Chapter 16: Pharmacotherapy of Psychosis and Mania

The Dopamine Hypothesis. The development of our understanding of the neurobiology and pharmacotheorapy of psychoses profited from the synthesis of chlorpromazine in 1950 and of haloperidol in 1958. The DA hypothesis of psychosis derived from the fortuitous discovery of chlorpromazine's therapeutic efficacy in schizophrenia, and the subsequent elucidation by Carlsson that postsynaptic DA D₂ receptor antagonism was the common mechanism that explained antipsychotic properties. Reserpine, derived from Rauwolfia, exhibited antipsychotic properties by decreasing dopaminergic neurotransmission; however, unlike D₂ receptor antagonists, reserpine exerted its effects through depletion of monoamines from presynaptic nerve terminals. The dopamine theory of psychosis was reinforced by the high risk for drug-induced psychosis among substances that directly increased synaptic dopamine availability, including cocaine, amphetamines, and the Parkinson's disease treatment L-dopa (Carlsson, 1978).

The psychoses related to delirium and dementia, particularly dementia of the Alzheimer type, may share a common etiology: the deficiency in cholinergic neurotransmission, either due to anticholinergic properties of medications (Chew et al., 2008), age- or disease-related neuronal loss, or both (Barten and Albright, 2008; Hshieh et al., 2008). Among hospitalized elderly patients, increased plasma concentrations of anticholinergic medications are directly associated with increased delirium risk (Flacker and Lipsitz, 1999); however, unlike in Alzheimer's dementia, where psychotic symptoms are directly related to cholinergic neuronal loss and may respond to acetylcholinesterase therapy, delirium may have numerous precipitants besides medication-associated anticholinergic properties, all of which require specific treatment (e.g., infection, electrolyte imbalance, metabolic derangement) in addition to removal of offending anticholinergic medications.

Schizophrenia is a neurodevelopmental disorder with complex genetics and incompletely understood pathophysiology. Certain environmental exposures confer an increased risk of developing schizophrenia, including fetal second-trimester viral and nutritional insults, birth complications, and substance abuse in the late teen or early adult years (Fanous and Kendler, 2008). Rather than being the result of a single gene defect, mutations or polymorphisms of many genes appear to contribute to the risk for schizophrenia. Implicated are genes that regulate neuronal migration and synaptogenesis (neuregulin 1), synaptic DA availability (Val{108/158}Met polymorphism of catechol-O-methyltransferase, which increases DA catabolism), glutamate and DA neurotransmission (dystrobrevin binding protein 1 or dysbindin, particularly with schizophrenia patients with prominent negative symptoms), nicotinic neurotransmission (α7-receptor polymorphisms), and cognition (disrupted-in-schizophrenia-1) (Porteous, 2008). Schizophrenia patients also have increased rates of genome-wide DNA microduplications termed copy number variants (Need et al., 2009) and epigenetic changes, including disruptions in DNA methylation patterns in various brain regions (Porteous, 2008). This genetic variability is consistent with the clinical disease heterogeneity, and suggests that any one specific mechanism is unlikely to account for large amounts of disease risk. In addition to increased mesolimbic DA activity related to positive symptoms, schizophrenia patients have decreased DA D₁ activity in the dorsolateral and ventromedial prefrontal cortex (PFC) that is associated with cognitive deficits and negative symptoms (Buchanan et al., 2007).

Cognitive dysfunction is the greatest predictor of poor functional outcome in schizophrenia and shows limited response to antipsychotic treatment. Experimental studies have provided new insights into the mechanisms of cognitive dysfunction. Glutamate NMDA receptor stimulation is involved in tonic inhibition of mesolimbic DA release, but facilitates mesocortical DA release (Sodhi et al., 2008). Ketamine infusion studies in animals and healthy volunteers demonstrate that decreased NMDA function results in a picture that more accurately encompasses all aspects of schizophrenia, including positive, negative, and cognitive symptoms, and social withdrawal. Several of the newer antipsychotic drugs remediate not only the positive psychotic symptoms but also the cognitive disruption induced by ketamine and other potent NMDA antagonists such as MK-801 (dizocilpine). These results have prompted the clinical investigation of agents without affinity for DA receptors, but with potent agonist properties at metabotropic glutamate receptor subtypes (Patil et al., 2007).

Delusional disorder, schizophrenia, and schizoaffective disorder are chronic diseases that require long-term antipsychotic treatment, although there are few reliable studies of treatment outcomes for delusional disorder patients. Individuals with monosymptomatic delusions (e.g., paranoia, marital infidelity) do not have neurocognitive dysfunction and may continue to perform work and social functions unaffected by their illness, aside from behavioral consequences related to the specific delusional belief (American Psychiatric Association, 2000). These patients often have limited or no psychiatric contact outside of mandated legal interventions, thus limiting the opportunity for clinical trials. For schizophrenia and schizoaffective disorder, the goal of antipsychotic treatment is to maximize functional recovery by decreasing the severity of positive symptoms and their behavioral influence, improving negative symptoms, decreasing social withdrawal, and remediating cognitive dysfunction. That only 15% of chronic schizophrenia patients are employed in any capacity and 11% are married has been attributed to the relatively limited effect of treatment on core negative and cognitive symptoms of the illness (Lieberman et al., 2005). Nonetheless, continuous antipsychotic treatment reduces 1-year relapse rates from 80% among unmedicated patients, to ~15%. Poor adherence to antipsychotic treatment increases relapse risk, and is often related to adverse drug events, cognitive dysfunction, substance use, and limited illness insight.

The best-tolerated doses in dementia patients are one-fourth of adult schizophrenia doses (e.g., risperidone 0.5-1.5 mg/day), although extrapyramidal neurological symptoms (EPS), orthostasis, and sedation are particularly problematic in this patient population (Chapter 22). Significant antipsychotic benefits are usually seen in acute psychosis within 60-120 minutes after drug administration. Delirious or demented patients may be reluctant or unable to swallow tablets, so oral dissolving tablet (ODT) preparations for risperidone, aripiprazole, and olanzapine, or liquid concentrate forms of risperidone or aripiprazole, are options. The dissolving tablets adhere to any moist tongue or oral surface, cannot be spit out, and are then swallowed along with oral secretions. Intramuscular (IM) administration of ziprasidone, aripiprazole, or olanzapine represents an option for treating agitated and minimally cooperative patients, and presents less risk for drug-induced parkinsonism than haloperidol. QT_c prolongation associated with intramuscular droperidol and intravenous administration of haloperidol have curtailed use of those particular formulations. Treatment continues until agitated or hallucinatory behaviors are controlled and the underlying etiologies are addressed (Lacasse et al., 2006). (See "Use in Pediatric Populations" and "Use in Geriatric Populations" later in the chapter.)

With few exceptions (e.g., loxapine, where the active metabolite amoxapine is a heterocyclic antidepressant), most antipsychotic drugs show limited antidepressant benefit as monotherapy agents. However, clinical data demonstrate efficacy of atypical antipsychotic agents as adjunct therapy in treatment-resistant depression. The clinical efficacy may be related to the fact that almost all atypical antipsychotic medications are potent 5-HT_2A antagonists, with 5-HT_2A occupancies exceeding 80% at modest dosages (Figure 16–2) (Mamo et al., 2007). Both 5-HT_2A and 5-HT_2C receptors are G-protein coupled, with stimulation resulting in increased production of intracellular inositol trisphosphate (IP₃) (Marek et al., 2003). In vitro assays demonstrate that atypical agents are inverse agonists at both receptor subtypes, capable of lowering basal IP levels (Rauser et al., 2001; Weiner et al., 2001). For the purposes of functional description, the term antagonism will be used throughout this chapter with respect to actions at these two serotonin receptor subtypes, even though the underlying mechanism is inverse agonism (Chapter 3). 5-HT_2A and 5-HT_2C antagonism not only facilitates DA release but also increases noradrenergic outflow from the locus coeruleus. Decreased noradrenergic tone in locus coeruleus projections to PFC and limbic areas is considered an important aspect of depression pathophysiology. Pure 5-HT_2A antagonists are not effective antidepressants, but administration of 5-HT_2A and 5-HT_2C antagonists in the form of low doses of atypical antipsychotic agents, along with selective serotonin reuptake inhibitors (SSRIs), increases responses rates in SSRI nonresponders (Berman et al., 2007; Mahmoud et al., 2007; Rothschild et al., 2004). A combination preparation of low-dose olanzapine and fluoxetine is approved for bipolar depression, and low-dose risperidone (i.e., 1 mg) increases clinical response rates when added to existing SSRI treatment in SSRI nonresponders. Aripiprazole is FDA-approved for adjunctive use in antidepressant nonresponders, again at low doses (2-15 mg). Unlike other atypical antipsychotic agents, aripiprazole's highest affinities at low doses (< 15 mg) are for D₂ and D₃ receptors, at which it is a partial agonist, with only 50% 5-HT_2A receptor occupancy at 15 mg (Mamo et al., 2007) (Figure 16–2C). Aripiprazole and most other antipsychotic drugs are ineffective as monotherapy for bipolar depression, with quetiapine being the sole exception. Two large clinical trials demonstrated quetiapine's efficacy for bipolar depression at both 300 and 600 mg/day doses (Calabrese et al., 2005a; Thase et al., 2006). Quetiapine also has antidepressant efficacy in unipolar major depression trials, suggesting that it could be used as monotherapy for depression. The explanation is likely that its primary metabolite, norquetiapine, is a potent norepinephrine reuptake inhibitor (Jensen et al., 2008).

Schizophrenia patients have a 2-fold higher prevalence of metabolic syndrome and type 2 diabetes mellitus (DM) and 2-fold greater cardiovascular (CV) related mortality rates than the general population (Meyer and Nasrallah 2009). For this reason, consensus guidelines recommend baseline determination of serum glucose, lipids, weight, blood pressure, and when possible, waist circumference and personal and family histories of metabolic and CV disease (American Diabetes Association et al., 2004). This cardiometabolic risk assessment should be performed for all schizophrenia patients despite the low metabolic risk for certain agents. With the low EPS risk among atypical antipsychotic agents, prophylactic use of anti-parkinsonian medications (e.g., benztropine, trihexyphenidyl) is not necessary; however, acute dystonic reactions can occur (see Table 16–4), typically among patients who have never before taken antipsychotic drugs, adolescents, and young adults. Lower dosages are recommended in these individuals. Drug-induced parkinsonism can also occur, especially among elderly patients exposed to antipsychotic agents that have high D₂ affinity (e.g., typical antipsychotic drugs, risperidone, paliperidone); recommended doses are ~50% of those used in younger schizophrenia patients. (See also "Use in Pediatric Populations" and "Use in Geriatric Populations" later in the chapter.)

There are many reasons for psychotic relapse or inadequate response to antipsychotic treatment in schizophrenia patients. Examples are ongoing substance use, psychosocial stressors, inherent refractory illness, and poor medication adherence. Outside of controlled settings, the problem of medication nonadherence remains a significant barrier to successful treatment, yet one that is not easily assessed. Serum drug levels can be obtained for most antipsychotic medications, but there is limited dose-response data for atypical antipsychotic agents (except for clozapine), leaving clinicians little guidance about the interpretation of antipsychotic drug levels. Even low or undetectable levels may not reflect nonadherence, but rather can be the result of genetic variation or induction of hepatic CYPs that decrease drug availability, such as the high prevalence of CYP2D6 ultrarapid metabolizers among individuals from North Africa and the Middle East. Nevertheless, the common problem of medication nonadherence among schizophrenia patients has led to the development of long-acting injectable (LAI) antipsychotic medications, often referred to as depot antipsychotics. They are more widely used in the E.U. Only < 5% of U.S. patients receive depot antipsychotic treatment. There are currently four available LAI forms in the U.S.: decanoate esters of fluphenazine and haloperidol, risperidone-impregnated microspheres, and paliperidone palmitate. Patients receiving LAI antipsychotic medications show consistently lower relapse rates compared to patients receiving comparable oral forms and may suffer fewer adverse effects. LAI risperidone and paliperidone palmitate are currently the only atypical antipsychotic agents in depot form. However, clinical trials of LAI aripiprazole and olanzapine are in progress and these agents should become available in the near future. LAI risperidone is administered as biweekly intramuscular (IM) injections of 25-50 mg. Based on extensive kinetic modeling of clinical data, paliperidone palmitate therapy in acute schizophrenia is initiated with IM deltoid loading doses of 234 mg at day 1 and 156 mg at day 8 to provide serum paliperidone levels equivalent to 6 mg oral paliperidone during the first week, obviating the need for oral antipsychotic supplementation. Serum paliperidone levels from these two injections peak on day 15 at a level comparable to 12 mg oral paliperidone. Maintenance IM doses are given in deltoid or gluteus muscle every 4 weeks after day 8. Unlike acute IM preparations or paliperidone palmitate, all other LAI antipsychotic medications require several weeks to attain therapeutic levels and months to reach steady state, necessitating the use of oral medications for the initial 4 weeks of treatment.

In addition to agranulocytosis risk that mandates routine ongoing hematological monitoring, clozapine has numerous other adverse effects. Examples are high metabolic risk, dose-dependent lowering of the seizure threshold, orthostasis, sedation, anticholinergic effects (especially constipation), and sialorrhea related to muscarinic agonism at M₄ receptors. As a result, clozapine use is limited to refractory schizophrenia patients.

