Early Detection and Preventive Intervention in Schizophrenia: From Fantasy to Reality

Scientific progress in understanding human disease can be measured by the effectiveness of its treatment. There are three ways in which it can be measured: alleviation of symptoms without affecting the disease course (e.g., l-dopa for Parkinson’s disease, cholinesterase inhibitors for Alzheimer’s disease, triptans for migraine headaches); prevention of progression, but not curative (e.g., glatiramer and interferon for multiple sclerosis, anticoagulants for cerebrovascular disease, anticonvulsants for seizure disorders); and curative or preventive (e.g., surgical procedures for tumors, antibiotics and vaccines for infectious diseases).

The first effective treatment for schizophrenia and related psychotic disorders was chlorpromazine, the prototypical antipsychotic drug (1). Numerous other compounds were subsequently developed, all of which work by a common mechanism of action: inhibition of dopamine at the postsynaptic D₂ receptor. More than four decades of research with antipsychotics has definitively proven their efficacy in alleviating psychotic symptoms and preventing symptom recurrence (2–4). Despite these results, antipsychotics were considered to be a symptomatic treatment, acting like cholinesterase inhibitors for Alzheimer’s disease or l-dopa for Parkinson’s disease, and were not believed to resolve the fundamental pathology or alter the course of the illness (5, 6).

However, beginning in the late 1980s, studies of first-episode psychosis patients (7), in conjunction with seminal reviews by McGlashan (8) and Wyatt (9), revealed two revolutionary findings that challenged that premise: 1) longer durations of untreated psychosis were associated with worse treatment responses and outcomes (10, 11); and 2) a significant portion of first-episode psychosis patients could achieve symptom remission and recovery if they received effective treatment in a timely fashion (12–14). These studies indicated that the timing of treatment in first-episode patients was a critical factor in determining their prognoses. More generally, they also suggested that active psychosis was, in effect, “bad for the brain” in that it reflected an underlying pathophysiologic process that was progressive unless alleviated by treatment.

These findings significantly influenced scientific thinking and clinical approaches to schizophrenia and raised the possibility of treatment being able to limit its morbidity and even prevent its onset—or, in other words, modify the course of the illness. In this article, we review the evidence for this radical reconceptualization of schizophrenia and therapeutic prospects for those affected.

Nosology: the Early Classification of Schizophrenia

Progress in understanding schizophrenia can be said to begin with the description of dementia praecox by Pick and Kraepelin (15, 16) (Table 1). (Before that, psychiatric nosology was chaotic and characterized by a conflicting mosaic of inconsistent systems in which disease categories were based on short-term and cross-sectional observations of patients, from which the putative characteristic signs and symptoms of a given disease concept were derived.) The dominant psychiatric constructs, which gave a semblance of order to this fragmentary nosology, were the theory of degeneration and the concept of “unitary psychosis” (Einheitspsychose) (17). What distinguished what we now call schizophrenia from other conditions whose clinical profiles included psychotic symptoms was the patients’ clinical deterioration over the course of their illness. Hence, Pick and Kraepelin applied the term dementia praecox because of the progressive impairment in cognitive function in young patients in contrast to senile dementia. In addition to establishing diagnosis for a disorder, their conceptualization of the illness hinted at its pathophysiology and defined a potential target for therapeutic intervention.

Soon after that, however, Bleuler shifted the focus to the symptoms of the illness within a given episode and renamed the illness schizophrenia, referring to the schism between thought and affect in patients (18). As a result of the emphasis on cross-sectional symptomatology, the goal of future treatments would be to control symptoms rather than altering the course of the illness.

Efforts to Understand the Etiology and Pathophysiology of Schizophrenia

The serendipitous discovery of antipsychotics gave rise to the first scientifically credible pathophysiological theory of schizophrenia, the dopamine hypothesis (19). Previous theories (humoral, psychoanalytic, social learning, existential, constitutional, developmental, family system, infectious, immune), with the exception of the genetic theory, which was supported by epidemiologic genetic data, lacked substantive evidentiary bases (20–23). Since its inception five decades ago, the dopamine hypothesis has undergone multiple iterations (24–30). Despite these, while durable and heuristic, the dopamine hypothesis has been limited in its ability to fully explain the clinical pathology and course of schizophrenia and lead to innovative and better treatments.

