The Evolution of Drug Development in Schizophrenia

Abstract

Schizophrenia is a disease syndrome with major public health implications. The primary advance in pharmacotherapeutics was in 1952 with the introduction of antipsychotic medications (ie, chlorpromazine, dopamine D2 antagonism). Barriers to progress have been substantial, but many will be subject to rapid change based on current knowledge. There are attractive psychopathology indications for drug discovery (eg, impaired cognition and negative symptoms), and drugs with efficacy in these domains may have application across a number of disease classes. These pathologies are observed prior to psychosis raising the possibility of very early intervention and secondary prevention. Success in drug discovery for cognition and negative symptom pathologies may bring forth issues in ethics as the potential for enhancing normal function is explored.

INTRODUCTION

Schizophrenia is a clinical syndrome, perhaps comprising several disease entities. Cases are observed worldwide with some variation in incidence and life-long prevalence, but afflicting 0.5-0.8 percent of the world’s population (Saha et al, 2005). The onset of some aspects of the disease may be observed from birth onward, but psychotic symptoms generally become manifest in late adolescence and early adulthood in males, with an extended onset period in females. Psychotic symptoms such as hallucinations, delusions, and disorganization of thought can impair function, and are stigmatizing. These symptoms lead to diagnosis, but are usually preceded by trait dysfunctions in cognition, affect, and motivation. These aspects of schizophrenia account for substantial decrements in social and occupational functioning and appear to be primary determinants of long-term morbidity (Matza et al, 2006; Velligan et al, 2006; Green et al, 2004; Harvey et al, 2006). Psychotic symptoms may persist from disease onset, but the general trend is a pattern of remission/exacerbation or partial remission/exacerbation. The combination of early onset, poor function, and stigmatizing symptoms lead to failure in many human pursuits. Patients with schizophrenia are over-represented among the non-married, childless, unemployed, underemployed, and low academic achievers. Homelessness, extensive hospitalization, joblessness, time in jail, disability support, supervised living arrangements, dependence on family, excess tobacco, and substance abuse, social isolation, poor health, victims of crime, and early death are all associated with schizophrenia. The cost of schizophrenia in human and financial terms is great to both society and to families (Murray and Lopez, 1996; Wyatt et al, 1995; Mauskopf et al, 2002). Schizophrenia is a leading public health challenge (Rupp and Keith, 1993; Murray and Lopez, 1996; Lopez et al, 2006). Critical therapeutic advances have been associated with humane care, specialized forms of psychosocial treatments, electroconvulsive therapy, community-based care, and rehabilitation. Although some patients do well off medication (Bola, 2006) pharmacotherapy is generally considered the essential component for reducing psychotic symptoms and relapse rates. Among the first pharmacological agents used in schizophrenia was the antihypertensive agent, reserpine. This drug acts to reduce synaptic dopamine release and its beneficial effects in schizophrenia preceded the discovery of the antipsychotic properties of chlorpromazine (for review see Seeman, 2002; Kapur and Mamo, 2003). Since chlorpromazine was introduced in 1952, about 50 additional antipsychotic drugs have been developed for the treatment of schizophrenia. Each of these drugs exerts therapeutic action at the dopamine D2 receptor and all but aripiprazole, are antagonists at the D2 receptor (Kapur and Mamo, 2003; Davies et al, 2004). Aripiprazole, on the other hand, is a partial agonist at the D2 receptor (for review see Tamminga and Carlsson, 2002). Clozapine, the only antipsychotic approved with a superiority claim for treatment-resistant and refractory patients, is also a D2 antagonist (Conley, 1998; Wahlbeck et al, 1999). Clozapine is distinguished from other antipsychotic drugs in several ways including a broad profile of receptor affinities and ‘rapid-on, rapid-off’ kinetics at the D2 receptor (Seeman, 2002). However, the basis for clozapine superiority is not known, and has not been replicated by other new generation drugs.

Despite the ability of dopamine antagonists to reduce psychosis and delay symptom exacerbations, the long-term outcome of schizophrenia has remained poor. During the 1950-1960s, two major changes occurred in the way schizophrenia was treated. The first was a shift in treatment focus from long-term custodial to community-based care. The second change was the introduction of efficacious pharmacotherapy. However, Hegarty et al (1994) found little evidence that these two major revolutions altered the outcome of schizophrenia during the twentieth century. Changing methodologies during this period make before and after antipsychotic drug therapy comparisons difficult, but schizophrenia remains a chronic illness with substantial functional impairments for most cases. A probable explanation will be found below when considering the diverse nature of schizophrenia pathology, and the association of functional outcomes with pathological domains that are not responsive to antipsychotic medication.

The majority of currently approved pharmacological agents for the treatment of schizophrenia target psychotic symptoms as their primary effects. In this critical aspect, the drugs are extensively similar in efficacy and effectiveness. Only clozapine has been documented to be modestly more effective in treatment resistant/refractory cases (Kane et al, 1988; Lewis et al, 2006; McEvoy et al, 2006; Conley, 1998; Wahlbeck et al, 1999). The first generation antipsychotic drugs, often termed neuroleptics, have robust adverse effects. Dysphoria, dystonia, akathisia, dyskinesia, and Parkinsonian motor symptoms are the most notable of these effects. First generation antipsychotic drugs also may increase or prolong the depressive/demoralization aspects of illness course, impair learning, and slow information processing, and akathisia may increase hostility, aggression, and suicidality (Conley and Kelly, 2002; Awad and Voruganti, 2004; Weickert and Goldberg, 2005). These adverse effects have been exaggerated in clinical practice where excess dosing and under utilization of prophylactic antiparkinsonian drugs is common. Second generation antipsychotic drugs are similar to first generation drugs in their profile of therapeutic efficacy for core schizophrenia pathology (see cochrane reports http://www.mrw.interscience.wiley.com/cochrane/). Many of the adverse effects of first-generation antipsychotic drugs are diminished or absent in the second-generation antipsychotic medications, and this may account for advantages observed in some comparison studies. These advantages observed in some, but not all, studies include improvement in measures of negative symptoms, cognitive test performance, depression, adherence, time to relapse, aggression, and suicide. A few studies using low doses of first-generation antipsychotic drugs, and recent head-to-head comparisons with public sponsorship in the United States and United Kingdom fail to support the superiority of second generation drugs (Jones et al, 2006, Lieberman et al, 2003a, b, 2005; Geddes et al, 2005; Schooler et al, 2005; Lieberman, 2007). Some second-generation antipsychotic drugs cause very substantial adverse effects. Among these metabolic syndrome, already a concern in persons with schizophrenia based on life style risk factors, is a major concern with drugs that cause increased body mass index, hyperlipidemia, reduced insulin sensitivity, and are associated with an increased incidence of diabetes. Reduction in life span is great, and expected to worsen with increased exposure to pharmacological adverse effects (Hennekens et al, 2005; Auquier et al, 2006; Seeman, 2007; Colton and Manderscheid, 2006; Newcomer and Hennekens, 2007). In most respects, drug development for schizophrenia has not progressed appreciably since the introduction of chlorpromazine, a point which recently found emphasis in a large first episode clinical trial in Beijing comparing the original antipsychotic drug, chlorpromazine, with the only second generation drug with documented superiority, clozapine. This trial reported little therapeutic difference between these two drugs (Lieberman et al, 2003a). While not denying the clozapine superiority in treatment-resistant cases, these data reinforce the view that dopamine D2 receptor antagonists share a mechanism of action that produces similar efficacy.