Electroconvulsive therapy is considered a treatment of last resort in refractory schizophrenia and is rarely employed. Despite widespread clinical use of combining several antipsychotic agents, there is virtually no data supporting this practice, and metabolic risk increases with use of multiple antipsychotic agents (Correll et al., 2007). In one of few instances where a sound pharmacological rationale exists for combination treatment, the addition of a potent D₂ antagonist (e.g., risperidone, haloperidol) to maximally tolerated doses of clozapine (a weak D₂ blocker), the results have been decidedly mixed. That multiple antipsychotic agents ("polypharmacy") are commonly used in clinical practice attests to the limitations of current treatment. Lastly, antipsychotic drug therapy is the foundation of schizophrenia treatment, yet adequate management of schizophrenia patients requires a multimodal approach that also includes psychosocial, cognitive, and vocational rehabilitation to promote functional recovery.

Chemistry. In the past, the chemical structure of selected agents was informative regarding their antipsychotic activity, since most were derived from phenothiazine or butyrophenone structures. Phenothiazines have a tricyclic structure in which two benzene rings are linked by a sulfur and a nitrogen atom (Table 16–1). The chemically related thioxanthenes have a carbon in place of the nitrogen at position 10 with the R₁ moiety linked through a double bond. Substitution of an electron-withdrawing group at position 2 increases the antipsychotic efficacy of phenothiazines (e.g., chlorpromazine). The nature of the substituent at position 10 also influences pharmacological activity, and the phenothiazines and thioxanthenes can be divided into three groups on the basis of substitution at this site. Those with an aliphatic side chain include chlorpromazine and are relatively low in potency. Those with a piperidine ring in the side chain include thioridazine, which has lower EPS risk, possibly due to increased central antimuscarinic activity, but is rarely used due to concerns over QTc prolongation and risk of torsade de pointes. Several potent phenothiazine antipsychotic compounds have a piperazine group in the side chain (fluphenazine, perphenazine, and trifluoperazine) and have reduced affinity for H₁, M, and α₁ receptors. Those with a free hydroxyl can also be esterified to long-chain fatty acids to produce LAI antipsychotic medications (e.g., fluphenazine decanoate). Thioxanthenes also have aliphatic or piperazine side-chain substituents. Piperazine-substituted thioxanthenes include the high potency agent thiothixene. Since thioxanthenes have an olefinic double bond between the central-ring C10 and the side chain, geometric isomers exist. The cis isomers are more active. The antipsychotic phenothiazines and thioxanthenes have three carbon atoms interposed between position 10 of the central ring and the first amino nitrogen atom of the side chain at this position; the amine is always tertiary. Antihistaminic phenothiazines (e.g., promethazine) or strongly anticholinergic phenothiazines have only two carbon atoms separating the amino group from position 10 of the central ring. Metabolic N-dealkylation of the side chain or increasing the size of amino N-alkyl substituents reduces antidopaminergic and antipsychotic activity. Additional tricyclic antipsychotic agents are the benzepines, containing a 7-member central ring, of which loxapine (a dibenzoxazepine) and clozapine (a dibenzodiazepine) are available in the U.S. Clozapine-like atypical antipsychotic agents may lack a substituent on the aromatic ring (e.g., quetiapine, a dibenzothiazepine), have an analogous methyl substituent (olanzapine), or have an electron-donating substituent at position 8 (e.g., clozapine). In addition to their moderate potencies at DA receptors, clozapine-like agents interact with varying affinities at several other classes of receptors (α₁ and α₂ adrenergic, 5-HT_1A, 5-HT_2A, 5-HT_2C, M, H₁). The prototypical butyrophenone (phenylbutylpiperidine) antipsychotic is haloperidol, originally developed as a substituted derivative of the phenylpiperidine analgesic meperidine. An analogous compound, droperidol, is a very short-acting and highly sedating agent that was used almost exclusively for emergency sedation and anesthesia until QTc and torsade de pointes concerns substantially curtailed its use and the use of the related diphenylbutylpiperidine pimozide. Several other classes of heterocyclic compounds have antipsychotic effects, of which only the indole molindone has generated any interest due to its unusual association with modest weight loss (Sikich et al., 2008a). The enantiomeric, substituted benzamides are another group of heterocyclic compounds that are relatively selective antagonists at central D₂ receptors and have antipsychotic activity. Examples (not available in the U.S.) include sulpiride and its congener, amisulpride.

The introduction of clozapine stimulated research into agents with antipsychotic activity and low EPS risk. This search led to a series of atypical antipsychotic agents with certain pharmacological similarities to clozapine: namely lower affinity for D₂ receptors than typical antipsychotic drugs and high 5-HT₂ antagonist effects. The currently available atypical antipsychotic medications include the structurally related olanzapine, quetiapine, and clozapine; the benzisoxazoles risperidone, its active metabolite paliperidone, and iloperidone; ziprasidone (a benzisothiazolpiprazinylindolone derivative); asenapine (a dibenzo-oxepino pyrrole), and aripiprazole (a quinolinone derivative). Table 16–1.

Aripiprazole has an affinity for D₂ receptors only slightly less than DA itself, but its intrinsic activity is ~25% that of dopamine. As depicted in Figure 16–3, when dopamine is given in increasing concentrations in an in vitro model, 100% stimulation of the available D₂ receptors occurs (as measured by forskolin-stimulated cyclic AMP accumulation). In the absence of DA, aripiprazole produces a maximal level of D₂ activity ~25% that of DA (Burris et al., 2002). The potent D₂ antagonist haloperidol is capable of reducing dopamine's effect to zero, but when DA is incubated with increasing concentrations of aripiprazole, maximal inhibition of D₂ activity did not exceed 25% of the DA response, that is, the level of agonism provided by aripiprazole. Aripiprazole's capacity to stimulate D₂ receptors in brain areas where synaptic DA levels are limited (e.g., PFC neurons) or decrease dopaminergic activity when dopamine concentrations are high (e.g., mesolimbic cortex) is thought to be the basis for its clinical effects in schizophrenia. Evidence for its partial DA agonist properties is seen clinically as a reduction in serum prolactin levels. Unlike other antipsychotic agents, in which striatal D₂ occupancy (i.e., reduction in postsynaptic D₂ signal) > 78% is associated with EPS, aripiprazole requires significantly higher D₂ occupancy levels. Even with 100% receptor occupancy, aripiprazole's intrinsic dopaminergic agonism can generate a 25% postsynaptic signal, implying a maximal 75% reduction in DA neurotransmission, below the 78% threshold that triggers EPS in most individuals, although rare reports of acute dystonia exist, primarily in antipsychotic-naïve, younger patients.

Animal models of antipsychotic activity have evolved over a half-century of antipsychotic drug development to incorporate emerging knowledge of psychosis pathophysiology. Prior to the elucidation of D₂ antagonism as a common mechanism for antipsychotic agents, early behavioral paradigms exploited known animal effects that predicted clinical efficacy. As expected, these models, including catalepsy induction and blockade of apomorphine-induced stereotypic behavior, were subsequently found to be dependent on potent D₂ receptor antagonism (Lieberman et al., 2008). Clozapine failed these early trials and was not suspected to possess antipsychotic activity until experimental human use in the mid-1960s revealed it to be an effective treatment for schizophrenia, particularly in patients who had failed other antipsychotic medications, and with virtually absent EPS risk. By 1989, the pharmacological basis for some of these atypical properties, namely clinical efficacy without EPS induction, was found to result from a significantly weaker D₂ antagonism than existing antipsychotic agents, combined with potent 5-HT₂ antagonism. Subsequent research demonstrated that 5-HT_2A receptor antagonism is responsible for facilitation of dopamine release in both mesocortical and nigrostriatal pathways. The behavioral pharmacology of 5-HT_2A antagonism and related 5-HT_2C and 5-HT_1A receptor agonism has since been characterized, and relevant animal behavioral assays developed. Clozapine possesses activity at numerous other receptors including antagonism and agonism at various muscarinic receptor subtypes and antagonism at dopamine D₄ receptors; however, subsequent D₄ antagonists that did not also have D₂ antagonism lacked antipsychotic activity. Clozapine's active metabolite, N-desmethylclozapine, is a potent muscarinic M₁ agonist (Li et al., 2005). Although N-desmethylclozapine failed clinical trials as schizophrenia monotherapy, it increased interest in cholinergic agonists as primary treatments or adjunctive cognitive enhancing medications for schizophrenia. Important targets of agents in preclinical and clinical drug development include M₁ agonism (Janowsky and Davis 2009), M₄ agonism (Chan et al., 2008), and α7- and α4β2-nicotinic receptor agonism (Buchanan et al., 2007).

The glutamate hypofunction hypothesis of schizophrenia has led to novel animal models that examine the influence of proposed antipsychotic agents, including those in clinical development with agonist properties at metabotropic glutamate receptors mGlu₂ and mGlu₃ (Patil et al., 2007) and other subtypes. Atypical antipsychotic drugs are better than typical antipsychotic medications at reversing the negative symptoms, cognitive deficits and social withdrawal induced by glutamate antagonists. At the present time, however, it is unclear whether glutamate agonist that lack direct D₂ antagonist properties will prove as effective as serotonin-dopamine antagonists for schizophrenia treatment, or whether glutamate agonist mechanisms might be more useful when combined with D₂ modulation, much in the same manner as 5-HT_2A antagonism. Schizophrenia patients also exhibit specific neurophysiological abnormalities. For example, a disruption in the normal processing of sensory and cognitive information is exhibited in various preconscious inhibitory processes, including deficiencies in sensorimotor gating as assessed by prepulse inhibition (PPI) of the acoustic startle reflex. PPI is the automatic suppression of startle magnitude that occurs when the louder acoustic stimulus is preceded 30-500 ms by a weaker prepulse (Swerdlow et al., 2006). In schizophrenia patients, PPI is increased more robustly with atypical than typical antipsychotic agents, and in animal models, atypical antipsychotic agents are also more effective at opposing PPI disruption from NMDA antagonists. Increased understanding of the pharmacological basis for neurophysiological deficits provides another means for developing antipsychotic treatments that are specifically effective for schizophrenia and may not necessarily apply to other forms of psychosis.

Levels of central D₂ occupancy estimated by positron emission tomography (PET) brain imaging in patients treated with antipsychotic drugs support conclusions arising from laboratory studies that receptor occupancy predicts clinical efficacy, EPS, and serum level-clinical response relationships. Occupation of > 78% of D₂ receptors in the basal ganglia is associated with a risk of EPS across all dopamine antagonist antipsychotic agents, while occupancies in the range of 60-75% are associated with antipsychotic efficacy (Figure 16–2) (Kapur et al., 2000b). With the exception of aripiprazole, all atypical antipsychotic drugs at low doses have much greater occupancy of 5-HT_2A receptors (e.g., 75-99%) than typical agents (Table 16–2). Among atypical agents, clozapine has the highest ratio of 5-HT_2A/D₂ binding. Clozapine's D₂ occupancy 12-hours post-dose ranges from 51-63% (Kapur et al., 1999), providing evidence for its limited EPS risk. The trough D₂ occupancy for quetiapine is even lower (< 30%), but PET studies obtained 2-3 hours after dosing reveal D₂ receptor occupancies in the expected therapeutic range (54-64%), albeit transiently. Ziprasidone absorption is sensitive to the presence of food, but PET studies demonstrate that clinical efficacy occurs when D₂ occupancy exceeds 60%, which corresponds to a minimum daily dose of 120 mg (with food) (Mamo et al., 2004).

Among the typical antipsychotic drugs, receptor occupancy is best studied for haloperidol. Haloperidol has complex metabolism and is susceptible to modulation by CYP inhibitors, inducers, and CYP polymorphisms (Table 16–3), complicating the establishment of dose-response relationships in patients. Nonetheless, the use of serum levels can predict D₂ occupancy (Fitzgerald et al., 2000):

% D₂ receptor occupancy = 100 × (Plasma haloperidol ng/mL)/(0.40 ng/mL + Plasma haloperidol ng/mL)

Similar formulas also exist for several atypical antipsychotic drugs, although their plasma concentrations are rarely measured in the clinical setting.

D₃ and D₄ Receptors in the Basal Ganglia and Limbic System. The discovery that D₃ and D₄ receptors are preferentially expressed in limbic areas has led to efforts to identify selective inhibitors for these receptors that might have antipsychotic efficacy and low EPS risk. As previously noted, clozapine has modest selectivity for D₄ receptors, which are preferentially localized in cortical and limbic brain regions in relatively low abundance, and are up-regulated after repeated administration of most typical and atypical antipsychotic drugs (Tarazi et al., 2001). These receptors may contribute to clinical antipsychotic actions, but agents that are D₄ selective (e.g., sonepiprazole) or mixed D₄/5-HT_2A antagonists (e.g., fananserin) lack antipsychotic efficacy in clinical studies. In contrast to effects on D₂ and D₄ receptors, long-term administration of typical and atypical antipsychotic drugs does not alter D₃ receptor levels in rat forebrain regions (Tarazi et al., 2001). These findings suggest that D₃ receptors are unlikely to play a pivotal role in antipsychotic drug actions, perhaps because their avidity for endogenous DA prevents their interaction with antipsychotic agents. The subtle and atypical functional activities of cerebral D₃ receptors suggest that D₃ agonists rather than antagonists may have useful psychotropic effects, particularly in antagonizing stimulant-reward and dependence behaviors. Aripiprazole possesses affinity and intrinsic activity at D₃ receptors equivalent to that at D₂; the clinical advantage of this property is not readily apparent, although preclinical data suggest some effects on substance use.