A decade later, two competing pathophysiological theories emerged that subsequently framed the field’s thinking and research (31–34). The neurodevelopmental hypothesis of schizophrenia postulates that etiologic and pathogenic factors (such as genes and environmental insults), antedating the onset of the illness, disrupt the course of normal brain development, resulting in subtle alterations of specific cells, circuits, and their connectivity (31–33). This developmental disruption confers vulnerability to malfunction and the consequent manifestation of symptoms diagnostic of the illness. However, in contrast to other genetic neurodevelopmental disorders, such as autism, fragile X syndrome, or Down’s syndrome, the phenotype of schizophrenia is expressed after a latency period extending into the second and third decades of life. After its onset and over its course, the illness persists as a static encephalopathy, and the brain abnormalities only progress in the context of the aging process (35).

Although heuristic, the neurodevelopmental hypothesis of schizophrenia did not explain variation in the course of the illness after its onset or in response to treatment, and specifically its progressive nature. In addition, it fostered a therapeutic pessimism in which people with schizophrenia were “doomed from the womb” and had bleak prognoses regardless of treatment.

Because of these limitations and in an effort to accommodate the heterogeneity observed in the clinical presentations and research findings of schizophrenia, Crow proposed a “two-syndrome hypothesis” of schizophrenia (34) (Figure 1). Type I was characterized by acute episodes with mostly positive symptoms, minimal negative and cognitive symptoms, good antipsychotic treatment response, and indices of excessive dopamine activity. Type II was characterized by negative and cognitive symptoms (in addition to positive symptoms), poor treatment response, and evidence of brain structural pathology in vivo and postmortem. A novel feature of this hypothesis was the element of unidirectional progression of the illness from the type I to type II syndrome over time.

Shortly after publication of Crow’s two-syndrome hypothesis, McGlashan published a comprehensive review of the natural history of schizophrenia (8) that highlighted the clinical deterioration associated with the illness and specified that this occurred predominantly in the first 5 to 10 years of the illness. Subsequently, Wyatt (9) published an influential review that found that delays in treatment were associated with poorer outcomes. In addition, prospective studies of first-episode schizophrenia patients mapped the progressive nature of the illness and the therapeutic effects of treatment, beginning with the classic treatment study by May et al. (36). Concurrently, longitudinal MRI assessments of first-episode patients found progressive volumetric reduction of gray matter in specific brain regions and enlargement of the lateral ventricles and subarachnoid space (14, 37), in some cases reflecting illness course (38–41) and mitigated by treatment (39–43).

Studies of Natural History and Early Intervention

The consistently replicated correlation between longer duration of untreated psychosis and poorer outcome, along with the increasing reports of morphological changes demonstrated by longitudinal brain imaging studies, resonated with the historical descriptions of clinical deterioration by Kraepelin and subsequent long-term follow-up studies of (Manfred) Bleuler and Ciompi (8). These results prompted researchers to describe the natural history of schizophrenia as progressing through four stages of illness—premorbid, prodromal, onset/progressive, and chronic residual (44, 45). They also prompted researchers to redouble their efforts to develop a strategy to diagnose and treat patients as early in the course of their illness as possible.

The early identification and intervention strategy for schizophrenia offered hope to patients and clinicians alike that treatment could possibly limit the chronic morbidity and disability of schizophrenia (46, 47) (Figure 2). However, two critical challenges had to be addressed to achieve this goal: 1) reducing the time between the patient’s onset of symptoms and their diagnosis and treatment, and 2) engaging patients in treatment to foster recovery and prevent relapses.

Duration of Untreated Psychosis and the TIPS Study

First-episode studies revealed that the amount of time from when patients first developed psychotic symptoms until their initial treatment was on average 1 year in the developed countries and twice as long in developing countries (10). Thus, if the duration of untreated psychosis was a critical factor in determining treatment outcome, it would be important to reduce it to optimize outcomes.