The clinical trials data to date justify the following conclusions.

1. Discovery platforms for schizophrenia have repeatedly produced drugs with the same or similar mechanism of action. In spite of fifty years of development, virtually no new drugs have achieved superior efficacy for psychosis. The traditional discovery pathway has not produced drugs that address the cognitive impairments or negative symptom pathology.

2. Quality of life and functional outcomes are not adequately addressed by antipsychotic drug development.

Impaired cognition and negative symptom pathology remain unmet treatment needs, and substantially account for long-term morbidity, and poor functional outcomes associated with this disease (Buchanan et al, 2005; Green et al, 2004; Kirkpatrick et al, 2000, 2006; Matza et al, 2006).

History of the Concept and A Paradigm Shift: Relevance to Drug Discovery

In the late nineteenth century medical progress was facilitated by the power of the disease entity model, which identified similarities across patients in onset, manifestations, and course of illness. Distinctive patterns had been elusive among the insane, and the substantial heterogeneity between patients was sometimes resolved by identifying very narrow and specific proposed disease entities. Syphilitic insanity was common in that era, and manifestations included psychosis. When the cause was determined and these cases were recognized and separated from other forms of madness, it was possible for Kraepelin to distinguish the illness patterns on which he proposed the two major forms of chronic psychotic illness (Kraepelin, 1919). The manic-depressive psychoses were described and separated from dementia praecox. The latter combined paranoia, hebephrenia, and catatonia disease classes based on similarities in age and type of onset, symptomatic manifestations, and course of illness. Parenthetically, discovery of the spirochete as causative of a common and severe form of insanity and the subsequent treatment and prevention is one of the remarkable therapeutic triumphs in medicine.

Bleuler (1950), working in the era of associative psychology, identified dissociative pathology as fundamental to dementia praecox, and coined the term schizophrenia to denote mental splitting within thought, and between thought, affect, and behavior. Introduced in 1911, the concept of schizophrenia as a single disease entity with a unifying pathology has been the dominant paradigm for almost 100 years. The influence of this paradigm can be seen in most studies in that the design addresses schizophrenia as a class and views heterogeneity of manifestations as representing the same latent structure. Kraepelin described two maladies within dementia praecox: the dissociative pathology of Bleuler and the weakening of the well-springs of volition (today’s negative symptom concept). But he viewed these as two different phenomena arising from the same disease. The key question is whether schizophrenia is a disease or a syndrome (Carpenter, 2006). Consider dementia research and the importance of identifying specific disease entities rather than investigating a heterogeneous dementia syndrome.

The title of Bleuler’s text, Dementia Praecox or the Group of Schizophrenias, suggests heterogeneity, but the traditional subtypes such as paranoid schizophrenia, hebephrenic schizophrenia and catatonic schizophrenia were not validated as separate disease entities. Rather, by the midtwentieth century, influential proposals from Schneider (1959) and Langfeldt (1937, 1939) focused on symptoms with critical diagnostic importance proposing to distinguish true schizophrenia from pseudoschizophrenia and other forms of psychotic illness. During this time much of American psychiatry neglected classification in favor of psychodynamic formulations. The introduction of efficacious drugs for depression, psychosis, and mania gave impetus to classification. And the growing psychiatric research community needed classification criteria that were specific, valid, and reliable. Studies documenting a broader definition of schizophrenia in the US compared to the UK gave urgency to restructuring classification. The resulting DSM-III, influenced by the European concept of nuclear schizophrenia and the primacy of Schneider’s First Rank Symptoms, resulted in an unintended but dramatic change in the concept of schizophrenia (Tamminga and Carpenter, 1982; Eysenck et al, 1983). Symptoms such as hallucinations and delusions, considered secondary by Bleuler, became the foremost defining criteria, and special forms such as voices commenting on behavior or discussing the patient in third person pronouns became critical to the diagnosis of schizophrenia. They also became, if combined with duration and dysfunction criteria, sufficient to define a case as schizophrenia. Consequently, the schizophrenia concept was redefined as psychosis, with emphasis on reality distortion pathology.

During this time, clinical trials of antipsychotic drugs used primary endpoints that were weighted toward psychotic symptoms as defining therapeutic response. Hence, time to discharge, time to relapse, and rating scales using total scores or psychosis scores were used to assess change in drug treatment trials.

Taken together, the effect of these trends was to treat schizophrenia as a unitary disease entity, to equate psychosis (at least special psychotic phenomena) with schizophrenia, to measure treatment effects by measuring effects on psychosis, and to generally regard antipsychotic drugs as antischizophrenia drugs. Drug development models favored compounds based on their ability to block or reverse ‘hyperdopaminergic’ models. The result is 50 years of dopamine antagonists with a partial agonist as the only variation on the theme (Adams et al, 2005).

