison inmates covaried closely with the violence of the crime for which the inmate had been sentenced, but was not correlated with rated violence of the individuals (Dabbs, Frady, Carr, & Besch, 1987). Testosterone level may be related primarily to a disposition to aggress, and other stimuli must be present before the disposition is manifested in aggressive behaviour. Although testosterone level in boys has been shown to correlate positively to measures of both physical and verbal aggression (Olweus, Mattson, Schalling, & Low, 1980, 1988), the correlation was higher when provocation or threat were examined. Also, a positive correlation was found between testosterone level and lack of tolerance for frustration. At present there is no compelling evidence that testosterone is a direct cause of aggression, but it should be considered as part of a biosocial model (Mazur, 1995).

In women the relation between sex hormones and aggression is unclear. Increased irritability and hostility leading to aggression prior to menstruation have been reported in some studies. Dalton (1977) found a tendency for aggressive behaviour among women inmates during menstruation. In these cases, aggressiveness would be related to a drop in progesterone level during menstruation, along with a rise in the ratio between oestrogen and progesterone. Moreover, administration of progesterone decreased the likelihood of aggressive behaviours during menstruation by alleviating the feelings of irritability and hostility (Dalton, 1977). On the other hand, this associated effect was not confirmed in several other studies, and Bancroft and Backstrom (1985) argue against a direct cause-effect relationship between pre-menstrual mood change and levels of progesterone. Recently, modulation of both the serotonergic and dopamine systems has been reported (Wieck, 1996).

Summing up the evidence, the biological make-up of both humans and animals undoubtedly plays a crucial role as a background condition in the appearance of aggressive behaviours. The data reviewed here suggest that these body chemicals are not in themselves causes of aggression. Aggressive reactions are most likely to result from interactions between external stimuli and personal sensitivities.

DRUGS USED IN THE MANAGEMENT OF AGGRESSION

The laboratory work reviewed earlier is complemented by the empirical assessment of various drugs with a wide range of actions that have been used to treat aggressive behaviour.

Antipsychotics

The antipsychotics are the most used group of compounds, and the treatment of choice for controlling aggression in schizophrenia (Yudofsky, Silver, & Schneider, 1987). Many compounds have been used and their efficacy in controlling rather than treating aggressive behaviour was established in many early studies in the 1960s and 1970s (Itil & Wadud, 1975). A review some years ago suggested that chlorpromazine and trifluopromazine were the most effective of the phenothiazines (Levanthal & Brodie, 1981). Of the other groups, haloperidol is often used unless a depot formulation is required, when flupenthixol decanoate has been shown to be effective. Newer compounds have also been tried with some degree of success (Tuason, 1986). The rationale behind use of these drugs to control aggression is often pragmatic, weighing risks against side-effects, rather than comparing efficacy. It seems that thioridazine, chlorpromazine, and haloperidol are all commonly used in the USA and the UK (Wressell, Tyrer, & Berney, 1990; Yudofsky, Silver, & Hales, 1990) but it is difficult to establish differences between compounds as they have rarely been compared in controlled trials with properly matched samples. Antipsychotics are also used to control acute episodes of disturbed behaviour whatever the pathology. This is termed Rapid Tranquillisation and is primarily a sedative effect (Pilowsky, Ring, Shine, & Lader, 1992).

Although most studies have focused on the use of antipsychotics to control aggressive behaviour in schizophrenia, their efficacy has also been shown in other disorders. It is interesting that antipsychotics including thiothixene and haloperidol have proved helpful in treating impulsiveness and aggressive outbursts in both borderline and impulsive and antisocial personality disorder in low doses, where no obvious tranquillisation is present (Goldberg et al., 1986; Soloff et al., 1986, 1989, 1993; Tyrer & Seivewright, 1988). This group of drugs is also widely used in patients with learning disabilities, and apart from early studies, two more recent studies have shown improvements. One (Gualtieri & Schroeder, 1989) used low-dose (2–8mg) fluphenazine and found a decrease in both self-injury and aggression. The other found pipothiazine palmitate (a depot phenothiazine) to reduce aggression (Lynch, Eliatamby, & Anderson, 1985). Some antipsychotics have also been evaluated in patients with organic brain disease but the improvement in symptoms such as hostility and uncooperativeness is usually only modest (Coccaro et al., 1990). Antipsychotics must be used cautiously in epilepsy or related conditions as these drugs lower the seizure threshold.