The Role of Non-Dopamine Receptors for Atypical Antipsychotic Agents. The concept of atypicality was initially based on clozapine's absence of EPS within the therapeutic range, combined with a prominent role of 5-HT₂ receptor antagonism. As subsequent agents were synthesized using clozapine's 5-HT₂/D₂ ratio as a model, most of which possessed greater D₂ affinity and EPS risk than clozapine, there has been considerable debate on the definition of an atypical antipsychotic agent and its necessary properties. Aripiprazole in particular is problematic for the model based on ratios of 5-HT₂ to D₂, since its action as partial agonist necessitates very high D₂ affinity. Loxapine is another problematic agent for the model since its receptor pharmacology suggests atypical properties based on 5-HT₂/D₂ ratio; however, in clinical practice its use was associated with the expected higher level of EPS characteristic of typical antipsychotic drugs, perhaps related to the additive D₂ antagonist properties of the active metabolite amoxapine. These dilemmas have lead some to suggest abandonment or modification of the atypical/typical antipsychotic terminology, perhaps in lieu of the designation by generation (e.g., first, second, etc.), as is used with antibiotics, or some other organizing scheme (Gründer et al., 2009). Nonetheless, the term "atypical" persists in common usage and designates lesser (but not absent) EPS risk and other decreased effects of excessive D₂ antagonism, or more accurately, reduction in D₂-mediated neurotransmission.

5-HT_2A receptors are widely distributed, but antagonism exerts the greatest effect on prefrontal and basal ganglia DA release and midbrain noradrenergic outflow, with recent data implicating 5-HT_2A receptor polymorphisms in differential antidepressant response (McMahon et al., 2006). There are significant interrelationships between serotonin receptors, with evidence from animal studies that stimulation of either 5-HT_1A or 5-HT_2C receptors antagonizes the behavioral effects of 5-HT_2A agonists (Marek et al., 2003). This can be seen with agents that have direct effects on 5-HT_2A activity and also in studies of NMDA antagonists where there are opposing modulatory effects of 5-HT_2A and 5-HT_2C stimulation. By virtue of their impact on noradrenergic neurotransmission, 5-HT_2A antagonists pass many preclinical in vivo assays of antidepressant activity; conversely, 5-HT_2C antagonists exhibit a similar spectrum of antidepressant properties, although pure 5-HT_2A or 5-HT_2C agents are, by themselves, not effective antidepressants in human trials (Marek et al., 2003). Data also indicate that 5-HT_2C agonism decreases mesolimbic DA neurotransmission, a mechanism that is being explored in clinical trials of vabicaserin, a pure 5-HT_2C agonist. At the cellular level, stimulation of 5-HT_2A and 5-HT_1A receptors causes depolarization and hyperpolarization, respectively, of cortical pyramidal cells (Marek et al., 2003). No atypical antipsychotic agent is a potent agonist of 5-HT_1A receptors, but several, including clozapine, ziprasidone and aripiprazole are partial agonists. The extent to which this contributes to any clinical effect is unknown, but more potent selective 5-HT_1A agonists have anxiolytic effects and appear to exert procognitive effects in schizophrenia patients when added to existing antipsychotic treatment.

Based upon trials of the selective α₂ adrenergic antagonist idazoxan, researchers have postulated effects of α₂ blockade on mood, but ongoing research into this mechanism has not been promising. Risperidone is the one antipsychotic with relatively high affinity for α₂ adrenergic receptors, at which it is an antagonist; however, the clinical correlate of this unique profile is unclear. Any benefit on major depressive symptoms from low-dose risperidone augmentation of SSRI antidepressants is more likely conveyed by effects at the 5-HT_2A receptor, at which risperidone has > 100-fold greater potency than at α₂ adrenergic receptors.

Tolerance and Physical Dependence. As defined in Chapter 24, antipsychotic drugs are not addicting; however, tolerance to the antihistaminic and anticholinergic effects of antipsychotic agents usually develops over days or weeks. Loss of efficacy with prolonged treatment is not known to occur with antipsychotic agents; however, tolerance to antipsychotic drugs and cross-tolerance among the agents are demonstrable in behavioral and biochemical experiments in animals, particularly those directed toward evaluation of the blockade of dopaminergic receptors in the basal ganglia (Tarazi et al., 2001). This form of tolerance may be less prominent in limbic and cortical areas of the forebrain. One correlate of tolerance in striatal dopaminergic systems is the development of receptor supersensitivity (mediated by upregulation of supersensitive DA receptors) referred to as D₂^High receptors (Lieberman et al., 2008). These changes may underlie the clinical phenomenon of withdrawal-emergent dyskinesias and may contribute to the pathophysiology of tardive dyskinesia. This may also partly explain the ability of certain chronic schizophrenia patients to tolerate high doses of potent DA antagonists with limited EPS.

Elimination from the plasma may be more rapid than from sites of high lipid content and binding, notably the CNS, as evidenced by PET pharmacokinetic studies that demonstrate half-lives in the CNS that exceed those in plasma. For example, mean single-dose plasma half-lives of olanzapine and risperidone are 24.2 and 10.3 hours respectively, whereas a 50% reduction from peak striatal D₂ receptor occupancy requires 75.2 hours for olanzapine and 66.6 hours for risperidone (Tauscher et al., 2002). Similar discrepancies are seen between the time course of plasma levels and occupancy of extrastriatal D₂ and 5-HT_2A receptors (Tauscher et al., 2002). Metabolites of long acting injectable medications have been detected in the urine several months after drug administration was discontinued. Slow removal of drug may contribute to the typical delay of exacerbation of psychosis after stopping drug treatment. Depot decanoate esters of fluphenazine and haloperidol, paliperidone palmitate, as well as risperidone-impregnated microspheres, are absorbed and eliminated much more slowly than are oral preparations. For example, the t_1/2 of oral fluphenazine is ~20 hours while the IM decanoate ester has a t_1/2 of 14.3 days; oral haloperidol has a t_1/2 of 24-48 hours in CYP2D6-extensive metabolizers (de Leon et al., 2004), while haloperidol decanoate has a t_1/2 of 21 days (Altamura et al., 2003); paliperidone palmitate has a t_1/2 of 25-49 days compared to an oral paliperidone t_1/2 of 23 hours. Clearance of fluphenazine and haloperidol decanoate following repeated dosing can require 6-8 months. Effects of LAI risperidone (RISPERDAL CONSTA) are delayed for 4 weeks because of slow biodegradation of the microspheres and persist for at least 4-6 weeks after the injections are discontinued (Altamura et al., 2003). The dosing regimen recommended for starting patients on LAI paliperidone generates therapeutic levels in the first week, obviating the need for routine oral antipsychotic supplementation.

With the exception of asenapine, paliperidone and ziprasidone, all antipsychotic drugs undergo extensive phase 1 metabolism by CYPs and subsequent phase 2 glucuronidation, sulfation, and other conjugations (Table 16–3). Hydrophilic metabolites of these drugs are excreted in the urine and to some extent in the bile. Most oxidized metabolites of antipsychotic drugs are biologically inactive; a few (e.g., P88 metabolite of iloperidone, hydroxy metabolite of haloperidol 9-OH risperidone, N-desmethylclozapine, and dehydroaripiprazole) are active. These active metabolites may contribute to biological activity of the parent compound and complicate correlating serum drug levels with clinical effects. The active metabolite of risperidone, paliperidone (9-OH risperidone), is already the product of oxidative metabolism, and 59% is excreted unchanged in urine with a lesser amount (32%) excreted as metabolites or via phase 2 metabolism (≤ 10%). Ziprasidone's primary metabolic pathway is through the aldehyde oxidase system that is neither saturable nor inhibitable by commonly encountered xenobiotics, with ~ one-third of ziprasidone's metabolism through CYP3A4 (Meyer, 2007). Asenapine is metabolized primarily through glucuronidation (UGT4), with a minor contribution from CYP1A2. The potential for drug-drug interactions is covered in "Adverse Effects and Drug Interactions" later in the chapter.

Absorption for most agents is quite high, and concurrent administration of anticholinergic anti-parkinsonian agents does not appreciably diminish intestinal absorption. Most orally-disintegrating tablets (ODT) and liquid preparations provide similar pharmacokinetics since there is little mucosal absorption and effects depend on swallowed drug. Asenapine remains the only exception. It is only available as an ODT preparation administered sublingually, and all absorption occurs via oral mucosa, with bioavailability of 35% by this route. If swallowed, the first pass effect is > 98%, indicating that drug swallowed with oral secretions is not bioavailable. Intramuscular administration avoids much of the first-pass enteric metabolism and provides measurable concentrations in plasma within 15-30 minutes. Most agents are highly protein bound, but this protein binding may include glycoprotein sites. Kinetic studies indicate that antipsychotic drugs do not significantly displace other prealbumin- or albumin-bound medications. Antipsychotic medications are predominantly highly lipophilic with apparent volumes of distribution as high as 20 L/kg.

Anxiety Disorders. There are two anxiety disorders in which double-blind, placebo-controlled trials have shown benefit of adjunctive treatment with antipsychotic drugs: obsessive compulsive disorder (OCD) and post-traumatic stress disorder (PTSD). While SSRI antidepressants remain the only psychotropic medication with FDA approval for PTSD treatment, adjunctive low-dose quetiapine, olanzapine, and particularly risperidone significantly reduce the overall level of symptoms in SSRI-resistant PTSD (Bartzokis et al., 2005). OCD patients with limited response to the standard 12-week regimen of high dose SSRI also benefit from adjunctive risperidone (mean dose 2.2 mg), even in the presence of comorbid tic disorders (McDougle et al., 2000). For generalized anxiety disorder, double-blind placebo controlled clinical trials demonstrate efficacy for quetiapine as monotherapy, and for adjunctive low-dose risperidone.

Tourette's Disorder. The ability of antipsychotic drugs to suppress tics in patients with Tourette's disorder has been known for decades, and relates to reduced D₂ neurotransmission in basal ganglia sites. In prior decades, the use of low-dose, high-potency typical antipsychotic agents (e.g., haloperidol, pimozide) was the treatment of choice, but these nonpsychotic patients were extremely sensitive to the impact of DA blockade on cognitive processing speed, and on reward centers. Safety concerns regarding pimozide's QTc prolongation and increased risk for ventricular arrhythmias have large ended its clinical use. While lacking FDA approval for tic disorders, risperidone and aripiprazole have indications for child and adolescent schizophrenia and bipolar disorder (acute mania) treatment, and these agents (as well as ziprasidone) have published data supporting their use for tic suppression. Given the enormous sensitivity of preadolescent and teenage patients to antipsychotic-induced weight gain, aripiprazole has an advantage in this patient population due to somewhat lower risk for weight gain, and low risk for hyperprolactinemia or concerns over QTc effects.

Huntington's Disease. Huntington's disease is another neuropsychiatric condition, which, like tic disorders, is associated with basal ganglia pathology. DA blockade can suppress the severity of choreoathetotic movements, but is not strongly endorsed due to the risks associated of excessive DA antagonism that outweigh the marginal benefit. Inhibition of the vesicular monoamine transporter type 2 (VMAT2) with tetrabenazine has replaced DA receptor blockade in the management of chorea (Chapter 22).

Autism. Autism is a disease whose neuropathology is incompletely understood, but in some patients is associated with explosive behavioral outbursts, and aggressive or self-injurious behaviors that may be stereotypical. Risperidone has FDA approval for irritability associated with autism in child and adolescent patients ages 5-16, with common use for disruptive behavior problems in autism and other forms of mental retardation. Initial risperidone daily doses are 0.25 mg for patients weighing < 20 kg, and 0.5 mg for others, with a target dose of 0.5 mg/day in those < 20 kg weight, and 1.0 mg/day for other patients, with a range 0.5-3.0 mg/day.

Anti-emetic Use. Most antipsychotic drugs protect against the nausea- and emesis-inducing effects of DA agonists such as apomorphine that act at central DA receptors in the chemoreceptor trigger zone of the medulla. Antipsychotic drugs are effective antiemetics already at low doses. Drugs or other stimuli that cause emesis by an action on the nodose ganglion, or locally on the GI tract, are not antagonized by antipsychotic drugs, but potent piperazines and butyrophenones are sometimes effective against nausea caused by vestibular stimulation. The commonly used antiemetic phenothiazines are weak DA antagonists (e.g., prochlorperazine) without antipsychotic activity, but can be associated with EPS or akathisia.