A Paradigm Shift to Facilitate Drug Discovery

Heterogeneity in the manifestations and course of schizophrenia has long been observed. Within the disease entity paradigm, various aspects of pathology were viewed as emerging from the same latent structure. A unifying pathophysiology was expected, with neuroanatomic locations of the pathology perhaps determining symptom expression. The syphilitic insanities provided a compelling model, and knowledge of brain-behavior relations could account for different symptom patterns between cases. This view was challenged by work in the early 1970s suggesting an almost orthogonal relationship between symptom complexes and a lack of predictive relationships between symptom domains (Strauss et al, 1974). Further, pathologic manifestations within a domain were closely linked across time so that negative symptoms predicted future negative symptoms, past social functioning predicted future functioning, etc (Carpenter et al, 1978; Strauss and Carpenter, 1977). On the basis of these data, schizophrenia was reconceptualized as a tripartite construct with positive psychotic symptoms, negative symptoms, and pathology in interpersonal relating, constituting separate domains. It was envisioned that each domain would be a separate target for etiologic and treatment discovery (Strauss et al, 1974).

This tripartite domains model, proposed in 1974, has been modified in important ways that are highly relevant to new drug discovery. First, positive psychotic symptoms involve two domains: reality distortion (ie, hallucinations and delusions) and disorganization of thought and behavior. Although both are responsive to antipsychotic drugs, they separate repeatedly in factor analytic studies (Buchanan and Carpenter, 1994) and may have different associated biologic features. Second, impaired cognition as measured in various psychological and neuropsychological test procedures, is now viewed as central to the early manifestations of schizophrenia and critically related to functional outcomes. The impaired cognitive pathology is not significantly associated with the symptom domains (Gold and Harvey, 1993; Berman et al, 1997; Harvey et al, 2006; Cohen et al, 2007) but is comprised of seven independent areas of impairment, which were identified by the academic/pharmaceutical/governmental MATRICS initiative (Table 1). The specificity of these concepts should facilitate developing animal and human experimental models. Third, a number of endophenotypes have been proposed (Table 2), often based on physiologic measures of information processing. This is especially critical in the post-genomic era since genotype/endophenotype relations are likely to be far more robust than genotype/schizophrenia relations (Braff and Light, 2005; Braff et al, 2007; Thaker, 2007). Here too, facilitation of animal (Table 3) and human experimental models useful in drug development are promising so long as phenotypes can be translated from one species to the other. Tables 2 and 3 provide evidence that human phenotypes can be translated into non-human experimental endpoints and that these endpoints, ultimately, can be used to inform about the human disease. Fourth, the negative symptom construct derived from Kraepelin’s avolitional malady, is being revised to provide a more specific definition. Clinical ratings in treatment trials during the past 20 years have failed to differentiate primary (to the disease) from secondary (eg, paranoid social withdrawal, drug-induced anergia or akinesia, or depressive anhedonia) negative symptoms. Here too, animal and human models will be facilitated by currently evolving constructs and measurement. A consensus view, developed in a National Institute of Mental Health (NIMH) workshop, proposes anhedonia, blunted affect, alogia, avolition, and asociality as components of this construct (Kirkpatrick et al, 2006).

Table 1.

MATRICS: provisional consensus cognitive battery

Parameter evaluated	Test to assess parameter in patients
Speed of processing	Category fluency
	Brief Assessment of Cognition in Schizophrenia (BACS)—symbol-coding
	Trail making A
Attention/vigilance	Continuous Performance Test—identical pairs (CPT-IP)
Working memory	Verbal: University of Maryland—letter-number span
	Nonverbal: Wechsler Memory Scale (WMS)—III spatial span
Verbal learning	Hopkins Verbal Learning Test (HVLT)—revised
Visual learning	Brief Visuospatial Memory Test (BVMT)—revised
Reasoning and problem solving	Neuropsychological Assessment Battery (NAB)—mazes
Social cognition	Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT)—managing emotions

Table 2.

Phenotype	Description/neurobiology	Animal model	Heritability estimates	Comment
Eye tracking	Smooth pursuit of a target by the eye—involves motion processing (MT, MST), attention (posterior parietal) and predictive oculomotor (frontal eye fields) responses.	None	Depends on the measure used—recent studies with specific measures report high heritability (h2=0.7-0.9).1	Two independent studies found significant linkage of eye tracking phenotype to loci on chromosome 6p21.2,3 Association with COMT and DRD3 genes reported.4,5
P50	Sensory gating of the second of paired auditory stimuli. Modulated by nicotinic drugs, as well as adrenergic and 5HT system.	Yes-N40	Heritability estimate based on a twin study, h2=0.44.6	A finding of genetic linkage of P50 phenotype at 15q14 (near the location of α 7 nicotinic receptor gene.7,8
PPI	Sensory gating of the startle response to loud sound (pulse) by a muted auditory prepulse. Modulated by several neurotransmitter system including dopamine, glutamatergic, and 5HT.	Yes-PPI	Normal twin studies suggest that about 50% of the variance in PPI can be explained by the genetic factors.9	Early evidence suggest that the PPI phenotype is distinct from P50. QTL mapping identified loci on mouse chromosome 16.10
Sustained attention	Using a continuous performance task (CPT).	No (?)	Estimated heritability range between 0.48-0.62.11	Associated with the dopamine transporter gene in ADHD.12
Visual working memory	Neural correlates of oculomotor delayed response task are well described by human imaging and monkey neurophysiological studies.	Yes	Estimated heritability is reported to be ∼0.45 including in families of schizophrenia probands.11	Preliminary data show an association with DISC1 gene in families of schizophrenia probands.13
Verbal learning and memory	Different measures across studies tapping into this domain.	No	Estimated heritability based on healthy twin studies, h2=0.44. Similar findings in schizophrenia families, h2ranging from 0.21-0.49.11	Genome-wide data from 168 schizophrenia families suggested a locus for verbal learning and memory on 4q21.14 May be associated with BDNF gene.
Brain morphometry	Decreased total brain volume; decreased white matter volume in some tracts; and thalamic volume reductions noted in relatives of schizophrenia probands.	Yes (?)	Heritability estimates for several cortical regions are high (0.80-0.99 based on healthy twin studies).15-17	Further research with focus on specific brain morphometric measures needed.

Table 2 Courtesy of gunvant thaker, MD.

Table 3.