To examine whether early detection programs could reduce the duration of untreated psychosis in people seeking care, McGlashan and colleagues in Scandinavia carried out an observational cohort study of 281 individuals with primary psychotic disorders (48, 49). The Early Treatment and Intervention in Psychosis Study (TIPS) was conducted in four health care sectors in Norway and Denmark. A public awareness campaign and an early detection program were implemented in two sectors, and mental health care as usual was provided in the other two sectors. Once patients were diagnosed, they were treated by the same protocol regardless of which sector they were in. The early detection sectors broadcast public service announcements designed to raise awareness in the sector’s population about mental illness. Similar information was provided to primary care providers and other first-line health care personnel. In addition, a telephone number was included as part of the campaign that would allow patients, family, and friends to call the early detection teams as necessary. Usual care involved no outreach and patients seeking care on their own initiative.

Patients were followed for several years. At baseline, individuals in the early detection regions had significantly shorter duration of untreated psychosis (5 weeks) than those in the non–early detection group (16 weeks), indicating that the public outreach worked to encourage people to seek treatment sooner. Individuals in the early detection group were also less symptomatic and had better functioning (49). At 3 months, individuals in the early detection group had less negative and general psychopathology symptoms and better functioning. At 2 years, despite the fact that there were no differences between the two groups in treatment received, the early detection group still had lower levels of negative, cognitive, and depressive symptoms as measured by the Positive and Negative Syndrome Scale (PANSS) (48).

First-Episode Psychosis Studies and the RAISE Initiative

The second challenge was whether first-episode psychosis patients would remain engaged in treatment for sustained periods following their recovery, and whether they could be prevented from relapsing. Previous studies of first-episode patients had shown uniformly high attrition rates for various reasons, including lack of awareness about the potential chronicity of the illness, treatment side effects, and stigma.

To address this question, the National Institute of Mental Health (NIMH) launched the Recovery After an Initial Schizophrenia Episode (RAISE) initiative and funded two projects. RAISE Connection was a demonstration project of the feasibility of team-based multidisciplinary treatment tailored to first-episode patients that was implemented in the public mental health systems of New York and Maryland. The idea was to create a model treatment program to which first-episode patients would be referred and the outcomes of patients measured. The results of the project clearly indicated the feasibility of this model of care and its effectiveness through high rates of patient retention, treatment implementation, and seemingly good outcomes measured in symptom reduction, level of function, and work or school engagement (50–53). However, since it was a demonstration project, there was no comparison group to show whether RAISE Connection care was superior to usual treatment.

The RAISE Navigate study was a 2-year randomized clinical trial of a comprehensive, recovery-based model of care consisting of four core interventions—personalized medication management, family psychoeducation, resilience-focused individual therapy, and supported employment and education—compared with usual community care for 404 individuals with first-episode psychosis. The results indicated that individuals assigned to the experimental intervention continued in treatment longer and received more services. Moreover, the RAISE Navigate group experienced significantly greater improvement on the quality of life measure (Cohen’s d=0.31) and on total PANSS score (d=−0.29), PANSS depressive factor score (Cohen’s d=−0.23), and Calgary Depression Scale for Schizophrenia score (Cohen’s d=−0.18), but not on positive, negative, or disorganization symptoms, nor on the Clinical Global Impressions severity score over the 24-month period (54). Duration of untreated psychosis was found to moderate the treatment effect on outcomes, so that patients with shorter duration of untreated psychosis improved to a greater degree on measures of quality of life and symptom severity compared with those with longer duration of untreated psychosis.

Among the most important measures of effectiveness were whether individuals with first-episode psychosis would need hospitalization and require Social Security disability benefits or not. As first-episode patients were not disabled at their onset of illness, the need for enrollment in disability programs was an objective measure of treatment effectiveness. However, there were no differences between interventions in hospitalization rates (54) or in the number of study subjects who applied for disability benefits (55).

These results were generally consistent with other first-episode psychosis studies, which found superiority of treatment programs, albeit of modest effect sizes, but the superior effectiveness is dependent on continuing treatment (56).

Despite the RAISE Navigate study’s lack of effects on the economically consequential measures of hospitalization and Social Security disability benefits and modest impact on clinical measures, along with the uncontrolled nature of the study, researchers, clinicians, and policy makers remained enthusiastic about the idea of interdicting the illness at its inception and preventing illness progression and disability. Consequently, first-episode psychosis programs have proliferated (57), with a defined program of coordinated specialty services tailored specifically for first-episode patients fast becoming the new standard of care, such as OnTrack-NY and OnTrack-USA (58, 59) in the United States, EPPIC and HEADSPACE in Australia (60, 61), Forward Thinking Birmingham in the United Kingdom (62), Jigsaw in Ireland (63), ACCESS in Canada (64), and OnTrack-Chile.