The treatment of acute dystonia and antipsychotic-induced parkinsonism involves the use of anti-parkinsonian agents, although dose reduction should be considered as the initial strategy for parkinsonism. Muscarinic cholinergic receptors modulate nigrostriatal DA release, with blockade increasing synaptic DA availability. Important issues in the use of anticholinergics include the negative impact on cognition and memory, peripheral antimuscarinic adverse effects (e.g., urinary retention, dry mouth, cycloplegia, etc.), and the relative risk of exacerbating tardive dyskinesia. In patients receiving chronic anticholinergic therapy, there are also short-term risks of cholinergic rebound following abrupt anticholinergic withdrawal, which may include sleep disturbance (vivid dreams, nightmares) and also increased EPS if the patient continues to receive antipsychotic treatment. For parenteral administration, diphenhydramine (25-50 mg IM) and benztropine (1-2 mg IM) are the agents most commonly used. Diphenhydramine is an antihistamine that also possesses anticholinergic properties. Benztropine combines a benzhydryl group with a tropane group to create a compound which is more anticholinergic than trihexyphenidyl but less antihistaminic than diphenhydramine. The clinical effect of a single dose lasts 5 hours, thereby requiring 2 or 3 daily doses. Dosing usually starts at 0.5-1 mg bid, with a daily maximum of 6 mg, although slightly higher doses are used in rare circumstances. The piperidine compound trihexyphenidyl was one of the first synthetic anticholinergic agents available, and replaced the belladonna alkaloids for treatment of PD in the 1940s due to its more favorable side effect profile. Trihexyphenidyl inhibits the presynaptic DA reuptake transporter, and therefore has a concomitant higher risk of abuse than the antihistamines or benztropine. Trihexyphenidyl has good GI absorption, achieving peak plasma levels in 1-2 hours, with a serum t_1/2 ~10-12 hours, generally necessitating multiple daily dosing to achieve satisfactory clinical results. The total daily dosage range is 5-15 mg, given 2-3 times a day as divided doses. Biperiden (akineton) is another drug in this class.

Amantadine (symmetrel), originally marketed as an antiviral agent toward influenza A, represents the most commonly used non-anticholinergic medication for antipsychotic-induced parkinsonism. Its mechanism of action is unclear but appears to involve presynaptic DA reuptake blockade, facilitation of DA release, postsynaptic DA agonism, and receptor modulation. These properties are sufficient to reduce symptoms of drug-induced parkinsonism, and only rarely have been reported to exacerbate psychotic symptoms. Amantadine is well absorbed after oral administration, with peak levels achieved 1-4 hours after ingestion; clearance is renal, with > 90% recovered unmetabolized in the urine. The plasma t_1/2 is 12-18 hours in healthy young adults, but is sufficiently longer in those with renal impairment and the elderly that a 50% dose reduction is recommended. Starting dosage is 100 mg orally bid in healthy adults, which may be increased to 200 mg bid. A dose of 100 mg bid yields peak plasma levels of 0.5-0.8 μg/mL and trough levels of 0.3 μg/mL. Toxicity is seen at serum levels of 1.0-5.0 μg/mL. Amantadine's primary advantage, especially in older patients, is avoidance of adverse CNS and peripheral anticholinergic effects.

Tardive dyskinesia is characterized by stereotyped, repetitive, painless, involuntary, quick choreiform (tic-like) movements of the face, eyelids (blinks or spasm), mouth (grimaces), tongue, extremities, or trunk. There are varying degrees of slower athetosis (twisting movements), while tardive dystonia and tardive akathisia are rarely encountered as use of high-dose, high-potency typical antipsychotic medications has abated. The movements all disappear in sleep (as do many other extrapyramidal syndromes), vary in intensity over time, and are dependent on the level of arousal or emotional distress, sometimes reappearing during acute psychiatric illnesses following prolonged disappearance. The dyskinetic movements can be suppressed partially by use of a potent DA antagonist, but such interventions over time may worsen the severity, as this was part of the initial pharmacological insult. Switching patients from potent D₂ antagonists to weaker agents, especially clozapine has at times proven effective. When possible, drug discontinuation may be beneficial, but usually cannot be offered to schizophrenia patients. Trials of the antioxidant vitamin E have proved ineffective and are not recommended, especially in light of adverse cardiovascular effects from chronic vitamin E exposure.

Unlike antipsychotic-induced parkinsonism and acute dystonia, the phenomenology and treatment of akathisia suggests involvement of structures outside the nigrostriatal pathway. Akathisia was seen quite commonly during treatment with high doses of high potency typical antipsychotic drugs, but also can be seen with atypical agents, including those with weak D₂ affinities (e.g., quetiapine), and aripiprazole. Among the predominantly antipsychotic drug-naïve population with major depression in clinical studies with adjunctive aripiprazole, akathisia incidence was 23% using a starting dose of 5 mg, suggesting that dopaminergic partial agonism, rather than antagonism, may be etiologic for this medication (Berman et al., 2007; Marcus et al., 2008). Despite the association with D₂ blockade, akathisia does not have a robust response to anti-parkinsonian drugs, so other treatment strategies must be employed, including the use of high-potency benzodiazepines (e.g., clonazepam), nonselective β blockers with good CNS penetration (e.g., propranolol), and also dose reduction, or switching to another antipsychotic agent. That clonazepam and propranolol have significant cortical activity and are ineffective for other forms of EPS, points to an extrastriatal origin for akathisia symptoms.

The rare neuroleptic malignant syndrome (NMS) resembles a very severe form of parkinsonism, with signs of autonomic instability (hyperthermia and labile pulse, blood pressure, and respiration rate), stupor, elevation of creatine kinase in serum, and sometimes myoglobinemia with potential nephrotoxicity. At its most severe, this syndrome may persist for more than a week after the offending agent is discontinued, and is associated with mortality. This reaction has been associated with various types of antipsychotic agents, but its prevalence may be greater when relatively high doses of potent agents are used, especially when they are administered parenterally. Aside from cessation of antipsychotic treatment and provision of supportive care, including aggressive cooling measures, specific pharmacological treatment is unsatisfactory, although administration of dantrolene and the dopaminergic agonist bromocriptine may be helpful. While dantrolene also is used to manage the syndrome of malignant hyperthermia induced by general anesthetics, the neuroleptic-induced form of hyperthermia probably is not associated with a defect in Ca²⁺ metabolism in skeletal muscle. There are anecdotal reports of NMS with atypical antipsychotic agents, but this syndrome is now rarely seen in its full presentation.

The hyperprolactinemia from antipsychotic drugs is rapidly reversible when the drugs are discontinued. Hyperprolactinemia can directly induce breast engorgement and galactorrhea. Approximately 33% of human breast cancers are prolactin-dependent in vitro, a factor of potential importance if the prescription of these drugs is contemplated in a patient with previously detected breast cancer. By suppressing the secretion of gonadotropins and sex steroids, hyperprolactinemia can cause amenorrhea in women and sexual dysfunction or infertility in men. The development of amenorrhea is of concern as it represents low serum estradiol levels and ongoing risk for bone density loss. The development of amenorrhea becomes a sensitive marker for sex hormone levels, and should prompt clinical action, while asymptomatic measurable increases in serum prolactin levels do not necessarily merit any intervention. Dose reduction can be tried to decrease serum prolactin levels, but caution must be exercised to keep treatment within the antipsychotic therapeutic range. When switching from offending antipsychotic agents is not feasible, bromocriptine can be employed. There is also anecdotal evidence that aripiprazole augmentation may be effective.

All DA antagonist antipsychotic drugs possess antiemetic properties by virtue of their actions at the medullary chemoreceptor trigger zone (Figure 46–4). The D₂ partial agonists aripiprazole (and bifeprunox), however, can be associated with nausea. For aripiprazole this effect was noted in clinical trials of oral medication for acute mania, bipolar maintenance, or schizophrenia (nausea incidence 15% for aripiprazole vs. 11% for placebo; vomiting incidence 11% for aripiprazole vs. 6% for placebo), and also in studies of pediatric bipolar mania in patients ages 10-17 (nausea incidence 11%) and in acute IM trials for agitation in schizophrenia or acute mania (nausea incidence 9%). Bifeprunox possessed a level of intrinsic DA agonism that was slightly greater than that for aripiprazole (25-28%), but was significant enough to cause clinical problems with nausea and vomiting, that a 10-day titration from an initial starting dose of 0.125 mg was necessary to reach the proposed effective dosage range of 10-30 mg (Glick and Peselow, 2008).

There is an extended-release quetiapine preparation available, with markedly reduced C_max compared to the immediate-release form. Immediate-release quetiapine generates peak serum levels > 800 ng/mL within 2 hours of ingestion, while the extended release preparation has C_max ~50% lower, with peak serum levels seen 4-8 hours after ingestion. In clinical practice the onset of sedation for extended-release quetiapine is delayed for at least 3 hours after oral administration, and subjectively the sedation is much less profound than with the immediate-release form (Mamo et al., 2008).

Rapid discontinuation of sedating antihistaminic antipsychotic drugs is inevitably followed by significant complaints of rebound insomnia and sleep disturbance. If discontinuation of sedating antipsychotic treatment is deemed necessary, except for emergency cessation of clozapine for agranulocytosis, the medication should be tapered slowly over 4-8 weeks, and the clinician should be prepared to utilize a sedative when the end of the titration is reached. Generous dosing of another antihistamine (hydroxyzine), or the anticholinergic antihistamine diphenhydramine are reasonable replacement sedative medications, but others can used, including benzodiazepines, non-benzodiazepine hypnotics that act at specific benzodiazepine sites (e.g., zolpidem, eszopiclone), and sedating antidepressants (e.g., trazodone). Sedation may be useful during acute psychosis, but excessive sedation can interfere with patient evaluation, may prolong emergency room and psychiatric hospital stays unnecessarily, and is poorly tolerated among elderly dementia and delirium patients, so appropriate caution must be exercised with the choice of agent and the dose.

Weight gain is a significant problem during long-term use of antipsychotic drugs and represents a major barrier to medication adherence, as well as a significant threat to the physical and emotional health of the patient. Weight gain has effectively replaced concerns over EPS as the adverse effect causing the most consternation among patients and clinicians alike. Appetite stimulation is the primary mechanism involved, with little evidence to suggest that decreased activity (due to sedation) is a main contributor to antipsychotic-related weight gain (Gothelf et al., 2002). Recent animal studies indicate that medications with significant H₁ antagonism induce appetite stimulation through effects at hypothalamic sites (Kim et al., 2007). The low-potency phenothiazine chlorpromazine, and the atypical antipsychotic drugs olanzapine and clozapine, are the agents of highest risk, but weight gain of some extent is seen with nearly all antipsychotic drugs, partly related to the fact that acutely psychotic patients may lose weight; in placebo-controlled acute schizophrenia trials, the placebo cohort inevitably loses weight (Meyer, 2001). For clozapine and olanzapine, massive weight gains of 50 kg or more are not uncommon, and mean annual weight gains of 13 kg are reported in schizophrenia clinical trials, with 20% of subjects gaining ≥ 20% of baseline weight. High-potency typical antipsychotic drugs (e.g., haloperidol, fluphenazine), and newer atypical antipsychotics asenapine, ziprasidone, and aripiprazole, are associated with mean annual weight gains < 2 kg in schizophrenia patients, with mean gains of 2.5-3 kg noted for iloperidone, risperidone, and quetiapine (Meyer, 2001). For unknown reasons, molindone is associated with modest weight loss. Younger and antipsychotic drug-naïve patients are much more sensitive to the weight gain from all antipsychotic agents, including those which appear roughly weight neutral in adult studies, leading some to conjecture that DA blockade may also play a small additive role in weight gain. 5-HT_2C antagonism is also thought to play an additive role in promoting weight gain for medications that possess high H₁ affinities (e.g., clozapine, olanzapine), but appears to have no effect in the absence of significant H₁ blockade, as seen with ziprasidone, a weight-neutral antipsychotic drug with extremely high 5-HT_2C affinity. Switching to more weight-neutral medications can achieve significant results; however, when not feasible or when unsuccessful, behavioral strategies must be employed, and should be considered for all chronically mentally ill patients given the high obesity prevalence in this patient population.

Low-potency phenothiazines were known to elevate serum triglyceride values, but this effect was not seen with high-potency agents (Meyer and Koro, 2004). As atypical antipsychotic drugs became more widely used, significant increases in fasting triglyceride levels were noted during clozapine and olanzapine exposure, and to a lesser extent, with quetiapine (Meyer et al., 2008). Mean increases during chronic treatment of 50-100 mg/dL are common, with serum triglyceride levels exceeding 7000 mg/dL in some patients. Effects on total cholesterol and cholesterol fractions are significantly less, but show expected associations related to agents of highest risk: clozapine, olanzapine, and quetiapine. Risperidone and paliperidone have fewer effects on serum lipids, while asenapine, iloperidone, aripiprazole, and ziprasidone appear to have none (Meyer and Koro, 2004). Weight gain in general may induce deleterious lipid changes, but there is compelling evidence to indicate that antipsychotic-induced hypertriglyceridemia is a weight-independent adverse event that temporally occurs within weeks of starting an offending medication, and which similarly resolves within 6 weeks after medication discontinuation. The finding that serum triglycerides may change 70-80 mg/dL during a period when weight has changed relatively little propelled the search for adiposity-independent physiological mechanisms to explain this phenomenon.