Animal preparation	Phenotypes assessed
	DA-related behavior	Gating	Cognitive behavior	Social behavior	Molecular signature	Response to APD
Genetic preparations
NMDA NR1 receptor hypomorph (Mohn et al, 1999; Fradley et al, 2005)	Enhanced response to amphetamine	Disrupted PPI		Impaired social interaction	95% reduction in NRI expression	Behaviors improved by APD
STOP KO (Fradley et al, 2005)	Hyperactive	Disrupted PPI				PPI deficit not blocked by clozapine
GluR-A receptor KO (Bannerman et al, 2004)	Hyperactive		Disrupted spatial working memory			Anxiety prone
GABAAα3 receptor KO (Yee et al, 2005)	Spontaneous locomotor activity slightly increased but not after amphetamine	Disrupted PPI				PPI defect improved by haloperidol Rx
Dishevelled 1 KO (Lijam et al, 1997)		Disrupted PPI		Impaired social interaction
Calcineurin Aγ KO (Miyakawa et al, 2003)	Enhanced response to amphetamine	Disrupted PPI and latent inhibition	Decreased working memory	Impaired social interaction	Inducible KO
Catecholamine O-methyl transferase (COMT) KO (Gogos et al, 1998; Huotari et al, 2004)	No potentiation of amphetamine-induced locomotion				Increased DOPAC, D1 and D2 unchanged	Increased anxiety and aggression
Heterozygous Reeler mouse (Ballmaier et al, 2002; Costa et al, 2002; Podhorna and Didriksen, 2004) Kruger et al, 2006	Enhanced mesolimbic dopamine		Decreased working memory; no decrease in prefrontal cortex dependent task	Impaired social interaction	Reduced GAD 67, increased DNA methylation
Neurexophillin 3 KO (Beglopoulos et al, 2005)	Reduced rotorod performance	Disrupted PPI but increased startle response			Expressed in Cajal-Retzius cells
Neuregulin 1 hypomorph (Stefansson et al, 2002)	Hyperactive in open field test	Disrupted PPI			Reduced NMDA receptor activity	Open field behavior reversed by clozapine but not PPI
DISC1 KO (Hikida et al, 2007; Duan et al, 2007; Pletnikov et al, 2007; Clapcote et al, 2007)	Enhanced locomotor activity	Disrupted PPI; Decreased latent inhibition		Impaired social interaction	Decreased cortical parvalbumin-containing cells, accelerated neurogenesis with aberrant connectivity
ErbB4 KO (Golub et al, 2004)	Reduced spontaneous activity		Reduced Morris water maze learning
nPAS 1/3 KO (Erbel-Sieler et al, 2004; Pieper et al, 2005)	Enhanced open field locomotion	Disrupted PPI	Decreased social recognition		Reduced reelin interneurons
Heterozygous Nurr1 KO (Rojas et al, 2007)	Hyperactive in novel environment and also after amphetamine		Decreased emotional memory		Reduced DA turnover in striatum; increased DA turnover in PFC	Haloperidol reverses spontaneous hyperactivity
Retinoic acid receptor KO (Krezel et al, 1998)	Reduced open field locomotion				Reduced D1R & D2R
Vasopressin 1a receptor KO (Bielsky et al, 2005)			Decreased social recognition			Phenotype rescued by increase V1a expression in lateral septum
Vasopressin 1b receptor KO (Wersinger et al, 2002)	Reduced PFC DA	Disrupted PPI	Impaired social recognition			Atypical APD improve PPI but not haloperidol
Oxytocin and oxytocin receptor KO (Ferguson et al, 2001; Takayanagi et al, 2005)				Impaired social discrimination		More aggressive
Dopamine Transporter KO (Trinh et al, 2003; Rodriguiz et al, 2004)	Increased DA and decreased D1R, D2R, Hyperactive	Disrupted PPI		Impaired social behavior		More aggressive
Regulator of G-protein signalling 4 (RGS4) KO (Grillet et al, 2005)		Subtle PPI deficits	Impaired working memory
GDI1 knockout (DAdamo et al, 2002)			Impaired short-term memory	Diminished social behavior		Less aggression
Cannabinoid receptor 1 (CB1) KO (Haller et al, 2005)	Decreased PCP-induced locomotion			No effect on social interaction
Complexin I KO (Glynn et al, 2005)	Decreased amphetamine-induced locomotion
Complexin II KO (Yamauchi et al, 2005)			Decreased LTP; reduced morris water maze performance only after stress
Homer1a KO (Szumlinski et al, 2005)	Enhanced locomotor behavior to MK-801 and methamphet-amine	Disrupted PPI	Decreased radial arm maze performance		Decreased glutamate release in PFC following cocaine Rx
Glycine transporter KO (Tsai et al, 2004)	Locomotor response to psycho-stimulants same as wild type	Reduced sensitivity to amphetamine to disrupt PPI but more MK-801 induced disruption	Improves memory retention		Increased NMDA receptor expression and function
GSK-3 beta knockout (Amar et al, 2004)		Disrupted PPI correlates with enzyme activity
mGluR1 knockout (Brody et al, 2003)		Disrupted PPI				Not reversed by raclopride
mGluR5 knockout (Kinney et al, 2003; Brody et al, 2004)		Disrupted PPI				APD not effective
Proline Dehydrogenase (ProDH) KO (Gogos et al, 1999; Paterlini et al, 2005)	Reduced open-field behavior, enhanced response to amphetamine and MK801	Diminished PPI			COMT, calcineurin upregulation, reduced D1 and DARPP-32 expression,
Chromosome 22 deletion (Paylor et al, 2001)		Disrupted PPI	Impaired cognitive function
GAP-43 KO (Metz and Schwab, 2004)	Hyperactive in open field, reduced anxiety	Disrupted PPI
NCAM-180 KO (Wood et al, 1998)		Disrupted PPI, no changes induced by apomorphine Rx				Increase lateral ventricle size
Phosphodiesterase 1B KO (Reed et al, 2002)	Enhanced behavioral response to methamphetamine		Morris water maze performance impairment		Increased DARPP-32 phosphorylation
Beta-arrestin 2 KO (Beaulieu et al, 2005)	Decreased locomotor response to amphetamine				Normal DARRPP-32 phosphorylation after amphetamine
Trace Amine 1 Receptor KO Wolinsky et al, 2006	Enhanced locomotor response to amphetamine	Disrupted PPI			Increased psychostimulant-induced DA release
Insulin Receptor KO Zhao et al, 2006		Decreased startle amplitude			Decreased insulin receptor and Akt signaling; reduced phosphorylated GSK-3	Clozapine alleviates insulin resistance
Corticotropin releasing factor (CRF) overexpression (Dirks et al, 2003)		Disrupted PPI