Benzodiazepines

Benzodiazepines have been much used in the past in the control of aggressive behaviour, with many studies demonstrating their efficacy (Bond & Lader, 1979). The use of benzodiazepines has also been advocated in the control of acute behavioural disturbance (Mendoza, Djenderedjian, Adams, & Ananth, 1987). This has met with some well-founded criticism (Graham, 1988) but benzodiazepines are often used in conjunction with antipsychotics in rapid tranquillisation (Pilowsky et al., 1992; Salzman, 1988). The anti-aggressive efficacy of benzodiazepines apart from sedation has never been established and recently evidence has been accruing of their pro-aggressive properties. Reports of rage reactions stem from early marketing days, but it was always considered to be a paradoxical response and associated with high doses in predisposed individuals. However, it has been suggested that it may be much more common than hitherto realised and may represent an under-reported general rise in irritability sometimes leading to uncontrolled aggressive outbursts (DiMascio, Shader, & Harmatz, 1969). One review (Dietch & Jennings, 1988) estimated that less than 1% of patients treated with benzodiazepines experienced aggressive dyscontrol.

Animal studies have shown both anti- and pro-aggressive actions for benzodiazepines. These drugs may merely strengthen the current or prevailing behavioural tendency (Dantzer, 1977), but this does not explain why people supposedly behave out of character and may do the same when rechallenged despite the apparent prevailing mood at the time (Regestein & Reich, 1985). Mos and Olivier (1987) conclude from a review of the pro-aggressive effects of benzodiazepines, mainly in animals, that benzodiazepines enhance aggression when basal levels are low to moderate. This may be true of studies in anxious or neurotic patients and in laboratory models used in normal subjects (Gardos, Di Mascio, Salzman, & Shader, 1968; Salzman et al., 1974; Wilkinson, 1985), but does not apply when they are used to treat aggressive behaviour per se. Two or three reports have detailed the incidence of hostility following a newer high-potency benzodiazepine, alprazolam, in different patient groups. In one study, 8 out of 80 patients treated with this benzodiazepine displayed extreme anger or hostile behaviour (Rosenbaum, Woods, Groves, & Klerman, 1984). The initial hostility occurred within the first week in all patients and after a single dose in two. Discontinuing the drug led to a resolution of symptoms within a few hours. Three patients were rechallenged and two of the three became hostile. The authors conclude that alprazolam-induced hostility is an early and idiosyncratic effect and may be more likely in patients with well-suppressed chronic anger and resentment.

It should be noted here that triazolam, a high-potency benzodiazepine hypnotic, was withdrawn from the UK market because of problems with adverse reactions including disinhibition and aggression (Drug and Therapeutics Bulletin, 1991). Considering this evidence, it would seem advisable to avoid benzodiazepines in the treatment of aggression, but this may be premature as there appear to be differences among compounds: Thus, oxazepam has not been shown to increase aggressive behaviour in normal subjects (Gardos et al., 1968) and has been recommended as a specific anti-hostility tranquilliser (Salzman, Kochansky, Shader, Harmatz, & Ogletree, 1975).

Lithium

In psychiatric patients, most evidence for the efficacy of lithium in the control of aggressive behaviour comes from case studies and uncontrolled reports. A consistent improvement over previous treatments, e.g. the use of phenothiazines or benzodiazepines for years, clearly reveals the failure of other drugs (Shader, Jackson, & Dodes, 1974). There is convincing evidence for the efficacy of lithium in both explosive and aggressive personality disorder and in violent prisoners with mixed diagnoses. A series of studies in the USA showed a reduction in violent incidents and serious assaults without a dulling of affect (Sheard, 1971, 1975; Sheard, Marini, Bridges, & Wagner, 1976). Despite this seemingly favourable evidence, lithium does not appear to have been studied much in patients with antisocial or borderline personality disorder. This could be because of potential side-effects, but it has been suggested that when administered every second day, lithium maintains efficacy with much reduced side-effects (Mellerup & Plenge, 1990). There is also evidence from retrospective studies that lithium may be useful in patients with learning disabilities. The use of lithium in patients with temporal lobe epilepsy or inter-ictal aggressive behaviour is controversial as one study has shown a deterioration in these patients (Jus, Villeneuve, & Gautier, 1973).

Beta-adrenergic antagonists

Numerous case reports and open studies have established the efficacy of high dose beta-adrenergic antagonists in controlling aggression in patients in whom other medication (antipsychotics, benzodiazepines, antidepressants, anticonvulsants) has failed (Elliott, 1977; Greendyke, Schuster, & Wooton, 1984; Yudofsky, Williams, & Gorman,1981). This effect has been supported in a few double-blind trials (Greendyke & Kanter, 1986; Greendyke, Kanter, Schuster, Verstreate, & Wooton, 1986) and is particularly true for patients with organic brain disease or injury. It has been speculated that the beta blockers may have a bimodal action, first in the periphery and second in the CNS, and this is supported by work showing nadolol, a peripherally acting beta blocker, to be as effective as propranolol when added to mesoridazine (Polakoff, Sorgi, & Ratey, 1986). A recent review of studies using beta blockers in people with learning disabilities cautioned that the impressive response rate was based solely on case reports and open clinical trials (Ruedrich, Grush, & Wilson, 1990).