In individuals not exposed to antipsychotic drugs, elevated fasting triglycerides are a direct consequence of insulin resistance since insulin-dependent lipases in fat cells are normally inhibited by insulin. As insulin resistance worsens, inappropriately high levels of lipolysis lead to the release of excess amounts of free fatty acids that are hepatically transformed into triglyceride particles (Smith, 2007). Elevated fasting triglyceride levels thus become a sensitive marker of insulin resistance, leading to the hypothesis that the triglyceride increases seen during antipsychotic treatment are the result of derangements in glucose-insulin homeostasis. The ability of antipsychotic drugs to induce hyperglycemia was first noted during low-potency phenothiazine treatment, with chlorpromazine occasionally exploited for this specific property as adjunctive presurgical treatment for insulinoma. As atypical antipsychotic drugs found widespread use, numerous case series documented the association of new-onset diabetes and diabetic ketoacidosis associated with treatment with atypical antipsychotic drugs, with most of cases observed during clozapine and olanzapine therapy (Jin et al., 2002, 2004). Analysis of the FDA MedWatch database found that reversibility was high upon drug discontinuation (~78%) for olanzapine- and clozapine-associated diabetes and ketoacidosis, supporting the contention of a drug effect. Comparable rates for risperidone and quetiapine were significantly lower. The mechanism by which antipsychotic drugs disrupt glucose-insulin homeostasis is not known, but recent in vivo animal experiments document immediate dose-dependent effects of clozapine and olanzapine on whole-body and hepatic insulin sensitivity (Houseknecht et al., 2007).

Antipsychotic-induced weight gain, and other diabetes risk factors (e.g., age, family history, gestational diabetes, obesity, race, ethnicity, smoking) all contribute to metabolic dysfunction. There may also be inherent disease-related mechanisms that increase risk for metabolic disorders among patients with schizophrenia, but the medication itself is the primary modifiable risk factor. As a result, all atypical antipsychotic drugs have a hyperglycemia warning in the drug label in the U.S., although there is essentially no evidence that asenapine, iloperidone, aripiprazole, and ziprasidone cause hyperglycemia. Use of the metabolically more benign agents is recommended for the initial treatment of all patients where long-term treatment is expected. Clinicians should obtain baseline metabolic data, including fasting glucose, lipid panel, and also waist circumference, given the known association between central obesity and future type 2 diabetes risk. Ongoing follow-up of metabolic parameters is commonly built into psychiatric charts and community mental health clinic procedures to insure that all patients receive some level of metabolic monitoring. As with weight gain, the changes in fasting glucose and lipids should prompt reevaluation of ongoing treatment, institution of measures to improve metabolic health (diet, exercise, nutritional counseling), and consideration of switching antipsychotic agents.

Cardiac arrhythmia is the most common etiology for SCD, but it is important to determine whether the arrhythmia is primary or secondary to structural changes related to cardiomyopathy, myocarditis, or acute myocardial infarction. Secondary ventricular arrhythmia is probably the most frequent form of fatal tachyarrhythmia, but the exact distribution of SCD deaths by etiology among antipsychotic-treated patients is unclear in the absence of large autopsy samples. The true incidence of drug-induced ventricular arrhythmia can only be estimated, due partly to underreporting, and the fact that drug-induced torsade de pointes is rarely captured with confirmatory EKG (Nielsen and Toft, 2009).

Multiple ion channels are involved in the depolarization and repolarization of cardiac ventricular cells, which are discussed in detail in Chapter 29. Myoycte depolarization is seen as the QRS complex on EKG, and is primarily mediated by ion channels that permit rapid Na⁺ influx. Antagonism of these voltage-gated Na⁺ channels causes QRS widening and an increase in the PR interval, with increased risk for ventricular arrhythmia. Thioridazine has been shown to inhibit Na⁺ channels at high dosages, but other antipsychotic medications have not (Nielsen and Toft, 2009). Repolarization is mediated in part by K⁺ efflux via two channels: the rapid I_kr and the slow I_ks channel. The I_kr channel is encoded by the human-ether-a-related-go-go gene (HERG), polymorphisms of which are involved in the congenital long QT syndrome associated with syncope and SCD. Many antipsychotic drugs block the I_kr channel to an extent comparable with that seen in congenital long QT syndrome. Antagonism of I_kr channels is the mechanism responsible for most cases of drug-induced QT prolongation, and is the suspected mechanism for the majority of antipsychotic-induced sudden cardiac deaths (Nielsen and Toft, 2009).

Aside from individual agents, where anecdotal and pharmacosurveillance data indicate risk for torsade de pointes (e.g., thioridazine, pimozide), most of the commonly used newer antipsychotic agents are associated with known risk for ventricular arrhythmias, including ziprasidone in overdose up to 12,000 mg. One exception is sertindole, an agent not available in the U.S. that was withdrawn in 1998 based on anecdotal reports of torsade de pointes, and then reintroduced in Europe in 2006 with strict EKG monitoring guidelines (Nielsen and Toft, 2009). Although in vitro data revealed sertindole's affinity for the K⁺ rectifier channel, several epidemiological studies published over the past decade were unable to confirm an increased risk of sudden death due to sertindole exposure, thereby providing justification for its reintroduction. Aside from sertindole, the apparent safety of newer antipsychotic medications appears at odds with retrospective findings of SCD among antipsychotic users (Ray et al., 2009), thus confronting the clinician with contradictory information regarding antipsychotic drug-related SCD risk, and leading to conflicting clinical recommendations. Currently, there are no data that would suggest a benefit of routine EKG monitoring for prevention of SCD among antipsychotic drug users.

Other Adverse Effects. Seizure risk is an unusual adverse effect of antipsychotic drugs, with anecdotal reports of uncertain causality present for many agents. In the U.S., there is a class label warning for seizure risk on all antipsychotic agents, with reported incidences well below 1%. Among commonly used newer antipsychotic drugs, only clozapine has a dose-dependent seizure risk, with an incidence of 3-5% per year. The structurally related olanzapine had an incidence of 0.9% in premarketing studies. Seizure disorder patients who commence antipsychotic treatment must receive adequate prophylaxis, with consideration given to avoiding carbamazepine and phenytoin due to their capacity to induce CYPs and P-glycoprotein. Carbamazepine is also contraindicated during clozapine treatment due to its bone marrow effects, particularly leucopenia. Redistribution and increased spacing of doses to minimize high peak serum clozapine levels can help, but patients may eventually require antiseizure medication. Valproic acid derivatives (e.g., divalproex sodium) are often used, but will compound clozapine-associated weight gain.

Clozapine possesses a host of unusual adverse effects aside from seizure induction, the most concerning of which is agranulocytosis. Clozapine's introduction in the U.S. was based on its efficacy in refractory schizophrenia, but came with FDA-mandated CBC monitoring that is overseen by industry-created registries. Now that several generic forms of clozapine are available in addition to proprietary CLOZARIL, clinicians must verify with each manufacturer the history of prior exposure. The overall agranulocytosis incidence is slightly under 1%, with highest risk during the initial 6 months of treatment, peaking at months 2-3 and diminishing rapidly thereafter. The mechanism is immune mediated, and patients who have verifiable clozapine-related agranulocytosis should not be rechallenged. Increased risk is associated with certain HLA types and advanced age. An extensive algorithm guiding clinical response to agranulocytosis, and lesser forms of neutropenia, is available from manufacturer web sites, and must be followed, along with the current recommended CBC monitoring frequency.

While rarely used due to its risk of QTc prolongation, thioridazine is also associated with pigmentary retinopathy at daily doses ≥ 800 mg/day. Low-potency phenothiazines are associated with the development of photosensitivity, which necessitated warnings regarding sun exposure. Phenothiazines are also associated with development of a cholestatic picture on laboratory assessments (e.g., elevated alkaline phosphatase), and rarely elevations in hepatic transaminases.

Increased Mortality in Dementia Patients. Perhaps the least understood adverse effect is the increased risk for cerebrovascular events and all-cause mortality among elderly dementia patients exposed to antipsychotic medications. All antipsychotic agents carry a mortality warning in the drug label regarding their use in dementia patients. The cerebrovascular adverse event rates in 10-week dementia trials range from 0.4-0.6% for placebo to 1.3-1.5% for risperidone, olanzapine and aripiprazole (Jeste et al., 2008). The mortality warning indicates a 1.6-1.7 fold increased mortality risk for drug versus placebo. Mortality is due to heart failure, sudden death, or pneumonia. The underlying etiology for antipsychotic-related cerebrovascular and mortality risk is unknown, but the finding of virtually equivalent mortality risk for typical agents compared to atypical antipsychotic drugs (including aripiprazole) suggests an impact of reduced D₂ signaling regardless of individual antipsychotic mechanisms.

Changes in smoking status can be especially problematic for clozapine-treated patients, and will alter serum levels by 50% or more. Within 2 weeks of smoking discontinuation (e.g., hospitalization in nonsmoking environment), the absence of aryl hydrocarbons will cause upregulated CYP1A2 activity to return to baseline levels, with a concomitant rise in serum clozapine concentrations (Rostami-Hodjegan et al., 2004). For smoking patients with high serum clozapine levels, this loss of enzyme induction may result in nonlinear increases in clozapine levels, with potentially catastrophic results. Conversely, patients discharged from nonsmoking wards to the community will be expected to resume smoking behavior, with an expected 50% decrease in clozapine levels. Monitoring of serum clozapine concentrations, anticipation of changes in smoking habits, and dosage adjustment can minimize development of subtherapeutic or supratherapeutic levels.

Use in Pediatric Populations. Both risperidone and aripiprazole have indications for child and adolescent bipolar disorder (acute mania) for ages 10-17, and for adolescent schizophrenia (ages 13-17). Risperidone and aripiprazole are FDA-approved for irritability associated with autism in child and adolescent patients ages 5-16. As discussed in the sections on adverse effects, antipsychotic drug-naïve patients and younger patients are more susceptible to EPS and to weight gain. Use of the minimum effective dose can minimize EPS risk; aripiprazole and ziprasidone have the lowest risk, olanzapine the highest. Risperidone's effects on prolactin must be monitored by clinical inquiry, but long-term follow-up studies indicate some tolerance to hyperprolactinemia, with levels after 12 months of exposure significantly lower than peak levels, and close to baseline. Delayed sexual maturation was not seen in adolescents in clinical trials with risperidone, but should nevertheless be monitored.

Use in Geriatric Populations. The increased sensitivity to EPS, orthostasis, sedation, and anticholinergic effects are important issues for the geriatric population, and often dictate the choice of antipsychotic medication. Avoidance of drug-drug interactions is also important, as older patients on numerous concomitant medications have multiple opportunities for interactions. Dose adjustment can offset known drug-drug interactions, but clinicians must be attentive to changes in concurrent medications and the potential pharmacokinetic consequences. Vigilance must also be maintained for the additive pharmacodynamic effects of α₁ adrenergic, antihistaminic, and anticholinergic properties of other agents. Elderly patients have an increased risk for tardive dyskinesia and parkinsonism, with TD rates ~5-fold higher than those seen with younger patients. With typical antipsychotics, the reported annual TD incidence among elderly patients in 20-25% compared to 4-5% for younger patients. With atypical antipsychotic, the annual TD rate in elderly patients is much lower (2-3%). Increased risk for cerebrovascular events and all-cause mortality is also seen in elderly patients with dementia (see "Increased Mortality in Dementia Patients"). Compared to younger patients, antipsychotic-induced weight gain is lower in elderly patients.

Use During Pregnancy and Lactation. Antipsychotic agents carry pregnancy class B or C warnings. Human data from anecdotal case reports and manufacturer registries indicate limited or no patterns of toxicity and no consistent increased rates of malformations. Nonetheless, the use of any medication during pregnancy must be balanced by concerns over fetal impact, especially first-trimester exposure, and the mental health of the mother. Haloperidol is often cited as the agent with the best safety record based on decades of accumulated human exposure reports, but newer antipsychotic drugs, such as risperidone, have not generated any signals of concern (Viguera et al., 2009). As antipsychotic drugs are designed to cross the blood-brain barrier, all have high rates of placental passage. Placental passage ratios are estimated to be highest for olanzapine (72%), followed by haloperidol (42%), risperidone (49%), and quetiapine (24%). Neonates exposed to olanzapine, the atypical agent with highest placental passage ratio, exhibit a trend towards greater neonatal intensive care unit admission (Viguera et al., 2009). Use in lactation presents a separate set of concerns due to the low level of infant hepatic catabolic activity in the first 2 postpartum months. While breast milk presents an important source of protective antibodies for the baby, the inability of the newborn to adequately metabolize xenobiotics presents a significant risk for antipsychotic drug toxicity. The available data do not provide adequate guidance on choice of agent.