Pre-Syndromal Detection and Intervention Research

First-episode psychosis studies prompted researchers to extend the strategy of early intervention to people who did not meet criteria for a syndromal psychotic disorder but were deemed to be at clinical high risk for psychosis. (Individuals at clinical high risk are those who are identified as having nonspecific and attenuated psychotic symptoms and are believed to be at increased risk of developing psychotic disorders, while a diagnosis of prodromal symptoms can only be determined retrospectively for those who go on to develop a syndromal psychosis.) However, the knowledge base and methodology developed in clinical high-risk research were neither as extensive nor as well developed as those of the first-episode psychosis research that prompted and informed the RAISE initiative. Nevertheless, researchers pursued what would be psychiatry’s first attempt at prevention of a mental disorder (46).

History

While contemporary researchers may have used what they learned from first-episode psychosis studies to extend their focus to clinical high-risk patients, it was the eminent psychoanalyst Harry Stack Sullivan who first introduced the idea of prevention of schizophrenia. Sullivan claimed that this most serious of mental disorders was the result of maladaptive relationships and experiences, which could be repaired, rather than hereditary or biological factors (65). Over a half century later, Falloon proposed screening individuals in primary care clinics for the prodromal symptoms of schizophrenia and then applying family therapy to prevent onset of the illness (66).

Sullivan and Falloon were trained in the psychoanalytic and family therapy traditions and focused more on treatment than diagnostic or prognostic measures. While their therapeutic efforts were based on causal hypotheses, they were not supported by empirical evidence. Other investigators focused on predictive diagnostic measures by examining individuals at genetic high risk who had not yet reached schizophrenia’s onset age, hoping to identify measures predictive of schizophrenia (67–72), while others focused on nonspecific symptoms, such as cognitive deficits and general abnormalities in perception and thinking, which they referred to as “basic symptoms” (73).

Diagnostic Criteria

What might be considered the modern era of high-risk research was pioneered by Hafner in Germany, McGorry and colleagues in Australia (74), and McGlashan and colleagues (75) in the United States. They were inspired by the first-episode psychosis studies demonstrating the relationship between duration of untreated psychosis and prognosis and Wyatt’s and McGlashan’s influential reviews. (McGlashan coined the term “duration of untreated psychosis,” and McGorry, who organized the International Early Psychosis Association, established an award in Richard Wyatt’s name.) A key difference between theirs and the earlier high-risk research (e.g., Sullivan, Erlenmeyer-Kimling, Mednick) was that it was now motivated by a temporal urgency for effective intervention to avoid the risk of disease onset and progression.

Hafner emphasized gender and age at onset as key features of natural history and developed the Interview for the Retrospective Assessment of the Onset and Course of Schizophrenia and Other Psychoses (IRAOS). The Australian group initially incorporated genetic risk and functional impairment into their criteria for individuals at high risk and then enhanced them with a focus on attenuated positive symptoms. From their work came the Comprehensive Assessment of At-Risk Mental States (CAARMS) (76), an instrument that both assessed psychopathology and identified individuals at high risk for psychosis. Similar work by McGlashan et al. led to the Structured Interview for Prodromal Symptoms/Structured Interview for Psychosis-Risk Syndromes (SIPS) (75).

While their studies were seminal, their results reflected significant methodologic limitations. The case identification criteria produced high false positive rates (>70% of study subjects who met clinical high-risk criteria did not develop a diagnosable psychotic disorder [by DSM or ICD criteria] over the 2- to 5-year follow-up period), which increased as conversion rates to syndromal psychoses decreased in subsequent studies. This led to efforts to ascertain larger sample sizes, broadening the diagnostic criteria, and, with the encouragement of funding agencies, forming consortia, including the North American Prodrome Longitudinal Study (NAPLS) (77), the Early Detection and Intervention for the Prevention of Psychosis Program (EDIPPP) (57), PRONIA (78), PSYSCAN (79), and the European Prediction of Psychosis Study (EPOS) (80).