In schizophrenia, beta blockers have been added to maintenance antipsychotic treatment and where this has been done systematically, there has been some evidence of improvement (Sorgi, Ratey, & Polakoff, 1986). Although a more recent study with adjunctive nadolol (80–120mg/day) showed only a trend towards decreased aggression (Alpert et al., 1990), the authors rightly point out that any consistent trend when a drug is added double-blind to a behavioural programme is important in a severely disturbed group of patients. However, beta blockers can elevate the bodily concentrations of some antipsychotics by competing for liver enzymes, and so caution with this combination is advocated (Hanssen et al., 1980; Silver, Yudofsky, Kogan, & Katz, 1986). There is some preliminary work to show that propranolol can be useful in intermittent explosive disorder (426mg/day), although patients preferred carbamazepine and tolerated it better (Mattes, Rosenberg, & Maya, 1984).

Anticonvulsants

Electrical disturbances in the brain have been implicated as a cause of episodic human violent behaviour but anticonvulsants, e.g. phenytoin, have had mixed results in the control of such aggressive behaviour over and above the control of epilepsy. Some open trials have been favourable but controlled trials have not always replicated these results, leading to the view that most anticonvulsants are poorly effective in controlling aggressive behaviour in adults (Eichelman, 1987) or children (Conners, Kramer, Rothschild, Schwartz, & Stone, 1971). However, carbamazepine, which has a mood-stabilising effect in addition to its anticonvulsant action, has shown efficacy in managing aggressive behaviour in various patient groups (Cowdry & Gardner, 1988; Mattes, 1990). Carbamazepine has also been added to antipsychotic treatment in schizophrenics. In a double-blind 15-week crossover study of carbamazepine (200mg three times a day) and placebo, all 11 patients showed some improvement and 5 were markedly clinically better (Neppe, 1983). Another retrospective study showed a reduction in recorded aggressive episodes after carbamazepine treatment (Luchins, 1984), and there are many other open and single-case studies confirming improvement in symptoms such as aggression and hostility. There is no evidence, however, that carbamazepine is effective in schizophrenics not exhibiting symptoms of dyscontrol. In fact, it may produce deterioration because of a lowering of plasma levels of antipsychotics (Kidron, Averbuch, Klein, & Belmaker, 1985). It has therefore been suggested that the rationale for adjunctive carbamazepine treatment in nonresponsive psychosis should be the presence of target symptomatology such as hostility or affective lability (Neppe, 1988, 1990).

Stimulants and antidepressants

Methylphenidate, dextroamphetamine, and pemoline have all shown some efficacy in the treatment of attention-deficit disorder in children and adolescents (Miczek, 1987) and there is some evidence that this effect is also shown in adults who exhibit persistent symptoms of minimal brain dysfunction (Wender, Reimherr, & Wood, 1981; Wender, Reimherr, Wood, & Ward, 1985). Hyperactivity and poor concentration usually improve with age, but where aggressive behaviour is part of the syndrome, it may worsen. Stimulants have limited therapeutic indications as abuse of stimulants and high-dose intoxication can lead to extreme aggression (Ellinwood, 1971), although this is often secondary to a psychotic paranoid state. Although tricyclic antidepressants generally are not effective in the control of aggression, there is some limited evidence, from open trials, of the efficacy of monoamine oxidase inhibitors in attention-deficit disorder and borderline personality disorder, and as newer selective and reversible inhibitors such as moclobemide are developed, they deserve evaluation in this context.

Effects of serotonergic compounds

Tryptophan, the precursor of serotonin, has some beneficial effects when added to maintenance antipsychotic treatment in schizophrenics who have been convicted of violent crime or exhibited episodes of threatening aggression (Morand, Young, & Ervin, 1983). In another study, it showed an indirect effect by decreasing the need for p.r.n. injectable medication in aggressive psychiatric inpatients (Volavka et al., 1990). Both lithium and carbamazepine enhance 5-HT activity. Propranolol has both 5-HT₁ and 5-HT₂ antagonist properties, and although the efficacy of the antipsychotics is generally attributed to their anti-dopaminergic and sedative effects, most also block 5HT₂ receptors (Glennon, 1990). Indeed, both clozapine and risperidone are very active in this respect.