Acute IM forms exist for many typical antipsychotic drugs, and also for aripiprazole, ziprasidone, and olanzapine, with the latter being the most sedating due to its antihistaminic and anticholinergic properties. The standard IM doses are 9.75 mg for aripiprazole, 10 mg for olanzapine, and 20 mg for ziprasidone (Altamura et al., 2003). The 20-mg IM ziprasidone dose generates serum levels that exceed those from 160 mg/day orally, but has not been associated with reports of torsade de pointes or cardiac dysrhythmia. There are numerous LAI (long-acting injectable) formulations of typical antipsychotics, but in the U.S., the only available LAI typical agents are fluphenazine and haloperidol (as decanoate esters), with usual dosages 12.5-25 mg IM every 2 weeks and 50-200 mg (not to exceed 100 mg as the initial dose) IM monthly, respectively. Haloperidol decanoate has more predictable kinetics, while fluphenazine decanoate's serum level can increase within days after administration and induce adverse effects (Altamura et al., 2003). EPS remains a significant barrier for using these agents. LAI risperidone was studied at doses of 25, 50, and 75 mg IM biweekly, but the higher dose was found to be without increased efficacy and with greater EPS rates. Current available doses, all distributed as complete, single-dose kits, are 12.5, 25, 37.5, and 50 mg, given as a 2-mL water-based injection. LAI risperidone is impregnated into organic microspheres that require several weeks to liberate free drug. The impact of any dosage change (or missed dose) of LAI risperidone is seen 4 weeks later. LAI paliperidone has recently become available, and dosing is initiated with two loading IM deltoid injections given at days 1 (234 mg) and 8 (156 mg), followed by injections every 4 weeks starting at day 36. This loading strategy can be omitted if transitioning patients from another LAI antipsychotic, with LAI paliperidone given in lieu of the next injection, and the every 4 weeks thereafter. The package insert provides detailed information on dose equivalency with oral paliperidone, and on handling missed injections. ODT forms are available for aripiprazole, clozapine, olanzapine, and risperidone, and are the only form for asenapine. These ODT preparations are commonly used in emergency and inpatient settings where patients may be prone to cheeking or spitting out oral tablets.

Mania is distinguished from its less severe form, hypomania, by the fact that hypomania, by definition, does not result in functional impairment or hospitalization, and is not associated with psychotic symptoms. Patients who experience periods of hypomania and major depression have bipolar II disorder, those with mania at any time, bipolar I, and those with hypomania, but less severe forms of depression, cyclothymia (American Psychiatric Association, 2000). The prevalence of bipolar I disorder is roughly 1% of the population, and the prevalence of all forms of bipolar disorder 3-5%.

Antipsychotic Agents. The chemistry and pharmacology of antipsychotic medications are addressed earlier in this chapter.

Anticonvulsants. The pharmacology and chemistry of the anticonvulsants with significant data for acute mania (valproic acid compounds, carbamazepine) and for bipolar maintenance (lamotrigine) are covered extensively in Chapter 21. These compounds are of diverse chemical classes, but share the common property of functional blockade of voltage-gated Na⁺ channels, albeit with differing binding sites. Valproate exhibits non-specific binding to voltage-gated Na⁺ channels, while carbamazepine (and its congeners) and lamotrigine have specific high affinity for the open-channel configuration of the alpha subunit (Söderpalm, 2002). These anticonvulsants have varying affinities for voltage-dependent Ca²⁺ channels, and differ in their ability to facilitate GABA-ergic (valproate) or inhibit glutamatergic neurotransmission (lamotrigine). The extent to which any of these actions is necessary for antimanic or other mood stabilizing activity is unknown, but the failure of phenytoin, gabapentin, and topiramate to be effective antimanic and mood-stabilizing medications suggests that potent blockade of voltage-gated Na⁺ channels (which gabapentin and topiramate lack) is necessary but not sufficient, since phenytoin is very active at these channels.

Mechanism of Action. Therapeutic concentrations of Li⁺ have almost no discernible psychotropic effects in individuals without psychiatric symptoms. There are numerous molecular and cellular actions of Li⁺, some of which overlap with identified properties of other mood-stabilizing agents (particularly valproate), and are discussed below. An important characteristic of Li⁺ is that, unlike Na⁺ and K⁺, Li⁺ develops a relatively small gradient across biological membranes. Although it can replace Na⁺ in supporting a single action potential in a nerve cell, it is not a substrate for the Na⁺ pump and therefore cannot maintain membrane potentials. It is uncertain whether therapeutic concentrations of Li⁺ (0.5-1.0 mEq/L) affect the transport of other monovalent or divalent cations by nerve cells.

Hypotheses for the Mechanism of Action of Lithium, and Relationship to Anticonvulsants. Plausible hypotheses focus on lithium's impact on monoamines implicated in the pathophysiology of mood disorders, and on second-messenger and other intracellular molecular mechanisms involved in signal transduction, gene regulation, and cell survival (Quiroz et al., 2004). In animal brain tissue, lithium at concentrations of 1-10 mEq/L inhibits the depolarization-provoked and Ca⁺⁺-dependent release of NE and DA, but not 5-HT, from nerve terminals. Li⁺ may even transiently enhance release of 5-HT, especially in the limbic system, but has limited effects on catecholamine-sensitive adenylyl cyclase activity or on the binding of ligands to monoamine receptors in brain tissue, although it does influence response of 5-HT autoreceptors to agonists (Lenox and Wang, 2003). Li⁺ modifies some hormonal responses mediated by adenylyl cyclase or PLC in other tissues, including the actions of vasopressin and thyroid-stimulating hormone on their peripheral target tissues (Quiroz et al., 2004). Li⁺ can inhibit the effects of receptor-blocking agents that cause supersensitivity in such systems. In part, the actions of Li⁺ may reflect its ability to interfere with the activity of both stimulatory and inhibitory G proteins (G_s and G_i) by keeping them in their inactive αβγ trimeric state (Jope, 1999).

Considered by many as relevant for lithium's therapeutic efficacy is its inhibition of inositol monophosphatase and interference with the phosphatidylinositol pathway (Figure 16–1), leading to decreased cerebral inositol concentrations (Williams et al., 2002). Phosphatidylinositol (PI) is a membrane lipid that is phosphorylated to form phosphatidylinositol bisphosphate (PIP₂). Activated phospholipase C cleaves PIP₂ into diacylglycerol and inositol 1,4,5- trisphosphate (IP₃), with the latter stimulating Ca²⁺ release from cellular stores. IP₃ is dephosphorylated to inositol monophosphate (IP) and thence to inositol by inositol monophosphatase. Within its range of therapeutic concentrations, Li⁺ uncompetitively inhibits this last step (Chapter 3), with resultant decrease in available inositol for resynthesis into PIP₂ (Shaldubina et al., 2001). The inositol depletion effect can be detected in vivo with magnetic resonance spectroscopy (Manji and Lenox, 2000). A recent genome-wide association study has implicated diacylglycerol kinase in the etiology of bipolar disorder, strengthening the association between Li⁺ actions and PI metabolism (Baum et al., 2008). Further support for the role of inositol signaling in mania rests on the finding that valproate, and valproate derivatives, decrease intracellular inositol concentrations (Shaltiel et al., 2007). Unlike Li⁺, valproate decreases inositol through inhibition of myo-inositol-1-phosphate synthase. Carbamazepine exposure in cultured sensory neurons alters the dynamic behavior of neuron growth cones, effects that are remediated through inositol supplementation, implicating inositol depletion as a mechanism underlying carbamazepine's mood stabilizing properties (Williams et al., 2002).

Stimulation of NMDA receptors results in IP₃ accumulation in primate brain, leading some to postulate that lithium's activity in the inositol pathway might also relate to glutamatergic effects (Dixon and Hokin, 1998). Subsequent experiments demonstrated the ability of acute high lithium levels to increase synaptic glutamate in mice and non-human primates, primarily through inhibition of glutamate reuptake at presynaptic nerve terminals by lowering the V_max of glutamate transport, but not glutamate binding to the transporter (Dixon and Hokin, 1998). The impact of supratherapeutic Li⁺ on synaptic glutamate may be implicated in lithium-induced neurotoxicity during overdose; conversely, chronic Li⁺ administration at therapeutic CNS concentrations results in increased glutamate uptake and increased levels of glutamate in presynaptic synaptosomes (Dixon and Hokin, 1998). This effect occurs at therapeutic Li⁺ serum concentrations, leading to speculation that the neuroprotective effects of Li⁺ seen in models of excitotoxic cell death may be mediated through modulation of synaptic glutamate availability (Bauer et al., 2003).

Li⁺ treatment also leads to consistent decreases in the functioning of protein kinases in brain tissue, including PKC, particularly isoforms α and β (Jope, 1999; Manji and Lenox, 2000; Quiroz et al., 2004). Among other proposed antimanic or mood-stabilizing agents, this effect is also shared with valproic acid (particularly for PKC) but not with carbamazepine (Manji and Lenox, 2000). Long-term treatment of rats with lithium carbonate or valproate decreases cytoplasm-to-membrane translocation of PKC and reduces PKC stimulation–induced release of 5-HT from cerebral cortical and hippocampal tissue. Excessive PKC activation can disrupt prefrontal cortical regulation of behavior, but pretreatment of monkeys and rats with lithium carbonate or valproate blocks the impairment in working memory induced by activation of PKC in a manner also seen with the PKC inhibitor chelerythrine (Yildiz et al., 2008). The impact of Li⁺ or valproate on PKC activity may secondarily alter the release of amine neurotransmitters and hormones as well as the activity of tyrosine hydroxylase (Manji and Lenox, 2000). A major substrate for cerebral PKC is the myristolated alanine-rich C-kinase substrate (MARCKS) protein, which has been implicated in synaptic and neuronal plasticity. The expression of MARCKS protein is reduced by treatment with both Li⁺ and valproate, but not by carbamazepine, antipsychotic medications, or antidepressants (Watson and Lenox, 1996). This proposed mechanism of PKC inhibition has been the basis for therapeutic trials of tamoxifen, a selective estrogen receptor modulator that is also a potent centrally active PKC inhibitor (Einat et al., 2007). In acutely manic bipolar I patients, tamoxifen has shown evidence of efficacy as adjunctive treatment, and in two double-blind, placebo-controlled monotherapy trials (Yildiz et al., 2008; Zarate et al., 2007).

Both Li⁺ and valproate treatment also inhibit the activity of glycogen synthase kinase-3β (GSK-3β) (Williams et al., 2002). Of relevance to mood disorders, GSK-3 inhibition increases hippocampal levels of β-catenin, a function implicated in mood stabilization (Kozikowski et al., 2007). GSK-3β has been found to regulate mood stabilizer-induced axonal growth and synaptic remodeling and to modulate brain-derived neurotrophic factor response. Mice with mutant forms of the GSK-3β gene demonstrate abnormalities of biological rhythms that are proposed endophenotypes for the circadian disruption seen in manic patients. In animal models, Li⁺ induces molecular and behavioral effects comparable to that seen when one GSK-3β gene locus is inactivated. Potent and selective GSK-3β inhibitors show expected activity in mouse models of mania, thus providing further stimulus for exploring this mechanism (Kozikowski et al., 2007).

Another proposed common mechanism for the actions of Li⁺ and valproate relates to reduction in arachidonic acid turnover in brain membrane phospholipids (Rao et al., 2008). Rats fed Li⁺ in amounts that achieve therapeutic CNS drug levels have reduced turnover of PI (↓ 83%) and phosphatidylcholine (↓ 73%); chronic intraperitoneal valproate achieves reductions of 34% and 36%, respectively. Li⁺ also decreased gene expression of PLA₂ and decreased levels of COX-2 and its products (Rapoport and Bosetti 2002).

5-HT release from presynaptic terminals is regulated by 5-HT_1A autoreceptors located on the cell body and 5-HT_1B receptors on the nerve terminal. In vitro electrophysiological studies suggest that Li⁺ facilitates 5-HT release. Li⁺ augements effects of antidepressants, and in animal models of depression, lithium's activity appears to be mediated through desensitizing actions at 5-HT_1B sites; Li⁺ also antagonizes mouse behaviors induced by administration of selective 5-HT_1B agonists (Shaldubina et al., 2001).

Li⁺ and valproic acid both interact with nuclear regulatory factors that affect gene expression (e.g., AP-1, AMI-1β, PEBP-2β; both agents increase expression of Bcl-2, which is associated with protection against neuronal degeneration/apoptosis (Manji and Chen, 2002).

Absorption, Distribution, and Elimination. Li⁺ is absorbed readily and almost completely from the GI tract. Complete absorption occurs in ~ 8 hours, with peak plasma concentrations occurring 2-4 hours after an oral dose (Sproule, 2002). Slow-release preparations of lithium carbonate minimize early peaks in plasma concentrations, and to some extent may decrease local GI adverse effects, but the increased trough levels may increase the risk for, and the extent of, nephrogenic diabetes insipidus (Schou et al., 1982). Li⁺ initially is distributed in the extracellular fluid, and gradually accumulates in various tissues; it does not bind appreciably to plasma proteins. The concentration gradient across plasma membranes is much smaller than those for Na⁺ and K⁺. The final volume of distribution (0.7-0.9 L/kg) approaches that of total body water and is lower than agents that are lipophilic and protein bound. Passage through the blood-brain barrier is slow, and when a steady state is achieved, the concentration of Li⁺ in the cerebrospinal fluid and in brain tissue is ~ 40-50% of the concentration in plasma. The kinetics of Li⁺ can be monitored in human brain with magnetic resonance spectroscopy (Soares et al., 2000).