Studies by these and other groups further characterized the phenomenology of the high-risk state (81–84) and eventually led to the creation of algorithms and risk calculators for predicting development of syndromal psychosis (85–88). While incrementally better, these instruments still lacked adequate specificity and predictive validity. The limitations of the clinical high-risk criteria were dramatically reflected in the fifth revision of the DSM. The DSM-5 Task Force considered a proposal to include attenuated psychosis syndrome as a new diagnosis purported to identify clinical high-risk patients in the prodromal stage of schizophrenia (89). However, because of its limited specificity and predictive validity, it was judged to be premature and was referred for inclusion in an appendix for insufficiently validated diagnoses requiring further research.

Consequently, intensive efforts have been made to develop confirmatory diagnostic biomarkers by applying a range of modalities (neuroimaging, electrophysiologic, neurocognitive, serologic) to the study of this population in an effort to enhance the diagnostic precision and utility of the clinical high risk designation.

Diagnostic Biomarkers

Much of the initial biomarker research with functional and neuroanatomical MRI applications found individuals at high risk to have values intermediate between healthy individuals and individuals with syndromal psychosis (Table 2). Later work focused more on neurochemical imaging with MR spectroscopy and positron emission tomography (PET) and MRI-derived measures of connectivity and function, which found abnormalities similar to those seen in individuals with syndromal psychosis in clinical high-risk patients but of lesser magnitude. Among these were neuroimaging studies that added to past postmortem research (90–92) and highlighted the involvement of medial temporal lobe structures, specifically the hippocampus, as reflecting structural and functional abnormalities in the early stages of, or at high risk for, schizophrenia (93–97).

Some of the most promising biomarkers were measures of midbrain and striatal dopamine concentrations, metabolic activity in the (predominantly) left CA1 region of the hippocampus, glutamatergic activity in the associative striatum, decreased mismatch negativity, automated speech analysis, thalamic dysconnectivity, reduced number of sleep spindles, salivary cortisol, blood-based biomarkers (oxidative stress, inflammatory, immune analytes), and neuroanatomical maturity (93, 98–113). However, the ability of even these measures to predict which clinical high-risk individuals would progress to syndromal psychosis was insufficiently reliable to be clinically applied.

Preventive Intervention Studies

The ultimate goal of schizophrenia research is to prevent (or delay) the onset of the illness. Suffice it to say, despite efforts to develop treatments for prevention (Table 3), as well as success developing treatments that can decrease the burden of attenuated symptoms, no treatment has yet been proven effective for prevention of transition to syndromal psychosis. Experimental medicine approaches to develop preventive treatments for schizophrenia is a major emphasis of the NIMH Accelerating Medicines Partnership between government, industry, and academic investigators. These studies are intended to test new chemical entities aimed at specific molecular targets with compelling scientific rationales for putative therapeutic efficacy.

The criteria for considering candidate preventive treatments necessarily differ from those that apply to treatments for the full-blown illness. In the latter case, the diagnosis is already established, symptoms are severe, treatments with proven efficacy and known safety exist, and there is some understanding of the treatments’ mechanism of action and relationship to the disease pathophysiology. For the former, there is uncertainty about the goals of treatment (amelioration of the attenuated and nonspecific symptoms versus prevention or delay of progression to syndromal psychosis), limited awareness of the pathophysiology that putative treatments should target, and a lower threshold for safety and risk of iatrogenic harm.

To meet the challenge of developing treatments for prevention in people who are in their prodromal stage of illness and may not be clearly diagnosable, greater methodologic rigor and a higher standard of proof are required. Future research to validate diagnostic or theragnostic biomarkers and treatment strategies for clinical high-risk and prodromal populations ideally should be guided by scientifically credible pathophysiological theories for which there is substantial supporting evidence.

Pathophysiological Theories to Guide Future Research

There are currently three viable theories with various permutations and many areas of overlap to guide research on biomarker and therapeutics development for early detection and preventive intervention in schizophrenia.