DBA/2 Mouse (Stevens et al, 1998)		Disrupted N40 gating & PPI			Phenotype reversed by α7-nicotinic receptor agonist
129S6/SvEv Mouse Koike et al, 2006			Working memory deficit		DISC1 mutation
Apomorphine Susceptible Rat (Ellenbroek and Cools, 2002)	Enhanced locomotor response to novel open field	Disrupted PPI and diminished latent inhibition
Developmental preparations
Rat prenatal variable stress (Kinnunen et al, 2003; Koenig et al, 2005)	Increased response to amphetamine and PCP with post-pubertal onset	Disrupted PPI & N40	Impaired object and social recognition	Impaired social interaction present in adolescent and adult rats; reversal by oxytocin; no effect of cross-fostering	NMDA, GABAergic and presynaptic protein dysregulation	Stress applied during period of fetal brain development that overlaps with period identified in human epidemiological studies
Mouse Prenatal Viral infections (Fatemi et al, 1999, 2005; Shi et al, 2003)	Decreased open-field exploration;	Impaired PPI		Impaired social interaction	Reduced reelin expression in cortex layer 1; reduced cortical thickness; increased pyramidal cell density	Clozapine and chlorpromazine increase PPI—hyper-reversal of PPI deficit
Rodent Prenatal PolyI:C Challenge (GD 9) (Borrell et al, 2002; Shi et al, 2003; Meyer et al, 2005, 2006)	No change in total distance traveled in open field but reduced center exploration; increased response to amphetamine	Disrupted PPI; latent inhibition changes appear after puberty; no effect of cross-fostering	Reduced escape latency in Morris water maze		Amphetamine-induced DA release increased; increased hippocampal pyknotic cells	Latent inhibition defect reversed by clozapine and haloperidol
Rat Borna disease virus infection (Solbrig et al, 2000; Pletnikov et al, 2002; Hans et al, 2004)	Enhanced novelty-induced locomotor activity in Fisher rats; enhanced amphetamine-induced locomotion	Disrupted PPI in Fisher rats			Impairs BDNF synaptogenesis; prefrontal cortex thinning
Rat neonatal ventral hippocampal lesion (Lipska, 2004)	Increased response to amphetamine	Disrupted PPI	Decreased working memory	Impaired social interaction	Reduced presynaptic protein expression	Deficits improved by atypical APD
Rat antimitotic agent—MAM or AraC (Elmer et al, 2004; Flagstad et al, 2004; Gourevitch et al, 2004) Moore et al, 2006; Featherstone et al, 2007	Enhanced response to amphetamine	Disrupted PPI	Learning deficits in Morris water maze and object recognition; no change in 5-choice serial reaction time task	Decreased social interaction	Decreased brain, hippocampus weight; increased neuron density in prefrontal cortex; enhanced NAc DA release to amphetamine
Rat maternal malnutrition (Palmer et al, 2004)		Disrupted PPI that becomes apparent on PND 56 but not PND 35				Striatal NMDA receptor-binding increased without DA change
Rat prenatal vitamin D insufficiency (Kesby et al, 2006; Eyles et al, 2006)	Enhanced response to MK-801	No disruption in PPI				Stress reactivity unchanged
Rat placental insufficiency/birth insults (Boksa, 2004)	Enhanced response to amphetamine				Reduced DA release in PFC, increased DAT in NAc (basal), decreased DA receptor expression
Rat isolation rearing (Geyer et al, 1993; Varty and Geyer, 1998; Heidbreder et al, 2000; Weiss et al, 2000)	Strain dependent enhancement of amphetamine locomotion	Disrupted PPI that is strain dependent			Increased amphetamine-induced DA release	Raclopride reversed PPI deficit
Monkey fetal irradiation (Selemon et al, 2005)						Mid-gestational irradiation decreases both gray and white matter in frontal cortex
Rat 24 h maternal deprivation on postnatal day 9 (Ellenbroek et al, 1998)		Disrupted PPI, effect develops after puberty				Haloperidol reverses PPI deficit
Drug-induced preparations
Acute NMDA receptor antagonist Rx (MK801, PCP, ketamine) (Jentsch and Roth, 1999)	Increases locomotor activity		Decreased working memory	Impaired social interaction		Enhanced locomotor responses blocked by APD
Chronic NMDA receptor antagonist Rx (MK801, PCP, ketamine) (Jentsch and Roth, 1999; Sams-Dodd, 1999; Lee et al, 2005)	Enhanced locomotor response to psychomotor stimulants	Disrupted PPI	Decreased working memory	Impaired social interaction	Diminished expression of NMDA receptor coupled IEG, Homer 1a	Enhanced locomotor responses blocked by APD, social behavior and PPI impairments blocked by clozapine but not haloperidol
Basolateral Amygdala Picrotoxin Infusion—Rat (Berretta et al, 2001; Gisabella et al, 2005)						GABA antagonism in BLA decreases GAD67 in HPC; GABA antagonism in BLA increases HPC LTP
Lesion models
Neonatal ventral hippocampal lesion (Lipska and Weinberger, 2000; Lipska, 2004)	Enhanced locomotor responses to amphetamine with post-pubertal onset	Disrupted PPI	Various impairments in learning and memory	Impaired social behavior	Reduced presynaptic protein and growth factor expression, reduced NMDA receptor expression, impaired DA receptor expression in frontal cortex	Locomotor responses blocked by APD, social impairments blocked by clozapine but not haloperidol
Amygdalar Lesion (Hanlon and Sutherland, 2000; Daenen et al, 2002, 2003; Weiner, 2003)	Enhanced amphetamine or apomorphine-induced locomotion	Increase acoustic startle response but impair PPI on animals lesioned on PND 7 but not PND 21; abnormally persistent latent inhibition	Impaired place navigation & spatial ability	Social behaviors diminished in animals lesioned on PND 7 but not 21 but ventral HPC lesions did not affect social behavior	Increased lateral ventricular volume
Prefrontal Cortical Lesion (Miner et al, 1997; Wilkinson et al, 1997; Lipska et al, 1998; Lacroix et al, 2000)	Lesion potentiation of amphetamine induced locomotion under high stress conditions only	Medial lesions only augment PPI & lesions have no effect on LI