Approximately 95% of a single dose of Li⁺ is eliminated in the urine. From one- to two-thirds of an acute dose is excreted during a 6-12 hour initial phase of excretion, followed by slow excretion over the next 10-14 days (Goodnick and Schorr-Cain, 1991). The elimination t_1/2 averages 20-24 hours, obviating the need for multiple daily dosing. Once-daily dosing not only improves adherence, but is also associated with decreased polyuria (Schou et al., 1982). With repeated administration, Li⁺ excretion increases during the first 5-6 days until a steady state is achieved after ~5 half-lives. When Li⁺ is stopped, there is a rapid phase of renal excretion followed by a slow 10- to 14-day phase. Since 80% of the filtered Li⁺ is reabsorbed by the proximal renal tubules, clearance of Li⁺ by the kidney is ~ 20% of that for creatinine, 15-30 mL/min (Goodnick and Schorr-Cain 1991; Hopkins and Gelenberg 2000); this rate is somewhat lower in elderly patients (10-15 mL/min) (Juurlink et al., 2004; Tueth et al., 1998). Loading with Na⁺ produces a small enhancement of Li⁺ excretion, but Na⁺ depletion promotes a clinically important degree of Li⁺ retention. Although the pharmacokinetics of Li⁺ varies considerably among subjects, the volume of distribution and clearance are relatively stable in an individual patient.

Li⁺ is completely filtered, and 80% is reabsorbed in the proximal tubules (Goodnick and Schorr-Cain, 1991). Li⁺ competes with Na⁺ for reabsorption, and Li⁺ retention can be increased by Na⁺ loss related to diuretic use, or febrile, diarrheal, or other GI illness. Heavy sweating leads to a preferential secretion of Li⁺ over Na⁺ (Jefferson et al., 1982); however, the repletion of excessive sweating using free water without electrolytes can cause hyponatremia, and promote Li⁺ retention. Thiazide diuretics deplete Na⁺ and can cause significant reductions in Li⁺ clearance that result in toxic levels. The K⁺-sparing diuretics triamterene, spironolactone, and amiloride have modest effects on the excretion of Li⁺, with concomitantly smaller increases in serum levels. Loop diuretics such as furosemide seem to have limited impact on Li⁺ levels (Juurlink et al., 2004). Renal excretion can be increased by administration of osmotic diuretics or acetazolamide, but not sufficiently for the management of acute Li⁺ intoxication. Through alteration of renal perfusion, some nonsteroidal anti-inflammatory agents can facilitate renal proximal tubular resorption of Li⁺ and thereby increase serum concentrations (Phelan et al., 2003). This interaction appears to be particularly prominent with indomethacin, but also may occur with ibuprofen, naproxen, and COX-2 inhibitors, and possibly less so with sulindac and aspirin (Phelan et al., 2003). Angiotensin-converting enzyme inhibitors, particularly the renally cleared lisinopril, also cause Li⁺ retention, with isolated reports of toxicity among stable lithium-treated patients switched from fosinopril to lisinopril (Meyer et al., 2005).

Less than 1% of ingested Li⁺ leaves the human body in the feces; 4-5% is secreted in sweat (Sproule, 2002). Li⁺ is secreted in saliva in concentrations about twice those in plasma, while its concentration in tears is about equal to that in plasma. As noted under "Pregnancy and Lactation" later in the chapter, Li⁺ is secreted in human milk, but serum levels in breast-fed infants are ~20% that of maternal levels, and are not associated with notable behavioral effects (Viguera et al., 2007).

Serum Level Monitoring and Dose. Because of the low therapeutic index for Li⁺, periodic determination of serum concentrations is crucial. Li⁺ cannot be used with adequate safety in patients who cannot be tested regularly. Concentrations considered to be effective and acceptably safe are between 0.6 and 1.5 mEq/L. The range of 1.0-1.5 mEq/L is favored for treatment of acutely manic or hypomanic patients (Sproule, 2002). Somewhat lower values (0.6-1.0 mEq/L) are considered adequate and are safer for long-term prophylaxis. Serum concentrations of Li⁺ have been found to follow a clear dose-effect relationship between 0.4 and 1.0 mEq/L, but with a corresponding dose-dependent rise in polyuria and tremor as indices of adverse effects. Nonetheless, patients who maintain trough levels of 0.8-1.0 mEq/L experience decreased relapse risk compared to those maintained at lower serum concentrations. There are patients who may do well with serum levels of 0.5-0.8 mEq/L, but there are no current clinical or biological predictors to permit a priori identification of these individuals. Individualization of serum levels is often necessary to obtain a favorable risk-benefit relationship.

The concentration of Li⁺ in blood usually is measured at a trough of the daily oscillations that result from repetitive administration (i.e., from samples obtained 10-12 hours after the last oral dose of the day). Peaks can be two or three times higher at a steady state. When the peaks are reached, intoxication may result, even when concentrations in morning samples of plasma at the daily nadir are in the acceptable range of 0.6-1 mEq/L. Single daily doses generate relatively large oscillations of plasma Li⁺ concentration but lower mean trough levels than with multiple daily dosing, and are associated with a reduction in the extent and risk for polyuria (Schou et al., 1982); moreover, single nightly dosing means that peak serum levels occur during sleep, so complaints regarding CNS adverse effects are minimized. While relatively uncommon, GI complaints are one compelling reason for multiple daily dosing or using delayed release Li⁺ preparations, bearing in mind the increased polyuria risk from these strategies.

Li⁺ is effective in acute mania, but is rarely employed as a sole treatment for reasons noted above, and because 5-7 days are required for clinical effect. A 600-mg loading dose of Li⁺ can be given to hasten the time to steady state, and can also be used to predict dosage requirements based on the 24-hour serum Li⁺ result, with a very high correlation coefficient (r = 0.972) (Cooper et al., 1973). Acutely manic patients may require higher dosages to achieve therapeutic serum levels, and downward adjustment may be necessary once the patient is euthymic. When adherence with oral capsules or tablets is an issue, the liquid Li⁺ citrate can be used. Each 5 mL of lithium citrate syrup provides 8.12 mEq of Li⁺, equivalent to 300 mg of lithium carbonate.

The anticonvulsant sodium valproate provides more rapid antimanic effects than Li⁺, with therapeutic benefit seen within 3-5 days. The most common form of valproate in use is divalproex sodium, preferred over valproic acid due to lower incidence of GI and other adverse effects. Divalproex is initiated at 25 mg/kg once daily and titrated to effect or the desired serum concentration. Serum concentrations of 90-120 μg/mL show the best response in clinical studies (Bowden et al., 2006). With immediate release forms of valproic acid and divalproex sodium, 12-hour troughs are used to guide treatment. With the extended-release divalproex preparation, patients respond best when the 24-hour trough levels are in the high therapeutic range (Bowden et al., 2006).

Carbamazepine is effective for acute mania. Immediate release forms of carbamazepine cannot be loaded or rapidly titrated over 24 hours as with valproate due to the development of neurological adverse effects such as dizziness or ataxia, even within the therapeutic range (6-12 μg/mL) (Post et al., 2007); the extended-release form of carbamazepin was FDA-approved for acute mania in 2005. The extended-release form is better tolerated compared to older preparations, and effective as monotherapy with once-daily dosing (Weisler et al., 2006). Carbamazepine response rates are lower than those for valproate compounds or Li⁺, with mean rates of 45-60% cited in the literature (Post et al., 2007). Nevertheless, certain individuals respond to carbamazepine after failing Li⁺ and valproate. Initial doses are 400 mg/day, with the larger dose given at bedtime due to the sedating properties of carbamazepine. Titration proceeds by 200-mg increments every 24-48 hours based on clinical response and serum trough levels. Due to increased risk of Stevens-Johnson syndrome in Asian individuals (especially those from China, Hong Kong, Malaysia, and the Philippines where HLA-B*1502 prevalence is > 15%), HLA testing must be performed prior to treatment in populations at risk.

Lamotrigine has no role in acute mania due to the slow, extended titration necessary to minimize risk of Stevens-Johnson syndrome (Hahn et al., 2004).

The overriding concern guiding bipolar treatment is the high recurrence rate. Individuals who experience mania have an 80-90% lifetime risk of subsequent manic episodes. As with schizophrenia, lack of insight, poor psychosocial support, and substance abuse all interfere with treatment adherence. While the anticonvulsants lamotrigine, carbamazepine, and divalproex have data supporting their use in bipolar prophylaxis, only concern has consistently been shown to reduce the risk of suicide compared to non-lithium users (Cipriani et al., 2005; Tondo et al., 2001), and compared to bipolar patients on valproate acid derivatives (Goodwin et al., 2003). Early case series and retrospective clinical literature indicated preferential benefit of valproate over concern in rapid-cycling patients, but a recent large trial failed to substantiate this effect, and found no significant differences in time to relapse between the two agents (Calabrese et al., 2005b). Based largely on subanalyses of acute mania trials, there are also claims that valproate may be superior to concern for patients experiencing mixed episodes, but the lack of prospective data from randomized trials also raises questions about the validity of these findings. Lamotrigine was effective in two large, 18-month long maintenance trials for bipolar patients whose most recent mood episode was manic or depressed, with greater effect on depressive relapse (Bowden et al., 2003; Calabrese et al., 2003). The ability to provide prophylaxis for future depressive episodes combined with data in acute bipolar depression (Hahn et al., 2004) has made lamotrigine a useful choice for bipolar treatment, given that bipolar I and II patients spend large amounts of time in depressive phases (32% and 50%, respectively) (Judd et al., 2003).

Bipolar disorder is a lifetime illness with high recurrence rates. Individuals who experience an episode of mania should be educated about the probable need for ongoing treatment. Stopping mood stabilizer therapy can be considered in patients who have experienced only one lifetime manic episode, particularly when there may have been a pharmacological precipitant (e.g., substance or antidepressant use), and who have been euthymic for extended periods. For bipolar II patients, the impact of hypomania is relatively limited, so the decision to recommend prolonged maintenance treatment with a mood stabilizer is based on clinical response and risk:benefit ratio. Discontinuation of maintenance Li⁺ treatment in bipolar I patients carries a high risk of early recurrence and of suicidal behavior over a period of several months, even if the treatment had been successful for several years. Recurrence is much more rapid than is predicted by the natural history of untreated bipolar disorder, in which cycle lengths average ~1 year. This risk may be moderated by slow, gradual removal of Li⁺, while rapid discontinuation should be avoided unless dictated by medical emergencies.

Other Uses of Lithium. Li⁺ has been shown to be effective as adjunct therapy in treatment-resistant major depression (Bschor et al., 2002). Clinical data also support Li⁺ use as monotherapy for unipolar depression. Meta-analyses indicate that lithium's benefit on suicide reduction extends to unipolar mood disorder patients (Cipriani et al, 2005; Tondo et al., 2001). While maintenance Li⁺ levels of 0.6-1.0 mEq/L are used for bipolar prophylaxis, a lower range (0.4-0.8 mEq/L) is recommended for antidepressant augmentation.

Based on its neuroprotective properties, there has been consideration of Li⁺ treatment for conditions associated with excitotoxic and apoptotic cell death, such as stroke and spinal cord injury, as well as in neurodegenerative disorders, including dementia of the Alzheimer type, stroke, Parkinson's disease, Huntington's disease, amyotrophic lateral sclerosis, progressive supranuclear palsy, and spinocerebellar ataxia type I (Aghdam and Barger 2007; Bauer et al., 2003; Chuang, 2004).

Adverse Effects

CNS Effects. The most common CNS effect of Li⁺ in the therapeutic dose range is fine postural hand tremor, indistinguishable from essential tremor. Severity and risk for tremor are dose-dependent, with incidence ranging from 15-70%. Some tolerance may develop over time, but tremor may be sufficiently bothersome to require intervention. In addition to the avoidance of caffeine and other agents that increase tremor amplitude, therapeutic options include dose reduction (bearing in mind the increased relapse risk with lower serum Li⁺ levels), or in those without contraindications for β adrenergic blockade, use of propranolol, starting at 20 mg/day in divided doses, and gradually increased until the desired clinical effect is achieved (typically at doses ≤ 160 mg/day). Valproate treatment has a similar problem, and the approach to valproate-induced tremor is identical. At peak serum (and CNS) levels, some individuals may complain of incoordination, ataxia, or slurred speech, all of which can be avoided by dosing Li⁺ at bedtime. Patients may also complain of mental fatigue or cognitive dulling at higher serum Li⁺ levels, but this should be carefully assessed to determine whether this reflects a true side effect or a desire to regain the mental high from hypomania.

Li⁺ routinely causes EEG changes characterized by diffuse slowing, widened frequency spectrum, and potentiation with disorganization of background rhythm. Seizures have been reported in non-epileptic patients with therapeutic plasma concentrations of Li⁺. Li⁺ treatment has also been associated with increased risk of post-ECT confusion, and is generally tapered off prior to a course of ECT.

Li⁺ (and valproate) treatment results in significant weight gain, a problem that is magnified by concurrent use of antipsychotic drugs. Mean weight gain at one year in prospective Li⁺ trials ranges from −1 kg to + 4 kg, but the proportion of individuals who gain > 5% of baseline weight is 13-62% (Keck and McElroy, 2003). Although the mechanism is unclear, central appetite stimulation at hypothalamic sites is the most plausible explanation (Keck and McElroy, 2003).