Dopamine Theory

The original dopamine hypothesis, in 1974, postulated that overactivity in neurotransmission from dopamine cell bodies located in the ventral tegmental area of the midbrain to their terminal fields in the nucleus accumbens results in the development of psychotic symptoms (19). The next iteration of this hypothesis came with the work of Pycock et al. (24) and was extended to include both frontal-cortical (mesocortical) and subcortical (mesolimbic) components and demonstrated their reciprocal relationship. Deafferentation of frontal cortical neurons has been shown to result in decreased cortical dopamine activity and increased subcortical dopamine activity. The hypodopaminergic state in the frontal cortical terminal fields of the dopamine neurons, whose cell bodies are located in the ventral tegmental area, was hypothesized to be the basis of the negative symptoms of schizophrenia—avolition, apathy, alogia, and asociality—while hyperactivity of dopamine neurons projecting to the nucleus accumbens and limbic cortex were believed to cause the positive (psychotic) symptoms of the disorder (25, 30, 114, 115).

More recently, PET studies have suggested that the rostral caudate (i.e., associative striatum) may be a pathological locus in schizophrenia (98–100, 116, 117). Abi-Dargham and colleagues (118) have proposed that abnormal brain development leads to disruptions in the connectivity of dopamine pathways to the rostral caudate, including altered differentiation of the associative striatum from the sensorimotor striatum, abnormal patterning of dopamine and other inputs to the associative striatum, abnormal interactions between striatal dopamine and other neurotransmitters, and abnormal cortical inputs to the associative striatum.

Another iteration of the dopamine hypothesis draws from the substance abuse literature. Sensitization is defined as a behavioral, physiological, or cellular response to a stimulus that is augmented as a result of repeated prior exposure to that stimulus (119). This definition is derived initially from behavioral models of learning and refers to the progressive increase in the behavioral effects of psychostimulant drugs (which are indirect dopamine agonists) that are elicited by intermittent drug administration (119). Applying this concept to the dopamine hypothesis postulates that dysregulation in presynaptic trafficking of endogenous dopamine in the mesolimbic pathway mimics the effect of exogenous stimulant administration by releasing excess dopamine, which overstimulates postsynaptic receptors and causes neurochemical sensitization (27–29). As this aberration increases, ideas of reference become delusions and illusions become hallucinations.

Kapur elaborated the model of dopamine hyperactivity and neurochemical sensitization by applying it to the cognitive function of salience (29). Salience refers to a process whereby environmental stimuli and internal thoughts command our attention and drive motivation and behavior. The same mesolimbic dopamine system in which hyperactivity produces psychotic symptoms is believed to be a key component in the attribution of salience, as it is in reward and hedonic perception. In Kapur’s innovative conceptualization of salience as a mechanism in the pathogenesis of schizophrenia, dysregulated dopamine leads to aberrant assignment of salience in which normative stimuli are reified to ideas of reference and illusions and ultimately to delusions and hallucinations.

While the earlier versions of the dopamine hypothesis would suggest dopamine antagonists (i.e., antipsychotics) as treatments, the sensitization and salience versions might predict that while antipsychotics could dampen the symptoms, they would not remove their underlying basis. However, novel agents like NMDA antagonists, which have been shown to block the induction of sensitization (although not its expression once established), could be effective (27).

Synaptic Pruning/Immune Theory

In 1982 Feinberg (120), prompted by the work of Huttenlocher (121), proposed another pathophysiologic hypothesis, postulating that the maturational process of synaptic pruning is excessive in persons who develop schizophrenia. Elimination of extraneous synapses normatively occurs in the human brain during adolescence and into the third decade of life. This late phase of cortical maturation is unique to humans and corresponds to the temporal window of symptom expression in schizophrenia. This theory is consistent with the loss of cortical gray matter without cell death, which is the principal finding of postmortem studies in schizophrenia brains (122).

Keshavan et al. subsequently integrated this pathogenic model with the neurodevelopmental hypothesis (123), and, most recently, genetics studies by Stevens and McCarroll and colleagues provided a mechanistic explanation for how excessive synaptic elimination may occur (124, 125). They found an up-regulation of complement genes (specifically C4) in people with schizophrenia and speculated that this gain of function could cause microglial activation, which could phagocytose synapses. This finding leveraged the work of Shatz, who had previously discovered that major histocompatibility class I genes were integral to the molecular mechanisms of synaptic pruning and circuit formation, in addition to their more commonly understood role in conferring cellular immunity (126). They further observed that C4 was expressed in the gray and white matter, with the greatest concentration of C4-positive neurons and synapses in the hippocampus.