The domains of pathology paradigm, in contrast to the disease entity paradigm, identifies several aspects of schizophrenia as pathology targets for treatment development. The paradigm suggests separate developmental pathways for treatment discovery for each domain. Cognitive impairment and negative symptoms are the domains with the most compelling case as unmet treatment needs. If the domains of pathology paradigm proves robust in treatment discovery, other domains will be defined.

Animal Preparations and their use in Schizophrenia Drug Discovery

Advances in schizophrenia have been retarded not only by the complexities of the disease and a lack of consensus about the central features of the disease phenotype but also by inconsistencies in the neurochemical and molecular signatures of the disease and the dearth of informative animal models. The lack of useful animal preparations for schizophrenia can also be linked to the uniquely human nature of schizophrenia, which reflects the inability of animals to experience hallucinations and delusions, or even to convey the presence of these disease features. However, recent research is beginning to change thinking about the utility of animal preparations to generate useful information to understand the pathophysiology of schizophrenia, the identification of new treatment targets or conduct early evaluations of putative antischizophrenia drugs. A feature of critical importance to any of these uses of animal models is the relevance of the behavioral endpoints being analyzed in animals to the array of disease symptoms. The translation of some endpoints is immediately obvious, example, amphetamine-induced locomotion compared to amphetamine-induced displacement of radiolabeled raclopride in the human striatum (Laruelle et al, 1996), prepulse inhibition of the acoustic startle response (Swerdlow et al, 1994, 2000), spatial and working memory (Robbins, 1998) or even episodic memory (Eichenbaum and Fortin, 2005). These comparisons require accurate knowledge about both human and animal behaviors and the relationship to the disease process. Given the spectrum of phenotypes associated with schizophrenia, this is not always easily assessed. Articulated in Table 2 are human phenotypes that might be translatable to animal modeling.

Animal preparations that may be informative about some aspect of either the schizophrenia phenotype or etiology can be divided into four categories: (1) genetic-based preparations, (2) environmental-based preparations, (3) drug-induced preparations, and (4) lesion models. Much of the information driving the creation of these animal preparations was generated from either post-mortem human brain analysis using chemical, molecular or neuroanatomical techniques or epidemiological findings. Table 3 lists many of the animal preparations that have been reported to have phenotypic overlaps with either a behavioral component of schizophrenia or an identified molecular characteristic of the disease (Table 3).

Because of the diverse nature of the disease, generation of reliable and informative animal models has proven to be a difficult task. It is, however, fair to say that the models based solely on genetic modifications do not recreate the spectrum of schizophrenia-related phenotypes. Developmental models appear to better recapitulate the breadth of the behavior diversity of schizophrenia. Lesion models also are able to address the diverse nature of the disease. This said, several of the models may prove useful for generating an increased understanding of the pathophysiology of schizophrenia. Arguably, the most informative genetic models appear to be the calcineurin conditional knockout mouse, the neuregulin hypomorphic mouse, and the recently described DISC1 mutant mouse (Stefansson et al, 2002; Miyakawa et al, 2003; Hikida et al, 2007). There are two drawbacks to the currently available genetic models that must be mentioned. First, postmortem findings in schizophrenic brain tissue never reported the complete absence of any protein or mRNA in the human brain. Therefore, most of the current genetic knockout models create artificial voids in protein expression that do not exist in patients, although the use of heterozygous animals and some conditional knockout strategies are beginning to address this concern. Second, schizophrenia is considered a polygenic disease and it is overly simplistic to think that knocking out a single gene will recreate the diverse phenotype manifest in schizophrenic patients. With regard to other models, the rat prenatal stress model has strong face and construct validity based on epidemiological findings associated with schizophrenia (Kinnunen et al, 2003; Koenig et al, 2005; Lee et al, 2007) and the rat neonatal ventral hippocampal lesion model appears to recreate many of the behaviors associated with schizophrenia. However, the limitation to this later preparation is the void created during the lesion process, the consequences of which have yet to be fully established (Lipska, 2004). Nonetheless, the advent of these models, will likely streamline the creation new antischizophrenia drugs by facilitating (1) identification of new targets for drug discovery programs and (2) evaluation of the usefulness of new drugs based on new targets for the treatment of the symptoms of schizophrenia. Arguably, the greatest potential of these models maybe the identification of the pathophysiology underlying the disease and the identification of mechanisms to reverse that pathophysiology. However, because of the recent advent of many of the more useful animal models, there is only limited information available regarding the translation of findings in animal models to the treatment of schizophrenic human beings.

As mentioned above, the molecular pathology of schizophrenia remains obscure. However, investigators have developed genetic animal models for schizophrenia based on information gained from a variety of genetic linkage and DNA microarray studies. The genes most widely accepted to be involved in schizophrenia are summarized in Table 4. At the present time, the animal models based on these genetic findings do not recapitulate the breadth of the schizophrenia clinical profile. Several issues continue to confound attempts to advance further in this area. First, schizophrenia is a complex disease and multigenic. Continuing to pursue single genes as causal agents for the disease is overly simplistic and more efforts need to be directed toward understanding how multiple genes or genes and the environment interact to generate the disease phenotype. A second conundrum is parsing primary effects from compensatory effects. Post-mortem human studies are of great value, but only limited information can be gained about primary vs compensatory genetic changes in this system. Animal models may prove exceedingly valuable in addressing these particularly problematic issues. Finally, the deletion of genes and hence, proteins, from the brain’s normal milieu does not accurately capture the molecular profiles generated in schizophrenic patients. It may be necessary to explore the utility of temporal or spatial genetic knockdowns to gain information about the importance of a number of genes in schizophrenia. Recent work with DISC1 confirms the value of this approach (Hikida et al, 2007; Pletnikov et al, 2007).

Table 4.

Gene of interesta	Animal model available
Alpha-7 nicotinic receptor	Yes
COMT	Yes
DISC1	Yes
Dysbindin	No
G72	No
GAD1	Yes
Metabotropic glutamate receptor 3	No
MRDS1	No
Neuregulin	Yes
Reelin	Yes
RGS4	No

^aHuman schizophrenia-related genes derived from (Harrison and Weinberger, 2004; Rapoport et al, 2005).