Renal Effects. Long-term renal effects of Li⁺ treatment are an important concern. The kidneys' ability to concentrate urine decreases during Li⁺ therapy, and ~60% of individuals exposed to Li⁺ experience some form of polyuria and compensatory polydipsia. The mechanism of polyuria is unclear, but the result is decreased vasopressin stimulation of renal reabsorption of water, and the clinical picture of nephrogenic diabetes insipidus. Mean 24-hour urinary volumes of 3 L/day are common among long-term Li⁺ users, but Li⁺ discontinuation or a switch to single daily dosing may reverse the impact on renal concentrating ability in patients with < 5 years of Li⁺ exposure (Kusalic and Engelsmann, 1996). Patients exposed to multiple daily dosing or sustained- release Li⁺ preparations are at greatest risk for renal effects. Renal function should be monitored with biyearly serum blood urea nitrogen and creatinine levels, calculation of estimated GFR using standard formulas (Morriss and Benjamin, 2008), and annual measurement of 24-hour urinary volume. Reassessment of Li⁺ treatment, more frequent renal function tests, and possible nephrology consultation should be considered when estimated GFR is < 60 mL/min on several periodic measurements, daily urinary volume exceeds 4 L, or serum creatinine continues to rise on three separate occasions (Morriss and Benjamin, 2008).

Thyroid and Endocrine Effects. A small number of patients on Li⁺ develop a benign, diffuse, nontender thyroid enlargement suggestive of compromised thyroid function, although many of these patients will have normal thyroid function. Measurable effects of Li⁺ on thyroid indices are seen in a fraction of patients: 7-10% develop overt hypothyroidism (Jefferson, 2000), and 23% have subclinical disease, with women at 3-9 times greater risk (Kleiner et al., 1999). There appears to be no plateau in the incidence of hypothyroidism in studies covering up to 10 years' exposure; thus, ongoing monitoring of TSH and free T₄ is recommended throughout the course of Li⁺ treatment. The development of hypothyroidism is easily treated through exogenous replacement, and is not a reason to discontinue Li⁺ therapy. Rare reports of hyperthyroidism during Li⁺ treatment also exist (Jefferson, 2000).

EKG Effects. The prolonged use of Li⁺ causes a benign and reversible T-wave flattening in ~20% of patients and the appearance of U waves, effects unrelated to depletion of Na⁺ or K⁺. At therapeutic concentrations there are rare reports of Li⁺-induced effects on cardiac conduction and pacemaker automaticity, effects that become pronounced during overdose and lead to sinus bradycardia, atrioventricular blocks, and possible cardiovascular compromise. Routine EKG monitoring is not recommended in younger patients, but may be considered in older patients, particularly those with a history of arrhythmia or coronary heart disease.

Skin Effects. Allergic reactions such as dermatitis, folliculitis, and vasculitis can occur with Li⁺ administration. Worsening of acne vulgaris, psoriasis, and other dermatological conditions is a common problem that is usually treatable by topical measures, but in a small number may improve only upon discontinuation of Li⁺. Some patients on Li⁺ (and valproate) may experience alopecia; daily supplementation with a multivitamin containing at least 100 μg of selenium and 15 mg of zinc may be beneficial.

Pregnancy and Lactation. Li⁺ is classified as risk category D (see Appendix I). The use of Li⁺ in early pregnancy may be associated with an increase in the incidence of cardiovascular anomalies of the newborn, especially Ebstein's malformation (Freeman and Freeman, 2006). The basal risk of Ebstein's anomaly of ~1 per 20,000 live births may rise several-fold with first-trimester Li⁺ exposure, but probably not above 1 per 2500. Moreover, the defect typically is detectable in utero by ultrasonography and often is surgically correctable after birth. Although the anticonvulsants valproic acid and carbamazepine are also pregnancy risk category D, these agents are associated with irreversible neural tube defects. In balancing the risk vs. benefit of using Li⁺ in pregnancy, it is important to evaluate the risk of inadequate prophylaxis for the bipolar disorder patient, and subsequent risk that mania poses for the patient and fetus. In patients who choose to forego medication exposure during the first trimester, potentially safer treatments for acute mania include antipsychotic drugs or ECT (Dodd and Berk, 2006; Yonkers et al., 2004).

In pregnancy, maternal polyuria may be exacerbated by Li⁺. Concomitant use of Li⁺ with medications that waste Na⁺ or a low-Na⁺ diet during pregnancy can contribute to maternal and neonatal Li⁺ intoxication. Li⁺ freely crosses the placenta, and fetal or neonatal Li⁺ toxicity may develop when maternal blood levels are within the therapeutic range. Fetal Li⁺ exposure is associated with neonatal goiter, CNS depression, hypotonia ("floppy baby" syndrome), and cardiac murmur. Most recommend withholding lithium therapy for 24-48 hours before delivery, and this is considered standard practice to avoid the potentially toxic increases in maternal and fetal serum Li⁺ levels associated with postpartum diuresis. This brief period of discontinuation results in a mean decrease of 0.28 mEq/L in maternal Li⁺ concentrations (Newport et al., 2005). The physical and CNS sequelae of late-term neonatal Li⁺ exposure are reversible once Li⁺ exposure has ceased, and no long-term neurobehavioral consequences are observed (Viguera et al., 2007).

Other Effects. A benign, sustained increase in circulating polymorphonuclear leukocytes (12,000-15,000 cells/mm³) occurs during the chronic use of Li⁺ and is reversed within a week after termination of treatment. Some patients complain of a metallic taste, making food less palatable. Myasthenia gravis may worsen during treatment with Li⁺.

Acute Toxicity and Overdose. The occurrence of toxicity is related to the serum concentration of Li⁺ and its rate of rise following administration. Acute intoxication is characterized by vomiting, profuse diarrhea, coarse tremor, ataxia, coma, and convulsions. Symptoms of milder toxicity are most likely to occur at the absorptive peak of Li⁺ and include nausea, vomiting, abdominal pain, diarrhea, sedation, and fine tremor. The more serious effects involve the nervous system and include mental confusion, hyperreflexia, gross tremor, dysarthria, seizures, and cranial nerve and focal neurological signs, progressing to coma and death. Sometimes both cognitive and motor neurological damage may be irreversible, with persistent cerebellar tremor being the most common (El-Mallakh, 1986). Other toxic effects are cardiac arrhythmias, hypotension, and albuminuria.

Treatment of Lithium Intoxication. There is no specific antidote for Li⁺ intoxication, and treatment is supportive, including intubation if indicated, and continuous cardiac monitoring. Levels > 1.5 mEq/L are considered toxic, but inpatient medical admission is usually not indicated (in the absence of symptoms) until levels exceed 2 mEq/L. Care must be taken to assure that the patient is not Na⁺ and water depleted. Dialysis is the most effective means of removing Li⁺ and is necessary in severe poisonings, that is, in patients exhibiting symptoms of toxicity or patients with serum Li⁺ concentrations ≥ 3 mEq/L in acute overdoses. A review of 213 case reports of acute Li⁺ toxicity between 1948 and 1984, found that complete recovery occurred with an average maximal level of 2.5 mEq/L, permanent neurological symptoms with mean levels of 3.2 mEq/L, and death with mean maximal levels of 4.2 mEq/L (El-Mallakh, 1986). The most common neurological sequelae are due to cerebellar damage, and manifest clinically as ataxia, tremor, dysarthria, and dysmetria (Mangano et al., 1997).

Use in Pediatric and Geriatric Populations

Pediatric Use. The anticonvulsants all have pediatric indications for epilepsy, but only Li⁺ has FDA approval for child/adolescent bipolar disorder for ages ≥ 12 years (Tueth et al., 1998). Recently, aripiprazole and risperidone were FDA-approved for acute mania in children and adolescents ages 10-17. Children and adolescents have higher volumes of body water and higher GFR than adults. The resulting shorter t_1/2 of Li⁺ demands dosing increases on a mg/kg basis, and multiple daily dosing is often required. In children ages 6-12 years, a dose of 30 mg/kg/day given in three divided doses will produce a Li⁺ concentration of 0.6-1.2 mEq/L in 5 days (Danielyan and Kowatch, 2005), although ongoing dosing is always guided by serum levels and clinical response. Use in children under 12 represents an off-label use for Li⁺ and care givers should be alert to evidence of the development of toxicity. As with adults, ongoing monitoring of renal and thyroid function is important, along with clinical inquiry into extent of polyuria.

A limited number of controlled studies suggest that valproate has efficacy comparable to that of Li⁺ for mania in children or adolescents (Danielyan and Kowatch, 2005). As with Li⁺, weight gain and tremor can be problematic; moreover, there are reports of hyperammonemia in children with urea-cycle disorders. Ongoing monitoring of platelets and liver function tests, in addition to serum drug levels, is recommended. There have been no controlled studies of carbamazepine for the treatment of children and adolescents with mania, although there is substantial safety data for epilepsy indications in this age group.

Geriatric Use. The majority of older patients on Li⁺ therapy are those maintained for years on the medication. Less than 10% of individuals with bipolar disorder experience their first manic episode at age 50 or above. Elderly patients frequently take numerous medications for other illnesses and the potential for contraindications or drug-drug interactions is substantial. For patients naïve to Li⁺, these issues can be addressed relatively easily prior to commencement of Li⁺ treatment. The more difficult clinical decision revolves around switching treatment in stable patients who have taken Li⁺ for years or decades with excellent clinical response. In addition, age-related reductions in total body water and creatinine clearance reduce the safety margin for Li⁺ treatment in older patients. Targeting lower maintenance serum levels (0.6-0.8 mEq/L) may reduce the risk of toxicity. As GFR drops below 60 mL/min, strong consideration must be given to a search for alternative agents, despite lithium's therapeutic advantages (Morriss and Benjamin, 2008). Li⁺ toxicity occurs more frequently in elderly patients, in part as the result of concurrent use of loop diuretics and angiotensin-converting enzyme inhibitors (Juurlink et al., 2004). Anticonvulsants, especially extended-release divalproex, are a reasonable alternative to Li⁺. Elderly patients who are drug-naïve may be more sensitive to CNS adverse effects of all types of medications used for acute mania, especially parkinsonism and tardive dyskinesia from D₂ antagonism, confusion from antipsychotic medications with antimuscarinic properties, and ataxia or sedation from Li⁺ or anticonvulsants.

Novel Treatments for Psychosis and Mania

There are several molecular targets for antipsychotic drug development, including glutamate, serotonergic, and cholinergic receptor systems. The glutamate hypofunction hypothesis for schizophrenia has opened the door to exploration of metabotropic glutamate receptor agonists, and agonists at co-transmitter sites. The glutamate receptor family is divided into the excitatory ionotropic receptors, broadly distributed in the CNS, and metabotropic glutamate receptors that exert modulatory functions (Chapter 14). The former are not viable targets for drug activity due to the global impact on CNS function that may result. Metabotropic glutamate receptors show significant promise, with agents synthesized based on the distribution of the three receptor groups (Krivoy et al., 2008).

The mGlu 2/3 receptor agonist LY2140023 showed efficacy in a 4-week phase 2 schizophrenia monotherapy trial (Patil et al., 2007), and while subsequent data were problematic, similar compounds are still in development.

Glycine is a co-transmitter at glutamate receptors whose synaptic action is terminated by presynaptic reuptake transporters. There are two human forms of the glycine reuptake pump: GlyT1, which is localized to cortical areas; and GlyT2, which is concentrated in spinal cord. The GlyT1 inhibitor SSR504734 was as effective as haloperidol in blocking PCP-induced CNS metabolic changes in rats. Lurasidone is a novel 5-HT/DA antagonist that filed for FDA approval in late 2009. Lurasidone possesses potent activity at 5-HT₇ receptor sites, actions that, based on preclinical and early clinical studies, may be associated with cognitive benefits (Meyer et al., 2009). Muscarinic receptor modulators at M₁ and M₄ sites have passed several preclinical antipsychotic tests (Chan et al., 2008; Janowsky and Davis, 2009), and xanomeline, an M₁/M₄ agonist, has shown antipsychotic and procognitive affects in a 4-week, double-blind placebo-controlled schizophrenia trial (Shekhar et al., 2008). Certain targets (e.g., α7 and α4β2 nicotinic receptors) are being exploited solely for the purposes of cognitive enhancement in schizophrenia, and may not confer antipsychotic activity (Buchanan et al., 2007).

The expanded role for atypical antipsychotics in bipolar treatment is primarily in acute mania, although quetiapine possesses unique effectiveness for bipolar depression based on antidepressant properties conferred by the metabolite norquetiapine. Improved understanding of the biological basis for Li⁺ and anticonvulsant actions in mania has generated several possible targets. No agents acting on IP₃ have been developed, but several inhibitors of GSK-3β that are 3-benzofuranyl-4-indolylmaleimide compounds have been identified, with one showing activity in a murine mania model (Kozikowski et al., 2007). PKC inhibition has emerged as the most viable novel candidate for antimanic activity, based on several positive tamoxifen trials in acute mania (Einat et al., 2007; Yildiz et al., 2008; Zarate et al., 2007). There is still no medication that possesses lithium's unique broad spectrum of pharmacological activities for bipolar disorder, and particular aspects of lithium's actions that are addressed in a limited fashion by current medications (e.g., lamotrigine and quetiapine for bipolar depression), or not at all (e.g., neuroprotection), remain important targets for novel treatments (Quiroz et al., 2004).