This work provided an important link between the immune system—long suspected to play a role in the pathogenesis of schizophrenia—and the trajectory of brain development and maturation (127). The possibility that neuron-microglia interactions via the complement cascade caused excessive synaptic pruning by activating microglia during adolescence offered a potential mechanism to explain the reasons for age at onset, symptom expression, brain volume change, and synapse loss.

The therapeutic implications of this hypothesis are varied, suggesting potential for compounds that would inhibit complement gene expression, block phagocytosis, or target inflammation. There is also the question of timing of the treatment administration to the critical period of development during which this excessive synaptic elimination is thought to occur.

Glutamate and GABA Theory

The glutamate hypothesis of schizophrenia postulated that dysregulation of glutamatergic chemical neurotransmission causes excessive synaptic concentrations of glutamate by pyramidal cells, which leads to the clinical manifestations of the illness and potential excitatory neurotoxicity (128, 129). Javitt and Zukin proposed this hypothesis in 1991 (128) based on their studies of the psychotomimetic effects of phencyclidine. Their initial focus on NMDA receptor hypofunction was expanded to include the surge in presynaptic glutamate release in response to impeded NMDA signal transduction (129).

An important addition to the glutamate hypothesis came from the laboratories of Lewis (130), Benes (131), and Krystal (132), which consistently found reduced numbers of GABA interneurons in the hippocampus and prefrontal cortex. The reduced perisomatic inhibition of interneurons on target pyramidal cells leads to excessive glutamate release, which in turn could result in further loss of GABA interneurons (130).

The work of Hensch and colleagues provides a possible explanation for the deficit in GABAergic regulation of pyramidal cells releasing glutamate involving the erosion of perineural nets surrounding GABA interneurons (133). Perineuronal nets are extracellular matrix chondroitin sulfate proteoglycan–containing structures that surround the soma and dendrites of various mammalian neuronal cell types, including parvalbumin staining basket cells, the class of fast-spiking interneurons that regulate the activity of pyramidal cells. Perineuronal nets are important for the onset and closure of the critical periods for developmental plasticity extending into the rodent equivalent of adolescence and early adulthood (134). These researchers found that when they induced oxidative stress in newborn mice, which disrupted the perineuronal nets that enveloped vulnerable GABA interneurons, they failed to provide normal inhibitory input to target pyramidal cells, leading to excess glutamatergic function. Antioxidant agents and glutamate modulating agents reversed this extended window of vulnerability and restored integrity to the protective perineuronal nets, normalizing GABAergic inhibition and glutamate levels (135).

Finally, the previously described studies of Shatz, Stevens, and McCarroll and colleagues (124–126) provide another mechanistic explanation for the dysregulation of glutamate and GABA. The toxic gain of function in C4 genes that was found to be expressed in neuronal cell bodies, dendrites, and synapses observed in the hippocampus can reduce the inhibitory control of GABA-secreting interneurons on glutamate-releasing pyramidal cells (126). The consequent glutamatergic dysregulation could cause additional GABA cell loss and further disinhibition of pyramidal cells.

In a parallel line of clinical investigation, Small and colleagues (93) demonstrated in humans that neuronal hyperactivity in the CA1 region of the hippocampus is driven by excessive glutamate and constitutes an early locus of pathology in schizophrenia and precedes the development of atrophy and progression to syndromal psychosis in clinical high-risk individuals. The persistence of dysregulated glutamate activity enables the pathophysiology to spread to the subiculum and other subfields of the hippocampus, and externally to the striatum and frontal cortex, eventually producing reduction of cellular matrix, reflected as atrophy (136).