Three Conceptual Approaches to Drug Discovery for Pathological Domains

The disease model of schizophrenia’s core pathologies presumes pathophysiologies that are distinguished from normal brain function. Nonetheless, the observed behaviors are on a continuum with normal human function. Patients exhibit impairments or decrements in normal functions, not the absence of these functions. Therapeutic interventions are intended to ‘normalize’ these functions. With this in mind, drug discovery can be based on three conceptual approaches. First, cognitive enhancing drugs could be developed that do not depend on the specific pathophysiology for effect. Pathways of normal cognitive processing, for example, may represent final common pathways for therapeutic effect. An example is the pro-cognitive effect of dopamine agonists in normal volunteers and patients with hyperactivity/attention disorder. A second conceptual approach is the direct correction of the pathophysiology underlying the impaired function. Animal models based on induced impairments could be useful screening tools for determining restitutive or restorative effects of drugs. A third construct involves compensatory mechanisms. Cognitive behavioral therapy appears more likely to be effective if it focuses on compensatory rather than restorative techniques (Bellack, 2003). This third possibility brings a new dimension to the discovery of cognitive enhancing compounds because molecular targets in compensatory pathways would now present an alternative to molecular targets involved in the pathology of schizophrenia per se.

A drug that acts in the pathways involved in compensatory effects may be more effective if the pathological domain is a longstanding trait, at least if functional outcome is the treatment target. The proposition here parallels analgesic development where a drug may target the analgesic pathways (eg, where morphine has its effect) or may target the expectancy pathways where placebo has its effect (Colloca and Benedetti, 2005).

Nine Items Impeding Drug Discovery in 2007 and Recommendations for Future Developments

The synopsis of schizophrenia research provided above, generates a variety of concerns about the current status of drug development for schizophrenia and possibly other neuropsychiatric disorders. Below is a list of nine points, which appear to represent significant hurdles in the current drug-development scheme. Attention to these nine points may lead to more rapid identification and development of new therapeutic agents for schizophrenia. This list of impediments is by no means exhaustive and more significant hurdles may be identified by others. Nonetheless, this list provides a starting point to focus attention on the logjam in drug development for schizophrenia, and to contribute to a discussion of methods for breaking the logjam.

1. The single-disease paradigm with psychosis-defining schizophrenia

Considering schizophrenia as a single-disease entity has skewed drug development to focus solely on psychosis. However, positive psychotic symptoms comprising disorganized thought and behavior, hallucinations, and delusions are only one aspect of schizophrenia pathology. Key domains of pathology-affecting course and functional outcomes are negative symptoms and impaired cognition. Focus on these domains and the critical components encapsulated by these domains will potentially generate endophenotypic markers that will be useful for drug discovery. This paradigm shift is beginning to be employed by many investigators in academic and industrial settings.

2. Failure to identify key molecular pathologic elements as treatment targets

Basic knowledge at the level of molecular etiology and pathophysiology is insufficient to define molecular targets for drug development with high predictive validity for therapeutic success. The study of etiology and pathophysiology at the syndrome level fails to address heterogeneity. It is increasingly imperative to develop information on the molecular basis of each aspect of the disease to perform definitive studies about the illness and generate models that are more informative (Carpenter et al, 1993).

3. Failure to develop and apply animal and human models for specific attributes to enhance early evaluation of candidate compounds

Early proof of concept or proof of principle studies would greatly streamline the flow of drugs through the developmental pipeline. Investment in creating and validating model systems for the specific pathologic domains is essential.

4. A complex multi-factorial, polygenetic brain disorder without known, specific neuropathology/pathophysiology; and limited approaches to identifying novel molecular targets

Sophisticated bioinformatic approaches combined with molecular and neuroanatomical studies are needed to identify the causes and pathologic changes for each symptom domain. The bioinformatic resources to accom-plish this task combined with the nucleic acid- and protein-based investigations that will be needed to break the drug discovery logjam are expensive. At present, conducting the studies needed to generate reliable information on the molecular level requires large resources, usually located in labs at pharmaceutical companies. An additional contribution to the logjam is the limited transmission of such molecular information from industry into the public sector. Facilitating this transfer of information without harming the intellectual property of the companies will be essential if discovery of molecular targets is to be maximized. A final consideration in this regard is the development of consistent information about putative molecular targets. Abundant DNA microarray data are available about the disease but understanding the limitations of working with disease-related post-mortem human brain tissue is an under-appreciated aspect of data interpretation. It may be that informative animal models based on etiological considerations may create new opportunities to identify molecular pathophysiology of the disease domains.

5. Substantial profit associated with marketing of new drugs without regard for novel mechanism or therapeutic advance, thus using discovery pathways that result in D2-active drugs

Developing antipsychotic drugs is relatively inexpensive and low risk for pharmaceutical companies, in part because the pathway for regulatory approval and marketing is clear. Potential profits are large, and a substantial advance over current drugs is not required. This results in resources being devoted to the development of ‘me too’ antipsychotic drugs that neglect novel mechanisms and pathologic targets other than psychosis. Appreciation of the potential market, especially for a pro-cognitive drug, combined with greater feasibility for proof of concept testing, will facilitate development. At the policy level, raising the bar for approval of drugs which are based on the same therapeutic mechanism and which fail to document superior efficacy would shift incentives toward more effective therapies.

6. A general neglect of schizophrenia by society

Schizophrenia remains highly stigmatizing with most patients being impoverished, unemployed and unmarried. Moreover, the burden on affected families is high. Clinical care is inadequately funded, and society’s willingness to pay high costs for drugs is becoming increasingly doubtful. Addressing the lack of public understanding of schizophrenia and the need for supporting additional care costs are beyond the scope of this report, but the reaction to the greatly increased cost of new drugs combined with public attention to the minimal advances in efficacy may create another cycle of neglect.

7. Industry bias against drug discovery for complex diseases affecting impoverished populations

A major impediment to developing new schizophrenia drugs is the idea that schizophrenia is too complex, and there is inadequate knowledge upon which to base rational drug discovery. This perception is based on the heterogeneity of a disease syndrome and a realistic appreciation of current knowledge of molecular pathophysiology. However, we propose that investing in appropriate animal models, molecular tools, Bioinformatics, and neuroimaging for the study of pathologic domains will result in accelerated progress. A complex neuropsychiatric syndrome may be resolved at the level of pathologic domains, each with specific pathways that can be targeted for new drug development. Industry needs to accept the challenge of discovery in the context of a multi-factorial, polygenic syndrome and apply new concepts. A similar statement could be made about other neuropsychiatric disorders. Resolving pathophysiology at the level of pathological domain rather than syndrome may also address the overlap between psychiatric illnesses, and present drug development opportunities that cut across present diagnostic categories.