Serotonin agonists

The prototype, buspirone, has been marketed for the treatment of anxiety in the post-benzodiazepine era, but there is evidence that the later compounds, gepirone, ipsapirone, and flesinoxan, are more specific in their action on serotonin without the additional dopaminergic effects of buspirone. These drugs have also been shown to have some antidepressant effect and may in fact have mood-stabilising qualities in neurotic disorders. Animal work has pointed to the importance of the 5-HT_1A and 5-HT_1B receptor subtypes in the modulation of aggressive behaviour, and a class of drugs that act specifically on the 5-HT₁ receptor (Sijbesma, Schipper, & De Kloet, 1990), aptly named the Serenics—Eltoprazine and Fluprazine—have been developed for their putative anti-aggressive effects (Rasmussen, Olivier, Raghoebar, & Mos, 1990). These are the first drugs to be specifically developed to treat aggressive behaviour but have not fulfilled their early promise. The 5-HT_1B receptor has not been identified in the human brain (Hoyer, Pazos, Probst, & Palacios, 1986) and other animal work has suggested that two serotonin classes of receptor may be differentially involved in the modulation of aggression: stimulation of an inhibitory 5-HTi receptor and blockade of the 5-HT₂ receptor (Lindgren & Kantak, 1987). As there is interaction between these two receptors, antagonism of 5-HT₂ may enhance the function of 5-HT₁ (Montgomery & Fineberg, 1989), still implying a major role for the 5-HTi receptor in aggression.

The only one of these specific compounds yet to be tested formally in aggression is buspirone. There was a suggestion of a reduction in irritability from the original anxiety studies (Glitz & Pohl, 1991) and a series of case reports on people with developmental disabilities has shown a decrease in aggression and self-injury (Ratey, Sovner, Mikkelsen, & Chmielinski, 1989). This has been confirmed in one double-blind study of aggression and anxiety in six patients with mild to moderate learning disabilities who showed decreased aggressive and self-injurious behaviour after buspirone (up to 45mg) (Ratey, Sovner, Parks, & Rogentine, 1991). Other case studies have shown buspirone to decrease aggressive, hostile, and threatening behaviour in patients with varying diagnoses (Balon, 1990; Colenda, 1988; Levine, 1988; Tiller, 1988).

Serotonin re-uptake inhibitors

The serotonin re-uptake inhibitors, which increase central serotonin indirectly (by blocking re-uptake and altering the availability of 5-HT to one or other receptor) have been marketed as antidepressants, but have also had success in treating obsessive-compulsive disorder, which may itself be linked to impaired impulse control. It has been suggested that they work by inducing a state of indifference to intrusive thoughts (Healy, 1990) and so it is also possible that they will increase the perceived threshold to potential sources of provocation and therefore reduce the likelihood of an impulsive aggressive response. Taken together with the work associating lowered CNS levels of serotonin and impulsive aggression, serotonergic compounds may be exerting a normalising effect on a pre-existing deficiency. Although it has been suggested that the serotonin re-uptake inhibitors are effective on symptoms of impulsivity and irritability associated with panic and anxiety, and evidence is emerging that they decrease the incidence of anger attacks in depressed patients (Rosenbaum et al., 1993), systematic studies of their use to treat aggressive behaviour are only now being conducted. There is some preliminary evidence that citalopram may reduce the frequency of aggressive incidents in chronically violent schizophrenic in-patients (Vartiainen et al., 1995) and that fluoxetine may reduce impulsive aggressive behaviour in patients with borderline personality disorder (Cornelius et al., 1991; Markovitz et al., 1991; Norden, 1989) but more well constructed, double-blind trials are necessary.

SUMMARY

The brain mechanisms currently implicated in aggressive behaviour are the amygdala, the hypothalamus, and the temporal lobe. Most of this work is based on invasive techniques, which intervene in normal CNS functioning, and the growth in noninvasive techniques such as brain imaging is likely to improve our knowledge in this field. Many neurotransmitter systems are involved in aggressive behaviour but work in humans implies a strong link between affective or hostile aggression and lowered central serotonin. An association has also been found with enhanced noradrenergic activity and between testosterone and a lack of tolerance for frustration. Although the biological make-up of humans is important in this context, aggressive behaviour is likely to result from interactions between these and other factors, e.g. external and interpersonal stimuli. Many drugs are used clinically in the management of aggression and these are outlined. Different drugs have proved helpful in different types of aggression. There is some evidence that serotonergic compounds may reduce impulsive, irritable aggression but it is suggested that more well-controlled, double-blind trials are needed to evaluate their efficacy.