It should be noted that proposals integrating dopamine and glutamate systems have also been described (137). An accelerator-brake model was proposed by Carlsson et al. (138) connecting excess cortical glutamatergic activity to striatal dopamine release. The model predicts that if striatal dopamine release is enhanced, feedback regulation will increase the activity of the brake to counteract dopamine release. If the brake fails because of a (hypothetical) NMDA receptor deficiency in schizophrenia, the release of striatal dopamine will be enhanced—a prediction that was demonstrated with neuroreceptor imaging of dopamine release in humans (139). More recently, a preclinical model connecting hippocampal glutamate hyperactivity to increased striatal dopamine function has been developed by Lodge and Grace (140) that explains the coexisting disturbances in these neurotransmitter systems in individuals with schizophrenia and presumably those at clinical high risk.

A Unified Model

A unified pathophysiological model can be derived from these different but interconnected theories that incorporates the principles of altered neurodevelopment, including the fine tuning of neural circuits through synaptic elimination, and disruption of the excitatory/inhibitory balance between glutamate and GABA systems that in turn leads to symptom expression and disease progression. The key components of this model follow (Figure 3).

Susceptibility genes (141), including MHC class I genes in the hypervariability region of chromosome 6, instigate aberrations in brain development and cell-circuit connectivity that create vulnerability to neural dysfunction, which is mediated by excessive synaptic elimination and/or erosion of perineuronal nets leading to the loss of interneurons that alters the GABAergic inhibitory control of glutamate neurotransmission. While these genetic factors may affect brain development diffusely, the initial (or an early) site of dysfunction is the CA1 region of the hippocampus. This assumption is based on the principle of regional vulnerability, which stipulates that no matter how complex a disorder’s clinical phenomenology and diffuse its pathology, it must begin somewhere and therefore initially emerges in one brain region (142). In schizophrenia, the region is the hippocampal formation (as also is the case in Alzheimer’s disease), which contains a circuit comprising anatomically discrete but interconnected subregions with distinct cytoarchitecture and connectivity (143). Because the subregions are interconnected, when one is targeted it can spread to other subregions regions within and external to the hippocampus.

Studies have localized early dysfunction in schizophrenia to the CA1 hippocampal subregion (93, 144). This local disturbance in CA1 hyperactivity, caused by increased glutamate, is transmitted through monosynaptic input to the ventral striatum, leading to increased dopamine release (143). This in turn stimulates symptom expression and ultimately leads to CA1 volume reduction and spread of dysfunction to other brain regions (93).

This formulation of extant theories approximates a comprehensive pathophysiological model that includes a neurodevelopmental diathesis and disease progression components. To verify, modify, or disprove this model, it is important to study patients from their earliest stages, ideally in their prodromal stage or even at premorbid stages. Presumably, these theoretical models will be useful in guiding research to validate diagnostic and theragnostic biomarkers and treatments for preventive interventions.

Conclusions

The possibility of disease modification and prevention in schizophrenia is no longer a fantasy but a realistic goal. What stands between us and achieving this milestone in mental health care is sensitive and specific diagnostic criteria for the prodromal stage of schizophrenia, the validation of biomarkers, and proof of effectiveness of therapeutic agents. Imaging (MRI, PET), electrophysiologic (event-related potentials, gamma band, delta wave sleep), and serologic (oxidative stress, inflammatory, immune analytes) measures show promise as diagnostic or theragnostic markers either individually or in combination as part of a profile. These must be validated in studies with large patient samples and rigorous and reliable biomarker methodology robust enough to be applied across multiple sites and, ideally, practical enough to be easily implemented and readily acquired in community settings. Candidate compounds for experimental medication trials must target molecular substrates integral to the aforementioned pathophysiologic models. Despite recent failures, glutamate neurotransmission remains the most promising target among currently viable candidates. In this context, mGluR partial agonists, allosteric NMDA modulators, d-amino acid oxidase inhibitors, and N-acetylcysteine warrant consideration for testing. Other targets of interest include muscarinic cholinergic agonists, trace amine-associated receptor agonists, phosphodiesterase inhibitors, anti-inflammatory agents, immune system inhibitors, and antioxidant neuroprotectants (145). In vetting therapeutic agents for testing in clinical high-risk samples, it is important to carefully select biomarkers that most clearly reflect target engagement in order to accurately determine the dose and confirm the mechanism of action to avoid failed studies and misleading results.

These caveats notwithstanding, we can reasonably expect future studies to build on the research of the past four decades to advance our knowledge and ultimately enable this game-changing model of care to become a reality.