8. A failure to appreciate the broad market likely for drugs, which effect specific pathologic domains (not disease specific)

Schizophrenia is a disease that has symptoms, which overlap with other disease entities. For example, gating sensory information deficits have been identified in a variety of other diseases, as have cognitive impairments. Once a drug is shown to have beneficial effects in one disease entity, there is a high probability that that drug may also have beneficial effects in another entity with a similar phenotype. The development of adjunctive therapies that have utility in several disease entities could provide pharmaceutical companies with other outlets for their products. These include but are not limited to attention, motivation, reward response, affect disturbance, processing speed, and social cognition. New approaches to FDA for approval for a domain indication rather than a schizophrenia indication may facilitate exploration of efficacy across disease classes.

9. The FDA’s system of disease as an indication

Until now, the FDA has reviewed applications for a schizophrenia class indication. This works well for antipsychotic drug applications. However, the FDA can review applications for an indication that may cross disease lines. For example, analgesic drugs. The FDA is prepared to take the first step in recognizing non-psychotic domains as indications in schizophrenia. Representatives of the FDA have participated in developing clinical trials guidelines for addressing an indication for cognition and for negative symptoms (Buchanan et al, 2005; Kirkpatrick et al, 2006).

Attention to these nine issues will facilitate the development of drugs with efficacy for pathological domains other than psychosis. Success is likely to substantially improve functional outcomes, a challenge not met by current drugs. Particularly encouraging in this regard is the progress associated with the NIMH MATRICS project (Stover et al, 2007), the identification of potential molecular targets for therapeutic development in schizophrenia (Gray and Roth, 2007), and involvement of industry in these developments (Breier et al, 2007). Information about the molecular targets and approaches being taken with these trials can also be obtained from the Schizophrenia Research Forum website (http://www.schizophreniaforum.org).

General Summary

The era of psychopharmacology developed in the same time frame as international emphasis on disease classification and highly specifiable diagnostic criteria. The single-disease paradigm was dominant, and differential diagnosis came to rely very extensively on psychotic symptoms, especially reality distortion symptoms. Treatment effects in schizophrenia clinical trials have been mainly evaluated in relation to psychosis, or to global or total scores on rating instrument. This encouraged the development of antipsychotic drugs while neglecting other pathological domains. Despite unequivocal antipsychotic efficacy using this approach, the long-term functional outcomes have not changed much. The treatment discovery process has produced a series of drugs acting at the D2 receptor, and no drug has been approved for marketing with a novel molecular target. This has been noted as a general problem for the pharmaceutical industry (Drews, 2000; Dutta and Garner, 2003; Mills, 2006; Norrby et al, 2005; Carpenter, 2004; Scolnick, 2004) and applies also to developing drugs for depression and other mental illnesses.

An alternative paradigm identifies domains of pathology within the schizophrenia construct and proposes independent therapeutic development for each domain. The two leading unmet therapeutic needs in schizophrenia, cognition impairment and primary negative symptoms, come into focus in this paradigm. This moves the challenge from developing drugs for schizophrenia to developing drugs for pathological domains within the schizophrenia syndrome. A shift in regulatory focus is also required, but substantial progress has been made within the FDA in preparation to evaluate a drug for an indication within schizophrenia. Clinical trial designs for this purpose relating to cognition and to negative symptoms have been presented with FDA involvement (Buchanan et al, 2005; Kirkpatrick et al, 2006). Therapeutic discovery for these domains will be based on different developmental models, will require new approaches to early drug evaluations and proof of concept testing, and new designs for randomized clinical trial. Specifying elements of these domains can facilitate animal and human model development. The current status in these fields has been explored in recent NIMH workshops (Buchanan et al, 2005; Geyer and Heinssen, 2005; Kirkpatrick et al, 2006). New knowledge on genotype/endophenotype relationships will create another paradigm in which to conceptualize novel drug development (Braff and Light, 2005; Braff et al, 2007; Thaker, 2007).

Polypharmacy is a prevalent and worrisome practice in schizophrenia therapeutics. At present this usually involves add-on administration of more than one drug from the dopamine antagonist class (assuring increased adverse effects without evidence of increased efficacy). If drugs are discovered for pathological aspects of schizophrenia other than psychosis, the approach would be comedication with each drug having a different mechanism of action and administered for a different clinical target. The issue is not polypharmacy or add-on therapy. Rather, independent domains of pathology will require novel drugs, which specifically target an aspect of schizophrenia. If successful, the field would evolve comedication strategies with antipsychotic drugs for psychosis, anti-negative symptom drugs for this pathology, cognition-enhancing drugs for cognition, and so forth. Whether a drug developed for one domain will be cross-reactive with other domains, or whether synergism between mechanisms occurs must await the development of efficacious compounds.

A practical question remains as to whether domains of pathology replace or augment current syndrome classes. There is insufficient evidence at present to determine whether, for example, anergia observed in cases drawn from schizophrenia and major depressive disorder will share the same causal pathway. Or whether cognitive impairment in bipolar disorder and schizophrenia share the same latent structure. Nonetheless, these and other pathological features cut across current diagnostic boundaries at the clinical manifestation level. Therapeutic efficacy may follow the pathological domains rather than being syndrome specific. This is clearly the case for antipsychotic efficacy. However, in the absence of evidence for domain pathophysiological similarity across diagnostic classes, the current syndrome nosology is likely to continue in DSM-V and ICD-11. What will be new is an effort to deconstruct psychotic illnesses (Allardyce et al, 2007; Dutta et al, 2007; Keller et al, 2007; Vieta and Phillips, 2007; Owen et al, 2007; Tamminga and Davis, 2007) into key dimensions or domains, and identify these dimensions as therapeutic targets. The FDA may grant an indication for a domain within a syndrome, and may still require evidence of efficacy for the same domain in another syndrome. In this regard, pathological domains remain disease-class bound in initial development. Unlike pain, where an analgesic with efficacy can be marketed for pain in various disease conditions, a pathological domain indication may be restricted to cases in the parent syndrome who manifest the domain. But testing efficacy hypotheses in other syndromes will be facilitated. It seems likely that a therapeutic mechanism related to final common pathways, will have efficacy across syndrome boundaries. It is also possible that the therapeutic mechanism may relate to specific pathophysiology, which is unique to a syndrome. Advancing therapeutic discovery for domains of pathology will greatly enhance our understanding of the pathologies, which are disease-specific and the pathologies, which define the porous boundaries of our present classification system.